.comment-link {margin-left:.6em;}

High Definition Video for Independent Filmmakers
A How To Guide for Digital Filmmakers
Welcome all! This is my blog to share my latest research,
thoughts, etc. on utilizing HD for independent filmmaking.

YES, I am available for consulting
Contact me at mike@hdforindies.com

All content copyright 2004-2007 Mike Curtis.

Tuesday, April 27, 2004

Audio report: Inside Mac Radio from NAB 

Listen to it here.

Monday, April 26, 2004

Austin, TX based Forgent sues Apple & others over purported JPEG compression patent infringement 

Austin, Texas-based Forgent Networks announced late Thursday that its Compression Labs Inc. subsidiary has filed suit against 31 companies for patent infringement -- among them, Apple. Compression Labs alleges that these companies have infringed its U.S. Patent No. 4,698,672. The patent describes a "coding system for reducing redundancy" that Forgent says is integral in JPEG image compression. [MacCentral]



Mike's Comments: File this one next to the SCO Linux lawsuits. Quite quite possibly frivolous. I don't know the nitty gritty yet, but it's quite striking how all of a sudden smaller companies are trying to exert pressure on larger, ah, shall we say, more profitable companies over possible patent infringement.



Perhaps when this most recent spate is over, maybe maybe MAYBE the government will reform our patent process, and hire more (and more qualified) examiners, etc.

Macintouch Readers Discuss X-SAN 

Readers discuss expensive, high-performance Mac storage area networks. [MacInTouch]



Mike's Comments: This thread has been running since XSan was announced at NAB. Folks are talking about other options already on the market. I finally got the salient answer to my question about "What's the big deal about XSan? What does it do that I can't do already?"



THE ANSWER: You can hook up MULTIPLE Macs to that 2 GB Fiber Channel connection, instead of just one Mac. Enables workground functionality, not just server or single Mac connection

Capture Now time limit not functioning [MacInTouch]

Saturday, April 24, 2004

"the motion picture film industry lives at the behest of the consumer 35mm film industry" 

Greetings all - I'm back in Austin, sitting at a funky little south Austin coffee shop (Bouldin Creek), wirelessly blogging whilst awaiting my breakfast tacos. Oh, God, I'm becoming one of THOSE people...

STILL haven't had a chance to sit down and go through my audio notes. And with 500 video clips to sift, color correct, and recompress into 1000 DV files for a client's project by Tuesday morning, I may not for a few days yet.

OK, sorry, I'm getting into personal blog mode which Charlie Wood of GlobeLogger.com made me promise not to do if he was going to help me.

SO, TO THE POINT:

While I was at the Digital Cinema track at NAB, one of the panelists made a point to the effect of (not an exact quote):

"When was the last time you heard a Kodak or Fuji rep talking about how much better film is than digital? (Mike's note: he was talking about reps for the consumer 35mm film industry)

"No? Me either."

"The motion picture film industry lives at the behest of the consumer film industry."

Keep in mind, that gorgeous 35mm film, that has better lattitude, better resolution, nearly infinitely variable frame rate, is all based on consumer 35mm still photography film. The professional still 35mm film industry is also just an offshoot based on the vast infrastructure, technology, and especially the economies of scale that allows motion picture film to leverage all those benefits I just mentioned of being a big industry.

As digital still photography continues it's aggressive takeover of the marketplace, the amount of film being sold decreases. (How many of your NON movie junkie friends still shoot film for their snapshots? How many digital pictures do YOU take as compared to film?) Polaroid filed for bancruptcy not too long ago. I need to check stats on it, but I'm sure Kodak, Fuji, Agfa, etc. sales are down for 35mm film.

So how long until the price of 35mm film stock goes up, because there isn't as much of a consumer industry to amortize those research and development efforts across? Do the new stocks being advocated for DI (Digital Intermediary) work, such as Kodak's 5218, cost more or less than their predecessors? It's always easy to claim something new costs more to amortize that recent R&D, but that is an easy way to hide the fact that it just NEEDS to cost more since there isn't as much consumer industry to support these efforts.

Assuming this logic is correct, it is inevitable that the cost of film stock will rise.

In the meantime, digital acquisition devices, such as the Thomson Viper, the Dalsa Origin, the Arriflex product (only a "functional demonstration unit" at this time), wil drop in price. Dalsa hopes to rent their bulky camera, with the bulkier drive storage device, for around $3000 a day. Ouch. But then again, that includes all the footage you want to shoot (assuming you don't exceed the storage capacity of the attached heavy heavy HEAVY disk array), and it's already digitized imagery, so no need to develop/process film and scan it in. And oh yeah, it's previewable immediately on set. And it has excellent dynamic range, blah blah blah.

I need to do some spreadsheet work here - what is the breakeven point, in terms of how much footage shot per day, developed and scanned, equals the cost of a Dalsa Origin? And if it isn't feasible to match those, how long until it is? At what price point will Dalsa be on par with film, at least from a cost perspective?

It is clear that the cost of digital acquisition devices will drop over time - that is the nature of digital technology. Moore's law and all. Plus, it's clear that this is the brand new, just invented tech - we're at the most expensive point in the price/performance scale. These technologies are just getting into the cost effective, product-izable point in the development chain.

So film is going to get more expensive, digital is going to get cheaper.

How many years until a movie will advertise "shot on glorious film, just like in the Golden Era of moviemaking, not this new cheap digital stuff"?

It'll happen. Or at least it's certainly possible. Just a matter of time and changing opinions.

Friday, April 23, 2004

Note to Readers about NAB postings 

First off, sorry so much of this is so raw, unedited, and lacking links. I've been in a huge hurry to get it all posted (late for a date as I type this even now).

I am a Bad Blogger. Bad bad blogger. Don't do it again.

Secondly, there were some issues with the Atom and RSS feeds such that you might have missed some posts. We think we have them fixed, much thanks to Charlie Wood from GlobeLogger.com for his ongoing help and support. (See? I linked! ...once....)

UPDATE Saturday afternoon - site adjusted so that all NAB content is still on the (ridiculously long) main page. But I have posted 30,000 to 35,000 words since I arrived at NAB last week (I'm back home in Austin, praise be.)

Notes from NAB Tradeshow Show Floor, Day 4 

Notes from the show floor, Day 4:

--------------------

posted some new pictures of the following:

-Rorke Data Systems' Galaxy 16i RAID setup, utilizing RAID 5 with 16 SATA drives. Tweaked firmware in RAID controller allows for awesome 500+ MB/sec transfers

-Jeff Cree from Band Pro - this guy (and everyone else from there I talked to) sell a LOT of HD gear, and REALLY know what they are talking about. Not the cheapest vendor, but you get what you pay for. Training etc. included in the package deal, I spoke to some of their clients wandering around the booth asking their own questions and EVERYONE spoke glowingly of them.

-Element Labs is an Austin, TX (where I'm from) based company that makes a pixel wall - each of those squares is a single color, and you can drive it from a QuickTime or AVI file. A few pictures showing different images being displayed.

-some images of the user interface of the Da Vinci color correction ystem. Here, they are doing secondary color corrections. They grey middle bar shows what color area in the spectrum you're adjusting, the pyramid/diagonals around it controls the falloff (this is very much like Photoshop's Hue/Sat controls). Above it are controls for Black level, gamma, gain, and knee, much like a traditional lab color timer would do. Also note in the second pic the arbitrary area defining tools (for color correction in a particular area)

-a JL Cooper control surface, the left side is about $3000 and the right side is about $2000. Synthetic Aperture is working on a a version 2 of Color Finesse that will allow use of this hardware (the controls drive Color Finesse.) Very cool, very high end.

-------------------

OK, Day 4, everybody's verrrrrrrrrry tired...

Went by the Da Vinci booth and checked our their product line, first time I'd been able to sit down and see how their workflow works. Finally got a tour of how the system works, their workflow for primary (overall scene) color correction vs. secondary (channel/color range specific) color correction. Interesting to see, sdie by side, old school style lab type color corrections next to modern, Photoshop like color correction tools. They can do realtime from telecine systems at 2K, but from disk they are 6-15fps, limited mainly because their isn't a sufficiently fast/robust data pipe to do it realtime. Drive throughput and a bus to run it over seem to be the limits. But a 2K realtime system runs into the 1.2 to 1.5 million dollar price range.


----------------

follow up on the Kodak Look Management System: it will be supported (and integrated into?) the Discreet Lustre color grading system. They talked about having the client version available for something like several hundred to one thousand dollars for a price point, which is reasonable considering all that it is going to do.

--------------

clarification: color correction and color grading are really the same thing. Sorta like the difference between lawyer and attorney: attorney sounds more expensive.


---------------------

Went by the Christie booth and talked geek with them some more. Internally it can handle data up to 15 bits per channel, but there is no transport that supports that as yet (no pipe, be it the digital transport protocol nor a defined hardware connection for it to run over). But it will handle 12 bit SMPTE 372M, which is 2048x1080 at 24fps 12bit, which looks GREAT. Also, the price point is the same as their 1280x1024 (obviously non-square) version, at $120,000. This is good news, my gut expectation was around $200K.


---------------------

Went by the Rorke Data Systems booth and had a long talk with John Rorke. Their primary user base is medical and industrial image archiving. Clearly they make some very powerful, robust, well thought out products. I haven't checked pricing, but I'm sure it's not cheap. They get a full team in on a bid, so they offer EXTREMELY thorough, competent solutions....that may be overkill for the lower end of the market that I'm addressing. But if I were working on a full sized Hollywood feature, I'd be talking to them.

-------------------------

At the Da Vinci booth, noticed Element Labs was right next door. These guys make configurable (different pixel dimensions) pixel walls that you can feed from a digital file (AVI/QuickTime type stuff). Not in our area of needs, but still way cool. See notes above about their color correction tools.


----------------

Stopped by Reflecmedia. They make a greenscreen backing that has the same kind of glass reflective microbead stuff that is in the shiny tabs on your running shoes. you know, the stuff that reflects light directly back at the source. But rather than having it on your shoes to keep from getting run over at night, they put a ring of LEDs around the lense of your camera that are either green or blue (or a combo kit that'll do both) so that you get green light (or blue) bouncing back to the camera. It makes FANTASTIC green screen backing. you can bunch it up, fold it, etc. and it still generates a GREAT matte. Imagine putting a 60 degree bend in it...and it generates no shadow. See their web samples to get the idea. The backing itself is neutral grey, it just reflects whatever color you throw at it DIRECTLY back at the source. So all your fill/key lights don't get into the key color, just the stuff from that ring around the lense. Very very clever idea, hampered by the cost. But it makes it possible to do things like have a grip walk along carrying this thing behind actors somewhere, and be able to key them out without requiring a huge painted, perfectly lit wall.

They have a 7x7 square of the stuff (It's $30 sq/ft), with the lighting ring that goes on/around the lense, the lighting intensity controller, for either $1700 or $2700, I can't recall without checking my notes.

---------------------

Went by the Plug In Pavillion and saw a few things:

Trapcode had a very interesting 3D particle system running in After Effects, and it could interact (including bouncing off the alpha layer) with 2 other AE layers (no more).

Jim Tierny of Digital Anarchy had a new plugin to do a 3D layer that you could feed an animated bump map. Both this and the Trapcode (another one man shop) show further integration of 3D into After Effects as true 3D stuff, not just 2D renderings of 3D stuff. Both of'em could use the AE camera to pan around'em and stuff. So they are "real" 3D within AE

Talked to Bob at Synthetic Aperture, makers of Echo Fire and Color Finesse. Bob was showing Color Finesse 2.0, due this summer, that is both a standalone and an AE plugin. Allows for 32 bit floating point color correction, and renders at about 1/4 real time (10 second clip takes 40 seconds to render). Talked about 8, 10, and 12 bit support. 10 bit in Final Cut Pro works, 12 bit was a bit vague, will have to talk to the FCP folks and see if/when that exists. Bob was also showing integration with a nice JL Cooper control surface to do color correction with controls like the big boys have. Very nice. See pictures, link at top of this report.

-------------------------------

Personal note: while talking to Bob from Synthetic, they made the "the show is OVER. Go home!" announcement. Fetched my bags from bag check, waited 1/2 hour for a cab, hooked up with Steve Roselle from Alias (friend of mine from Austin), went to airport, got on my delayed flight and CAME HOME.

While I was at NAB, I was debating the wisdom of pursuing all this stuff. I met so many people that knew so much more than I did (although I knew more than a lot of others, and nobody I talked to knew all the stuff from all the areas). Most often I was learning tons from the vendors, sometimes I had useful information for them that they'd write down (so I don'think they were just humoring me). So it was daunting being the diletante amongst the knowledgable and experienced.

So was I nuts to want to pursue a digital filmmaking career in Austin? Was I crazy to think I could make a living, let alone make a difference?

And then, from the airport, as I caught a cab home, I was talking to the cabbie (Doug Sivad, driving whilst he finishes an advanced degree), and he asked where I was coming from, and he tells me he's working on a documentary about native Seminoles down in Mexico (I may have this wrong, it was 1am). We chatted all the way home about new stuff I'd seen, and about the workflow they were undertaking, and about how to get all this stuff done, and what a cool doc it was going to be when finished.

See?

This stuff is EVERYWHERE. People DO want to get involved, are involved, are doing things, they just don't know how to get it all done.

I think there is a lot of this stuff going on in Austin, I'm always hearing about how there is a ton of low level film making/video making going in Austin, but everyone is off doing their own thing and not talking to each other. Not to be secretive, they're just busy doing their own thing, getting (or trying to get) it done and just DOING it, rather than talking about it and crowing about who they know etc. etc.

So yeah, that felt like a great vindication of where I'm going. I do think I can make a meaningful, empowering difference....

...but can I make a living? Time will tell.

Tell me your thoughts - email me at mike@hdforindies.com

PS-this sounds like such a convenient set-up, that the cabbie is a scholar making his own doc. Sounds like a perfectly made up, Shattered Glass type story. But it's absolutely true! : )

-mike

Thursday, April 22, 2004

NAB Show Floor Pictures: More Going Up Now from Days 2 & 3 

Day 2 Pics

Day3 Pics

NAB Tradeshow Report: Day 3 Afternoon/Evening Notes 

Update: I fixed my math on the Dalsa section.

Day 3 Afternoon/Evening Notes

It's taking too long each night to go through my audio notes, so I'm just going to go over what I recall off the top of my head from each booth I visited.

---------------------

I have updated pictures of the ARRI functional demonstration unit, a couple of screen grabs from Kodak's look management system, a picture of the Lumiere developer, a picture of an HD block camera from Sony, pictures of the mythical Sony products (pictures of pictures of them), Pinnacle's proposed workflows, Tony my new British friend, yet another incredibly high speed digital video camera (1000 fps at 1600x1200), indoor helicopters (I forgot to mention the unicylist), the Thomson Viper digital camera, it's capture box and inputs, the Dalsa Origin camera and it's side panel readouts, and some screenshots from the presentation on how Sky Captain and the World of Tomorrow was put together in post. Go here to check them out.

I'll come back and fill in the details after I go through my 10 or so hours of audio notes.

-------------------------------------

Went by the ARRI booth again, called a functional demonstration. They'll ship in hopefully a year, and they want it to be price competitive with a similarly equipped 35mm camera. They see TV shows that are shooting film that are being pressured to shoot HD as their primary target market.

It's 12-bit 4:4:4 dual link HD-SDI 2048x1080, which is SMPTE 372M....I think...I'll have to confirm against my notes.

Still, it's a big heavy beast...but intended for studio usage, and can record to a standard HDCAM SR tape...at least in 1920x1080 mode.

-------------------

Went back by the JVC booth to grill Frederick, the Lumiere developer, on writing back to HDV. HDV is an MPEG-2 transport stream, so you need an MPEG-2 encoder. The Heuris folk had expressed doubt that Lumiere had it's own encoder. I very explicitly asked Frederick about this, and he said yes they DO have an encoder, it's built from some open source/public domain stuff. The question then is how good is it as compared to Heuris', which has a long history of making high quality, broadcast grade encoders. I'll try to figure out a way to do a side by side comparison on some various test footage.

----------------------------

Went by the Sony booth after all the hype about them getting into HDV...and it's all RSN-Ware (Real Soon Now). Although it's not RSN at all. First the bad news: intending to ship by winter.

The good news: Sony is going to do HDV the usual way they do HD: 1080i. Although I was suprised to see a slide that said 1440x1080 interlaced. I'll check my notes, I coulda sworn we were talking 1920x1080i. Talked to a technical guy, he said 25 megabit 1080i (JVC's 720p is 19Mbit), NTSC for sure but PAL who knows (we want 25 close to 24 fps, we could work with that). Definitely going to be 3 chip as are all Sony high end products (they have dibs on 3 CCD HD stuff, everybody else has to go CMOS).

In the demo (of nothing, so I guess it was just a presentation) they mentioned monitoring, and that the Luma series LCD monitors would have FireWire in for viewing HDV. That implies the presence of a hardware codec, likely to be $400-$800 additional for the unit (probably an add-in board).

The deck would read/write HDV to mini-DV presumably, and I don't know what other tape formats. Deck would cost less than camera, not stipulated how much.

Camera and deck shipping winter hopefully, didn't catch when the improved Luma monitors were expected.

Oh, and he said it'll do "real" timecode as metadata, but not VITC or longitudinal timecode, since this is after all a digital bytestream of MPEG-2 transport stream. So they're fixing one of the big problems we've seen. How's that for torture - JVC does the progressive format, but only at 720 without real timecode, and Sony does the right resolution with timecode, but as interlaced...and perhaps as a 4:3 aspect.

All together now: AAAAAAARRRRRRGGGGHHHHHH!!!!!!

-------------------------------------

But video lenses and film lenses are set up ENTIRELY differently. Film lenses expect to focus their image on a perfectly planar 35mm imaging area of essentially zero depth. Video lenses expect to go through a prism for 3 CCDs, or a Bayer type filter for a CMOS imaging setup. When someone asked for commentary on qualitative comparisons between a Bayer filter CMOS and 3 CCDs, the cinematographer audience groaned, and the moderator said that was a seminar or conference unto itself (that interchange was during the Digital Cinema summit).


-----------------

"ingest" is replacing the word "capture" which replaced "digitize" in the industry lingo. You're not trendy if you didn't call it ingesting this year. Perhaps it will migrate to "ingesting content" rather than "ingesting footage" since the foot in footage was feet of film. (Say that 5 times fast.)

----------------

One nice thing about the new Panasonic $25000 DVCPRO HD deck is that it will read everything back to mini-DV. So mini-DV, DVCAM, DVCPRO, DVCPRO50, DVCPRO HD, and I might have missed something in there. But a pretty wide gamut of medium quality footage at each resolution (but no uncompressed). Even D-5 is 4:2:2, don't forget.

One bad thing that came up during the Final Cut Pro user group (and boy, that was detail laden!) was that the $25000 Panasonic deck would let you Print To Tape just fine from FCP, but that you couldn't do a frame accurate insert edit - such as if you wanted to replace a scene in the middle of a tape. For that, you'd need the more powerful/capable $66,000 deck. OOUCH!! So just lay down a new tape.

---------------------------------

Swung by the Leitch booth, they have a $10,000 HD editor called Velocity HD. But the most interesting things in their booth were the short scifi film by Blur Studios (Tim Miller, continue to kick it in the Major Bad Ass department, I was on a panel with him a few years ago and impressed with his forthrightness). Then I saw Tony my new British friend, so I was done with Leitch and left. It does some realtime stuff, blah blah blah, I'll read my literature on it later. Can you tell it's a long show? Can you tell I'm getting tired?

----------------------

Swung by the Kodak booth to check out their Kodak Look Management System that I'd heard such high praise for at the Digital Cinema Summit over the weekend. It looks very interesting, very promising. Let's you use it as both a previsualization tool ("What if we....") and a defining standard ("Yo! Everybody! THIS is how I want it to look when we're done!"). Kodak has DEEP knowledge in these areas, and I have a fairly high degree of confidence they'll be able to pull this off. It let's you tell it what film stock you used, what kind of positive you made from that, what film you used to make a print or what telecine device you used, etc. It can then generate a 3D lookup table. How big? 17 cubed data points, with "tetrahedral interpolation" if I heard that right from the very young but frighteningly sharp tech guy. 17 cubed 3D color lookup tables with tetrahedral interpolation. Yeah, um, right. 220, 221, whatever it takes...you want fries with that, sir?

-----------------------------

Went by the Pinnacle booth where they were showing off their Liquid line of products, all the way from DV to uncompressed HD. Booth was swamped, I'll have to read the literature to see if they told me anything real in there.

--------------------------

Checked out Canon's booth, no HDV products to show here. But they're working on it. Man, GL2's are small and cool. I want one.

----------------------

Went by Huge and Medea's booths. Both have similar products, both are expensive. Did learn that RAID 3 is what they recommend for video. Woops, I covered this already.

----------------------

Sauntered into Thomson's booth to check out the Viper. I've seen too many cameras. It's another contender in the 1080p24 space. Shoots high enough res that they have a digital capture box with drives in it. Details to follow, but pics online. I'm pretty sure this one can do 12 bit data, the guy was interested when I told him BlackMagic for sure and maybe AJA can handle 12 bit data, and BlackMagic was writing their own 12bit codec, and BlackMagic had 14 bit linear D/A component monitoring, which would allow for 12bit log conversion to 14 bit linear, with loadable CLUTs in their converter box for 23HD 8bit monitoring, blah blah blah I'm getting lost here but you get the idea and so did he since he wrote it down.


----------------------

As 6 pm approached, I zipped over to Dalsa's booth to see some more about their camera. 4K, 16 bit internally, they're sending 12 bit (log I think, could be linear) down their quad fiber connection to a rather LARGE box (padded rack width, about 18-24 inches tall, 2 plus feet deep) that is basically a whompin' big disc array with just enough brains to know where to put all those zeroes and ones. Again, sufficient data pipes are what is slowing them down. They are perfectly capable of 48 frames per second at present, but the imager could go to 60 fps (if you had LOTS of light), but there is no interface to handle that kind of data rate at the moment. So the math is this: 4K x 2K (I don't know the exact res off the top of my head, so go with me here) is a 46MB file in 16 bit color. 24 frames per second of that is 1.1 gigabytes PER SECOND. 48 frames per second, which they can do, is 2.2 Gigs PER SECOND. DAMN. 60 would be 2.7 GB/sec.

Now I see why storage is such an issue for 4K.

want a 2 hour movie? Even just the final movie would be 126 terabytes at that size, not including the source scans, the color corrected scans, the shooting ratio, the editing ratio, the DI copy & extra files, the effects files, etc.

Let's carry that out. You wouldn't keep everything 16 bit, but let's call a 2 hour movie's worth of 4K 16 bit 125 TB (terabytes). Multiply all that out and it would be EASY to have a petabyte of info for a movie. Obviously not everything would be 4K 16 bit, but still...It would be cake to use half a petabyte or more (or a LOT more) for one movie.

-----------------------

Finally, I then walked straight over to the Final Cut Pro User Group Meeting at the Stardust. Easily 500 people there for the meeting. Cool. Among other stuff they showed some footage from Sky Captain and the World of Tomorrow.

OK, that's cool, but why you ask. Because the whole movie was done on the desktop. They shot HD, edited and color corrected it in 8 bit in Final Cut Pro 3 and 4, did a lot of their compositing in After Effects, and did the 3D with Maya, all on desktop machines. All the editing was on Macs and digital dailies of effects and treated looks on DVD with DVD Studio Pro using Compressor or Bitvice, and After Effects work was rendered on a PC render farm. All done on Macs. I think Shake may have been used some too, and the original plan was to be all desktop in one facility to do the whole movie. The studio wanted it out in time for summer, so 1/3 of the effects shots were sent out. Now the render farm keeps getting added on to so they can get the movie rendered out in time.

Every frame of the movie is a render, and virtually always a bluescreen shoot composited on 3D backgrounds.

So very very impressive.

All done with KonaHD boards. Ironically, they had playback problems up on stage. Poor guys...in front of a room full of hard core Mac-head editors, no less. But the whole movie was done with that as the crucial lynchpin of production...

Wednesday, April 21, 2004

Day 3 mid-day notes 

Day 3 Notes from mid-day:

-went by the Apple X-RAID area to ask some more about disk throughput stuff, wanting throughput info to know how it performed when the array was full. The guy I spoke to said they have some patent pending tricks (check out the PTO office site to find details) about disk file location optimization. It doesn't just write from the outer edge towards the spindle, and it doesn't just counter stripe (one drive writes edge towards spindle, every other does spindle towards edge).

Went by Eyeon to see their Digital Fusion 4.04 (shipping soon). it's way cool, very powerful, does float, supports ILM's .efx extended exposure range format, etc.

Matrox booth was previewing their HD system to ship by end of year. they were demoing real time 2 layers uncompressed HD, with 4 layers of graphics. Wow. Also will do real time color correction, etc. Shipping end of year. What codec? He wouldn't/couldn't say. No comment

Medea was showing the same array with either SCSI or fiber channel connectors. Does 250 MB/sec, down to about 180 MB/sec at end. RAID 3. Still not cheap.

-Huge blocks off part of the drives as inaccessible to make sure the throughput is all there...but doesn't make it available for anything else...why not make it available for graphics, etc.? they do have some great looking cases that look like mini-G5's, with drilled fronts and round hoop handles. Nice idea, not as eloquently implemented as Apple's. You can see the difference quality machining makes. also expensive.

I happened to have caught a cab over with the the CEO of Popwire, maker of Compression Master. In the Sky Micro booth I finally got a power demo of Compression Master. Why does this project exist? I asked myself this when I saw the press release for the launch of the product. It is going after Discreet's Cleaner. Why? I wondered. And then I found out - Discreet laid off the Cleaner Mac team - they're walking away from it.

I'll cover this in depth later, but it's really powerful, better controls than Cleaner, faster/better MPEG-2 than Cleaner.

If you haven't bought cleaner and you need a solution, buy it. If you have Cleaner and want something better, buy it. About $500.

Went by Curious Software - not Curious Labs, maker of Poser - these are some guys that used to make Matador Paint. It's aimed at broadcast market, where their mapping software rules the broadcast world. It is for matte and effects artists. It's like Photoshop animatable, or After Effects with paint. Not sure if it's a sufficient Commotion replacement, but he was saying there are a lot of disgruntled Commotion users coming over to them, unhappy with the way things have gone since Pinnacle acquired Puffin Designs, original creators of Commotion, written by Scott Squires from ILM. (I used to office a few hundred feet away from these guys in Sausalito)

Went back by QuVis booth and talked to George Scheckel some more at QuVis, their VP of Biz Dev. Their product does 30:1 visually lossless compression of 4K footage at 16 bit (internally), and does a realtime conversion to 12-bit, dual link HD SDI 2048x1080 to a high end DLP projector such as a Christie. The biggest bottleneck right now for better projection is a physical connection standard to send higher than 2K imagery to a projector. Panasonic has a 3800 some-odd resolution projector in the works, Sony supposedly has a 4K they'll show at InfoComm in June.

I'm going to walk back over to the JVC booth to verify the write back to HDV capapbilities of Lumiere.

Mike, 2:12pm Wednesday

NAB Trade Show Day 2 Notes & Comments 

NOTE: My RSS feed seems to be screwy, use the Atom feed for most up to date results.

I couldn't get online this morning to post yesterday's reports, so here's my comments from Day 2 on the NAB trade show floor.


Had a long talk with various guys at the BlackMagic booth, and got lots of encouraging news. Talked to Erick who has done about 20 HD projects with BlackMagic's own PhotoJPEG and offlines with 1080p24 all the time, says it works great.

Talked to the CEO about the 14 bit D/A analong monitoring, he said 14 bit linear is about the same as 12 bit log, so that will work for film folks. The HDLink can load in CLUTs, they are working on software to build/load your own high bit depth CLUTs. They are working on their own 12-bit codec. The spec, if I am interpreting this correctly, allows for 40 bit - 4:4:4:4, with is 10 bits per channel, including an alpha. By dropping the alpha, you can add two more bits to each color value, thus allowing the 12 bit.

But nobody's gear supports that...yet. There is no 12 bit tape format. But for film and DI workflows, this could be quite quite useful.

The HDLink product supports 12 bit dual link connections (so 4:4:4 12 bit), and I have yet to hear any functional distinction between what it does and what the eCinema does and what

Talked to a guy from Bluefish 4:4:4 about why they left the Mac...basically they tried and tried and couldn't get G4 to work properly for their needs...found out Apple had never tested some things, and Bluefish was essentially doing Apple's QA. Then they saw in the high end industry that sequential files was the way all the high end gear was going - that was what they were using - so they wanted to go that way. Then the G5s came out and things worked, but they'd already committed to a #'d file approach, and were going on the PC. They could still come out with a Mac product if they wanted to...

had a long talk at the Panasonic booth with Stuart, he explained how the Varicam really works. The tape format is built around 720p60, so it has to write SOMETHING for each of those 60 "slots" as it were per second. The catch is, internally, it can buffer, and separate what it captures and what it writes.

So it can do true and correct time sampling from 4 to 60 fps. It just writes repeated frames whenever there isn't a one to one correlation between the shooting rate and the writing rate....then adds user bits to identify which frames are different and new, and other metadata identifies the correct duration of that frame.

They are working on an application (or acquiring one, or bundling somebody else's) to let you ramp your timing during a shot and then correctly re-interpolate it from the user bits in the datastream on the Mac. Or something like that. I'll have to listen to my recording of that again.

Spent some time at the Boxx booth to see their uncompressed HD stuff. No real time anything, they are using the BlueFish HD something card. Interestingly, all their literature art is by my friend Policarpo Wood who I used to work with at frogdesign. Rockin'.

So I wasn't impressed with the realtime capapbilities. About 10:1 render time for transitions, about 14:1 render time for color corrections.

However, Premiere Pro 1.5 is pretty interesting. A very After Effects looking interface. Most amazingly, you can highlight a clip in your Premiere Pro timeline, hit copy, go to After Effects, hit Paste, and it drops the clip into the timeline. NO EXPORT FROM PREMIERE AND IMPORT INTO AE. WOW Do some subtle keyframing of the layer - scale, rotate, etc., animate the values as you want...now copy it, go back to Premiere, and hit paste...and it drops into the timeline WITH ALL THE KEYFRAMING ATTRIBUTES!!! Many/most plugins will work interoperably. This easy moving back and forth clearly implies they have moved the incredibly high quality transform engine from After Effects and moved it into Premiere Pro. This is incredibly useful and powerful from a workflow perspective for all those needing to do some tweaking in their editing workflow. This is pretty genius, in that you get the best of both worlds. You wouldn't want the complexity of the AE interface in Premiere, but you can access all the power. This makes me want to have an FCP to Premiere workflow. Automatic Duck? This implies a possible desktop DI type workflow, even though very, VERY non-realtime. This ties into my next comment...

Integration and HD are the two big themes of the show. As the prior paragraph shows, Adobe is serious about tight integration amongst their products. Apple wanted similarly tight integration, so Audition, FCP, & Motion all play very nicely together. Avid is also in on this, with Avid XPress Studio - includes 3D (SoftImage), effects (Based on Boris stuff), audio (Pro Tools LE), editing, etc. etc. etc. They are REALLY trying to hit the soup-to-nuts solutions and have it all tie in nicely...and they've got top-quality parts across the product line up (well, I'm not a huge fan of Boris...). But still, it's mighty impressive.

I had a long talk with Algolith developers (plugins) about the need to port their image scaling and time adjusting plugins to FCP. They had been reluctant to do so because they had heard editors want everything real time....but it's understood that difficult effects take too long.

I also talked to RealViz aobut their Retimer product, and they have a $350 standard def and a $700 high def FCP version (only linear time transforms, no spline based, ramp-up retiming) in addition to their $1700 standalone app. 20% off at the show. The timing: a 1.3 GHz Pentium mobile takes about 30 minutes to retime a 20 second clip. Scale that accordingly to your computer. I'll be in touch with them about the FCP product.

OK, I'm off to see Dalsa, Christie, etc. today.

NAB Trade Show Day 2 Notes & Comments 

NOTE: My RSS feed seems to be screwy, use the Atom feed for most up to date results.

I couldn't get online this morning to post yesterday's reports, so here's my comments from Day 2 on the NAB trade show floor.


Had a long talk with various guys at the BlackMagic booth, and got lots of encouraging news. Talked to Erick who has done about 20 HD projects with BlackMagic's own PhotoJPEG and offlines with 1080p24 all the time, says it works great.

Talked to the CEO about the 14 bit D/A analong monitoring, he said 14 bit linear is about the same as 12 bit log, so that will work for film folks. The HDLink can load in CLUTs, they are working on software to build/load your own high bit depth CLUTs. They are working on their own 12-bit codec. The spec, if I am interpreting this correctly, allows for 40 bit - 4:4:4:4, with is 10 bits per channel, including an alpha. By dropping the alpha, you can add two more bits to each color value, thus allowing the 12 bit.

But nobody's gear supports that...yet. There is no 12 bit tape format. But for film and DI workflows, this could be quite quite useful.

The HDLink product supports 12 bit dual link connections (so 4:4:4 12 bit), and I have yet to hear any functional distinction between what it does and what the eCinema does and what

Talked to a guy from Bluefish 4:4:4 about why they left the Mac...basically they tried and tried and couldn't get G4 to work properly for their needs...found out Apple had never tested some things, and Bluefish was essentially doing Apple's QA. Then they saw in the high end industry that sequential files was the way all the high end gear was going - that was what they were using - so they wanted to go that way. Then the G5s came out and things worked, but they'd already committed to a #'d file approach, and were going on the PC. They could still come out with a Mac product if they wanted to...

had a long talk at the Panasonic booth with Stuart, he explained how the Varicam really works. The tape format is built around 720p60, so it has to write SOMETHING for each of those 60 "slots" as it were per second. The catch is, internally, it can buffer, and separate what it captures and what it writes.

So it can do true and correct time sampling from 4 to 60 fps. It just writes repeated frames whenever there isn't a one to one correlation between the shooting rate and the writing rate....then adds user bits to identify which frames are different and new, and other metadata identifies the correct duration of that frame.

They are working on an application (or acquiring one, or bundling somebody else's) to let you ramp your timing during a shot and then correctly re-interpolate it from the user bits in the datastream on the Mac. Or something like that. I'll have to listen to my recording of that again.

Spent some time at the Boxx booth to see their uncompressed HD stuff. No real time anything, they are using the BlueFish HD something card. Interestingly, all their literature art is by my friend Policarpo Wood who I used to work with at frogdesign. Rockin'.

So I wasn't impressed with the realtime capapbilities. About 10:1 render time for transitions, about 14:1 render time for color corrections.

However, Premiere Pro 1.5 is pretty interesting. A very After Effects looking interface. Most amazingly, you can highlight a clip in your Premiere Pro timeline, hit copy, go to After Effects, hit Paste, and it drops the clip into the timeline. NO EXPORT FROM PREMIERE AND IMPORT INTO AE. WOW Do some subtle keyframing of the layer - scale, rotate, etc., animate the values as you want...now copy it, go back to Premiere, and hit paste...and it drops into the timeline WITH ALL THE KEYFRAMING ATTRIBUTES!!! Many/most plugins will work interoperably. This easy moving back and forth clearly implies they have moved the incredibly high quality transform engine from After Effects and moved it into Premiere Pro. This is incredibly useful and powerful from a workflow perspective for all those needing to do some tweaking in their editing workflow. This is pretty genius, in that you get the best of both worlds. You wouldn't want the complexity of the AE interface in Premiere, but you can access all the power. This makes

Tuesday, April 20, 2004

Pictures? We got pictures... 

I'm posting pictures of interesting things from NAB. See the following:

Best of Digital Cinema Track

NAB Trade Show Day 1

Check the top of that screen, I'll post further pics on an ongoing basis.

Final Cut Pro choices for handling HDV: Lumiere vs. Indie HD Toolkit 

Saw both of these products yesterday back to back.

Both are doing HDV in, conversion to something the Mac can understand, then convert the footage to something you can edit with (your choice of offline codec), then you edit with the offline footage, then they both use the same Offline/Media Manager tricks in FCP to let you swap back out to the high quality footage. You then export that as a single file (not writing to tape from timeline) and can then write it back to HDV tape or D-VHS (both use the same format).

Both are standalone apps or single apps (Lumiere is just one program, Indie HD Toolkit is 3)

Lumiere seems to have a few ease-of-use advantages. It hands off to FCP from it's little program using XML so you can do batch type operations. If automatically "glues" the audio and video files back together in batch so you don't have to do it yourself, one at a time, in QT Pro, to EVERY SINGLE INDIVIDUAL CLIP YOU WANT TO CAPTURE....which is what Indie HD Toolkit requires you to do. I asked about this, and got an answer from a high level Heuris person that I found....less than convincing. They said it was essentially no big deal, that they'd rather do it themselves, and you're going to toss the audio half the time anyway.

Both require you to MANUALLY go through your footage and hit Capture Start and Capture Stop. THERE IS NO BATCH/LOG CAPTURE feature in either. Both realize that is what customers want, and neither is able to deliver it at this time. Part of the problem may be the lack of "real" timecode in HDV, which is an MPEG-2 transport stream. It has some kind of tape code, but frame accuracy appears to be a problem.

The Heuris guy (Heuris make Indie HD Toolkit, and has a LONG history making MPEG encoders, and have a great deal of experience with it) expressed extreme doubt that Lumiere had an HDV encoder. He said something to the effect of "No way...that's 5 man years of effort to do that right." That's not an exact quote, but it was something very close to that. Part of his doubt may have come from how their product is set up. If you just want to capture and get into FCP, it's an approximately $500 product ($400 at the show). If you want to write back to HDV, it's another $4800 or so ($3800 or so at the show.)

Lumiere is $179, $149 at the show. So I need to go back to their booth and see them write altered, edited, composited footage back to tape. We'll see if it does the whole soup to nuts thing right after all.

So far now, Lumiere sounds better. I have complete trust that the Heuris product will work front to back, allowing you to make your color corrections and stuff. Heuris was casting doubt (or at least expressing doubt, I need to be accurate...and nice) that Lumiere was able to write back to the HDV MPEG-2 format.

But when I specifically asked the developer (Frederick, ex-CIO of Hooked on Phonics) whether I could create stuff from scratch - like motion graphics from After Effects - and get it back to HDV, he said yes. So I need to ask what they use to encode back to MPEG-2. Who'd they license, etc.?

Some thoughts on LONG term digital archiving 

OK, this is jumping back a few days to the Digital Cinema track/symposium. There was a panel on long term archives for all these movies we are making, for both the digital assets and the analog film assets.

NOTE TO SELF - I SHOULD WRITE AN ARTTICLE ON DIGITAL ARCHIVING...THEY TALKED AOBUT SAVING THE DIGITGLA MASTER FINAL AND THE SOURCE FILMED OCN, AND THE FINAL PRINT FILM ASTER...BUT SOURCE FILM HAS SO LITTLE TO DO WITH FINAL IN THIS DAY AND AGE, IT IS ALMOST USELESS...LIKE HAVING A PERFECT COPY OF THE CANVAS PICASSO PAINTED ON, OR THE MARBLE BLOCK JUST LIKE THE ONE MICHAELANGELO CHISELED INTO - IT HAS SO LITTLE TO DO WITH THE ART THAT TOOK PLACE, IT IS MERELY ONE PIECE , one raw piece. It is a creative piece, but just an element.

and also about archiving how to put it all together again in paper docs, include the app and the OS...ideally you'd include the machine as well, but that's unlikely. As a society, we are FUCKED in that any data over 20 years old stored digitally is incrasngly nlikely to be radable anymore...or if we do have the softwre, do we still have access to compatible hardware? Ongoing forward migration is the only truly guaranteed solution, but is so pricey...are you really going to stick a top end SGI box on the shelf with a movie? No F-ing way.

So what to do?

1.) Store it in as obvious a way as possible. Compresion is great, until you can't decode it anymore. #d TIFF files wouldn't be a bad digital mastering format, or something similar that is grunt obvious, and each frame had structure info in the header. LOTS of redundancy at this point is important.

2.) Also store the analog version of it - 3 strip color seps isn't a bad idea.

3.) Paper instructions on low acid paper saying what you have, where it is,what's the format, and what's definition/structure of that format. Sprinkile copies of this metadata in throughout the archive. Only redundancy ensures future readability when there's partial loss. Think about getting tax audited in 6 years from now. Some of your records will be lost, and if you used any shorthand in our notes, you better have a keyguide somewhere or it's USELESS.

More on this later...

In short everything possible should be saved and iterated in different ways: a print of the film, analog color separations on black and white film (easy to create from digital), VERY stably filed simple, obvious file formats: #'d TIFF files or somesuch. Then as many of the digital assets as you can keep, including the apps and OS if at all possible...or contracts with the vendors to hold such in escrow (if intellectual property protection issues exist) for a LONG, LONG TIME. Then stable paper records of what you've got, how to read it (description of the TIFF format, and how to read the media it's on). Ideally an eternally archived drive or whatever it was stored on.

Big Question Answered: "Why no 1080p24 in HDV or DVCPRO-HD?" 

AS FOR MARKET PROTECTION ISSUES I MENTIONED YESTERDAY....finally got a good answer on that. PAL decks are built to maintain a precise tape transport speed to do either 25p or 50i. Similarly, NTSC decks precisely do 29.97 or 59.94. Sony, to do their HDCAM that handles 23.98, 24, 25, 29.97, 30, 50, 59.94, or 60 fields/frames per second, has to be able to be variable...and very precisely variable. That's why the other formats don't support the 24p standard. It's difficult. Now, why 720p24 is supported and not 1080p24 in HD-DV, I don't know. Maybe because that would be one more drum speed they didn't want to support.

Panasonic's Varicam is really shooting 60 frames per second, and then selecting SOME of those in order to make up it's 24 frames per second. This is necessary, because the drum speed (the spinning thingy that makes the tape go by) was originally set up to shoot 60 frames or fields per second. The math doesn't seem to divide out evenly, I suspect if you shot something moving at an exactly consistent speed through frame that the distance moved per frame would vary slightly - indicating that the frames were not all taken at precisely timed intervals. They say it does a good job, but that just sounds fishy to me. I don't know that I'd see the difference.

The DVCPRO HD deck from Panasonic is $25,000.

HDCAM Sony F500 deck is "$80,000 and change...a lot of change."

HD For Indies Exclusive (ha) Interview with Ted Schilowitz of AJA 

I met with Ted Schilowitz, Product Manager, Desktop Video Engines, AJA and had a talk with him about working with 1920x1080p24 footage, and some Kona specific workflow questions. Below are my notes. Sorry, no time to clean them up.

-they're based in Grass Valley
-shot on HDCAM, what is offline workflow option?
-couldn't do RT color correction with PhotoJPEG
-there are 40 SE's on FCP - which one did I talk to?
-current and past projects could do PhotoJPEG offline option
-was cool to use PhotoJPEG in past to tiny offline PhotoJPEG
-those PhotoJPEGs were framerate independent in QuickTime, could play it online,
-Techicolor has used it on 4 or 5 big projects
-PhotoJPEG story is going away
-PhotoJPEG was supported to make this stuff work in meantime knowing something better was coming
-Apple is strict about release announcements, so they were supporting this knowing there were better options coming down the road (DVCPRO HD)
-as of yesterday, "there's a better mousetrap, better than the PhotoJPEG mousetrap."
-with their Kona 2 shipping this summer, can go from any HD deck, capture to any of the DV-100 codecs. HD 720p24, 6MBytes/sec.
-there is no 1080p24 framerate. Supports 1080i59.94
-bring it in at 720p24 for offline
-finish as 1080p24 uncompressed batch recapture, media manager, etc.
-what about effects, scaling appropriately?
-talk to Sony & Panasonic about issues he says
-will effects etc. scale properly? Some no, because blurs or other pixel scale effects wouldn't be quite the same. You'd have to "resize" your effects, and/or build it at the bigger size and make an offline version to cut it
-when people did the PhotoJPEG, he told 'em to build the Photoshop at 1920x1080, make a PhotoJPEG size (NTSC sized), and when they relinked media using the Media Manager have it link up to the larger 1920x1080 sized artwork.
-can work, but a pain if you have a lot of effects, etc.
-"never a perfect mousetrap, but chipping away at it"
-can do a realtime transcode, which is contingent on their hardware...as compared to a Decklink or Cinewave...he doesn't know
-he knows that their product will do the realtime transcode, doesn't know about the others
-if have a 1080p24 and don't want to worry about the scaling issues...
-IS IT POSSIBLE TO CAPTURE AT 1920X1080 USING PHOTOJPEG? SEEMS LIKE NO
-the NTSC sized offline works pretty well...play with it

-IT IS POSSIBLE BUT NOT EASY, IT WAS NEVER DESIGNED FOR THAT, IT'S HARD TO PLAY IT BACK, THERE WAS A BUG IN FCP 4 THAT IT WOULDN'T CAPTURE FULL SIZED....LOOKS LIKE THAT IS WHAT HAPPENED TO ME WITH MY TEST WITH GERALD THE GERMAN GUY

-they've recommended 960x540 to work with (1/2 res, 1/4 # of pixels)
-SO DON'T TRY 1080P24 PHOTOJPEG? "I wouldn't think it would be the way to go at all" to do 1080p24 - "it's more trouble than it's worth, to try to use it full size...."{break}....he says it's great to use to export some clips to see if it plays back well, "run it to a digital cinema projector, that kind of stuff...but as an editing workflow, I don't think I'd do that."
-I asked if it's because it's CPU bound: "it's going through and encode and decode process." it's too much at this point with current hardware "it'll play, it's just...why...the 720p stuff is gonna work so much better, look so much better"
-for an effects heavy piece, what is recommended workflow for 1080p24 workflow for an offline choice for offline? HD-DV doesn't support it. He says it would be very interesting to experiment how difficult that scaling would be, or build a bank of effects in 1920 x 1080, then scaling down for the 720p offline, then re-link to the full res ones for the final. His bet it is the EASIER route than to try 1080p24.

Talk to another guy for another approach, Anton Linnaker who works for Technicolor, working with Final Cut, PhotoJPEG, He would be a real good guy to talk about down sides of PhotoJPEG and some of the upsides.

Realtime SD downconversion...it's the same broadcast quality conversion they sell to trucks and stuff all around the country, significant events like Monday Night Football and SuperBowl...NOT a monitoring downconversion.

The monitoring issue - previewing on Cinema Display - "very viable" previewing or monitoring? "Both." How close is 23HD? Doesn't have SMPTE phosphors, LCDs are 8 bit displays, can't go 10bit. Not gonna see a 23HD running in a big edit suite or telecine room to do critical color judgement with...they use a $45K monitor."

for indies, DO use the 23HD converter stuff.

Can buy 2005 20L5 monitor, CRT, about $3 grand, SMPTE C phosphors and is a CRT. combo HD/SD model, only shows 800 lines of high def....tradeoff is 23HD shows every pixel...Sony monitor is component input. "Different strokes for different folks...you could do a certain level of color accurate work on it...talk to Martin from eCinema...they've built a little box that does gamma correction"

Martin from eCinema made the $7K box

at that point, you're better buying monitor with C phosphors

if price sensitive, there are lots of options.

The 23HD is pretty cool, "I'm looking forward to cutting something on a second Cinema display at 1920x1080...that's cool...I know there are limitations there too, I have to know what they are, refresh rate issues, gamma issues, as technology moves forward there is always a price to be paid, it's a good bang for the buck."

if doing 720p stuff, maybe the 20" monitor would suffice with the Martin (eCinema) box

AJA Kona 2 HD/SD Card: notes from NAB Show Floor 

KONA 2 CARD

The Kona 2 is a new HD/SD card from AJA. It will come out in late June for around $2500. A rack mountable breakout box is $300 extra (standard card has a huge gob of cables coming out the back

It'll do analog monitoring with a 12 bit per channel digital to analog converter. High quality.

Broadcast grade up and down conversion - these guys make the converters that are used for little things like SuperBowl and Monday Night Football, so it's good enough for you.


---------------------

KONA BOOTH - FIRST GUY

Kona has a hardware codec that will scale it, so your data rate is 1/4 of normal, Qrez, or use Apple's 8 or 10 bit codec, or Kona's codec which is no longer BlackMagic based.

can do dual link 4:4:4 but they don't have a codec yet

Does SD downconvert, breakout box (optional), same set of realtime effects that Apple does, so G5 handles those.

Kona 2 about $2500 - about the same price point as DeckLink's

Ted Shilowitz is The Guy to talk to...

-------------------------


AJA BOOTH: KONA 2 CARD

-Kona 2: single/dual link HD SDI, 4:2:2 or 4:4:4 HD SDI, SDI, PCI-x 133, 8 channel AES embedded, 10 bit downconvert, 10 bit upconvert, 12-bit analog component output, Qrez 4:1 codec, comes with cables, breakout box (K-Box) optional. Card is $2500, K-Box is $300.

-Qrez is inhouse developed. very good at digital filtering. Qrez comes in full bandwidth, hardware does compression/downsamping to quarter sized. Hardware on the way out upconverts. He thinks a lot of people wouldn't notice the difference. It is a workflow option, including laptop work, a lot of folks would never see the difference if run over the air. Can still read and write codec so long as had the codec available. 133 MHz PCI-X.
-SD and HD
-cross work - SD footage output HD and vice versa, assuming same timebase
-24p workflow for 1080p as well
-RT color correction, eventually mixes and wipes
-CAN do a cross dissolve now
-12 bit D/A for analog monitoring ONLY
-codec is 8 or 10 bit
-no current product handles 12 bit in
-4:4:4 is possible
-board in beta, not released (late June)
-roughly $2500 price point
-they had footage running from Jude law Sky Captain of Tomorrow running on these. Something I read talked about Final Cut Pro, Shake, etc. were used in it's production. Total desktop movie...rock on!

---------------------------------------------------------
working the 24p workflow question...never got 1080p24 PhotoJPEG to work.

Notes on Final Cut Pro HD 4.5 from NAB Show Floor 

DVCPRO HD - in conjunction with Panasonic, it's their new thing announced yesterday. Follow on to DVCPRO 25 and DVCPRO 50. Nine months or so working on this HD stuff. Using the HD camera, shoot native HD at 720p60, 720p30, or 720p24 or 100i59.94. UNFORTUNATELY, NO 1920X1080 AT 24 PROGRESSIVE FRAMES. DRAT!

Once shot it, plug it in through FireWire and suck it across. Don't need FW800, just FW400. The datarate is between 6-14 MBytes/sec. There is NO 1080P24 in the new spec. Drat.

His timeline is ALL realtime.

In low quality you can do 10 streams of HD.

The number of realtime SD streams hasn't been changed, but he had it doing 9 streams off of the X-RAID.

On his timeline, dark green is full res, orange is slightly downsampled to make it work in RT. His timeline was 90% green.

What wasn't doing it -
3 streams (2 type layers and a video layer) was orange
posterize effect
4 way split of same footage, 4 up layout
cross iris transition effect
picture in picture reposition with change in motion parameters
timecode burn onto window AND 3way color correction
opacity change with a widescreen matte box

his name is: Dean MacClain

The Preview on Cinema Display - can do EITHER the Cinema display OR the "real" video output

can do 1080 - anything you can do in your timeline can do Apple Display

with 23HD it is YUV, it is NOT real video, it is PREVIEWING not MONITORING, and you wouldn't do "real" color work with it.

You'd use it to show client work in progress.

Instead of having an ungodly expensive true HD monitor, you'd use this to see what you've got.

COMMENTARY: THIS IS HOW ALL THE CHEAPIES ARE GOING TO DO IT....AND IT'LL BE "GOOD ENOUGH" FOR A LOT OF PEOPLE...AND SINCE THE ALTERNATIVE IS SO MUCH MORE EXPENSIVE FOR NOT A HUGE CLIENT PERCEIVED DIFFERENCE, HARD TO COST JUSTIFY.

It IS a pixel for pixel preview, however.

----------------------

Other new Apple Goodies - mobile HD editing, PipePRO SDI,

Aurora's answer to IO is PipePro, a PCI card, is cheaper ($799) instead of $1200

uses FCP RT engine, everything you could already do with others with standard FCP RT stuff.

He's running 9 layers of realtime uncompressed...the whole Apple booth was running off a bunch of X-Serves...impressive...

--------------------------

looking at Powerbook 17", he has the SAME timeline as the dual G5 had, but it was almost all orange (downsampled)

...but it was set to High Quality. Drop it to Low quality and it does on the laptop what the dual G5 did on high quality

Can do the 23HD display stuff just as well on the laptop. Works on an attached Apple display OR on the LCD on the laptop itself...interesting....

it was at low quality, see some stretching artifacting, but it does it

a la Premiere Pro, if you ask it to do something it can't - in this case, setting the Sequence to high quality, it degrades gracefully - it shows you high quality frames, but the frame rate decreases (substantially in this case, like 8fps instead of 24)

you can flip with a hotkey between fullscreen preview and normal viewing mode - makes it more viable to work with a single screen setup, just got big when previewing stuff.

and it's FREE - $199 for everybody else, $999 new - same pricing scheme as always.

----------------------------------------------------------------

Grass Valley News Edit - don't have the full scoop, but they've chosen FCP to work with/share stuff with. The demo has a timeline set up on a PC with an app called News Edit, and a Mac running FCP with the same timeline open sitting right next to it. Not germane to my stuff, but interesting. I'm guessing it's leveraging FCP's XML roots to do file sharing, perhaps with MXF data (Movie eXchange Format).

MIKE'S COMMENTS: These types of deep data interchanges with FCP is encouraging. This, and Automatic Duck's products, and Lumiere, show how extensible Apple's architecture can be...as well as their WILLINGNESS to do so. OMF and AAF are out there, but Apple has been very quick and aggressive to actively seek an interchange of data with others. Helps TREMENDOUSLY that Apple sells FCP not so much as a standalone profit center (and I suspect/hope it is, check their 10-Q if they break it out that way), but as a means of selling more computers.

-----------------------------------------------------
more on FCP 24p workflow -

now there's a change so you can go 24fps to 25 fps more easily - you could go 25 to 24 before, but now it is easier because it just changes the QT header.

They have a DVCPRO HD sampler that I grabbed 3 copies of. If I haven't given them away to friends, email me and I might mail you the disk....I don't know if they (Apple) are going to mail these out to those who ask.

----------------------------------------------------

can capture from a D-5 to final cut Pro. D-5 is an uncompressed HD format. I need to find out what res/framerate/i vs. p it can do.

------------------------------------------------------

the new DVCPRO HD decks are about $25000 dollars

-------------------------------------
more from German HD guru dude from Apple:

eCinema one2one device - how is that different from preview window? You don't really have the HD signal, the quality is not that good, and this really gets the HD signal... it converts it...about $3500 for the new one2one from eCinema, it is a pixel for pixel translator...it's more of a monitoring rather than a previewing solution.

It has a USB connector to adjust color lookup tables and stuff...converts HD SDI to DVI-D, you can change your gamma, change your lookup tables etc. There existts some software to change/make/edit those LUT, but what does it cost and how easy is it, and knowing WHAT TO CHANGE IT TO could potentially be a non-trivial exercise...

MIKE'S NOTES: the preview will be "good enough" for many. It'll be tough to cost justify the extra cost (as much as a G5) when you can get "pretty good" results alone. Only if you're doing crucial work will most folks pop for the difference.

Notes from Apple Booth at NAB-FCP 24p workflow, X-Serve, H.264, Pixlet, DVD Studio Pro 3 

more on FCP 24p workflow -

now there's a change so you can go 24fps to 25 fps more easily - you could go 25 to 24 before, but now it is easier because it just changes the QT header.

They have a DVCPRO HD sampler that I grabbed 3 copies of. If I haven't given them away to friends, email me and I might mail you the disk....I don't know if they (Apple) are going to mail these out to those who ask.

----------------------------------------------------

can capture from a D-5 to final cut Pro. D-5 is an uncompressed HD format. I need to find out what res/framerate/i vs. p it can do.

------------------------------------------------------

the new DVCPRO HD decks are about $25000 dollars

-------------------------------------
more from German HD guru dude from Apple:

eCinema one2one device - how is that different from preview window? You don't really have the HD signal, the quality is not that good, and this really gets the HD signal... it converts it...about $3500 for the new one2one from eCinema, it is a pixel for pixel translator...it's more of a monitoring rather than a previewing solution.

It has a USB connector to adjust color lookup tables and stuff...converts HD SDI to DVI-D, you can change your gamma, change your lookup tables etc. There existts some software to change/make/edit those LUT, but what does it cost and how easy is it, and knowing WHAT TO CHANGE IT TO could potentially be a non-trivial exercise...

MIKE'S NOTES: the preview will be "good enough" for many. It'll be tough to cost justify the extra cost (as much as a G5) when you can get "pretty good" results alone. Only if you're doing crucial work will most folks pop for the difference.

-------------------------------------

Graham Petty is the CEO and engineer and honcho...on what? eCinema? What?

-----------------------------------
H.264 is an industry standard, recently approved (by MPEG group? No?). It's a new delivery codec...they're showing 1920x1080p24 at 6.8 MBits (roughly 800 kb/sec) and it looks VERY good...compression time is a factor still that they need to optimize, so they weren't willing to talk specifics, but playback on a dual G5 was where it needed to be...rock solid....it is Apple's implementation of the industry standard codec. They were showing Scooby Doo 2: Monsters Unleashed trailers.

----------------------------------

Pixlet is intended to a working codec, a previewing codec, that will play back on any viable workstation. They were showing 1920x1080@24p. The engineer I spoke with actually encoded the content, it was at 61% quality.

---------------------------------------

DVD Studio Pro 3 - transitions that you don't have to pre-render (like iDVD 4), and HD to MPEG-2 encoding (no great trick there), import of iDVD 4 projects (way way cool for making more rich & complicated), GO BACK AND SEE MORE!

--------------------------------------

X-Serve notes:

-here at the show, they have 60 TB of storage here in the rack. (TAKE A PICTURE!!!)

-3.5 TB of storage for $11,000

-talking to System Engineer about X-Serve stuff, it will do 4 streams with REAL playback with DVCPRO HD, deliver 10 streams using preview function...implies it's sending LESS DATA from the drives to the computer when you are doing this kind of work...SEEMS to imply it is sub-selecting SOME of the data from the drives to do this...clearly need to follow up on this

Apple's Motion: Mike's notes from floor demo -"Hit play, mess around, make some really neat stuff." 

Talking to guy about Motion and title animation

every single object has apply modes a la Photoshop, has most of the new ones, works pretty much the same

can have a separate alpha channel like AE does

-has a lot of gestural UI controls - he does a doodle onscreen and it calls up different UI palettes

supports multi-layer Photoshop stuff...can tell it to bring in all layers or just one layer...can bring in all and drop out those not wanted too

doesn't do layer mattes discretely a la Photoshop files

Layer List - what AE calls a comp, Motion calls a Layer

can do track matte like stuff

when you bring in a Photoshop multilayer and drag it around, can move as a unit or sub pieces...so when it subcomps, can still directly click on all the children within the parent "comp" ...which is called a Layer here.

unlike AE track matte where it has to be one below, can take it out of project window so long as it's in media bin

Direct quote from the demo guy: "This is the one thing that is going to confuse the hell out of everybody coming from After Effects...AE any image is a layer, with us, images are objects, layers are comps basically....the cool thing is if you want to pre-comp something, you don't open a new comp, pre-comp it, bring it in, you just have a layer which is essentially a comp in and of itself...so I can have an arbitrary number of layers and organized it any way I want"

talking about layers (which can be a comp like assemblage): "it has all the same properties of any piece of footage"

You can manipulate a Photoshop layer in a Motion "Layer" (what AE users think of as a subcomp in a comp) and still discretely control all of it's subcomponents. Unlike AE, where there is lots of clicking and looking at two timelines at once to fix stuff.

it's realtime, what happens when you go beyond CPU abilities - it's all about your graphics card - it's OpenGL magic. NV40 is a newer graphics card and it'll push the capabilities of the program further-another direct quote from demo guy

"we have legs....every time a better video card comes out, the software will get better"

doesn't depend on us optimizing the engine a little bit more...as video cards get better, more layers, more filters, more resolution, more bit depth, etc.

the gigabertz bus sure helps

limited to the capabilities of the card - 2K texture size.

the R360 but next step has been seeded out to people, nobody's seen the 9800 Pro 256 MB version

It's an 8-bit tool at the moment. Video is the target, not going for film. But it will do HD, the demo he was showing me was 720p24.

The demo was DVCPRO HD, but it COULD be uncompressed footage, drive speed allowing, etc etc disclaimer disclaimer

can render it out to a QT if you wish

it has a native QT component...such that if you have Motion installed on your system, you can take any finished Motion project, open it up in ANY quicktime application and it will work anywhere QT will work if Motion is installed.

So can drop it into FCP and it all works, and works pretty well, since it is only taxing the GPU, not the CPU...which might bind in some circumstances, so you'd want to pre-render (or just render in FCP) to be SURE it behaved correctly.

But can drop it into Avid, into After Effects, into Cleaner, wherever you want it to go.

To share with anybody, you just render it out.

The whole file system library idea is XML based...so object can have animation attached to it as well, that you can use elsewhere...with all that animation and behaviors etc. attached to it.

POWERFUL.

It can handle relative paths not absolute file paths - drag everything to a folder and hand it off.

Behaviors are cool - is simulation stuff...for a "throw" or gravity, etc.

It has a keyframing environment, but doesn't give you as much subtle control as AE. No scaling of groups of keyframes, etc.

Realtime dragaround to define a motion path - like motion sketch

Like of gestural control for UI, Apple encouraging folks to use Wacom tablets in demos.

Doesn't have a smoothing keyframe tool to smooth them out temporally...yet.

Handles fields appropriately

Motion blur is present, but with a COST at the moment...It isn't per object the way AE works, it is full scene motion blurring.it does temporal oversampling, with the attendant cost in performance. Until they can access some OpenGL stuff.

It's fully configurable but with 8 steps it hobbles it pretty hard.

It's a broadcast oriented motion graphics application to supplement After Effects, "we think of it as a companion, we're not going after After Effects"

A motion graphics application for video guys, it's NOT a compositor (visual effects). There is some overlap but their goal is not to take on After Effects".

Want to be able to slap stuff together to keep the interface pretty simple. "Hit play, mess around, make some really neat stuff."

supports a lot of After Effects plugins. Sapphire works, Tinder, Delerium, even though they shag RT performance. Continuum (Boris) is in progress. Continuum will work better with Motion in the future he hinted.

MIKE'S COMMENTS:

So you wouldn't use this in a DI process at ALL. I liked the tight integration they mentioned with FCP, and that it allows realtime positioning, scaling, and coloring...but it's not a big heavy serious app...and it's limited (at this point the demo guy said) to 8 bit.

At first I was concerned this would instigate a war between Apple and Adobe...or at least continue to overheat their relations...but Apple is pitching this as a low end, fairly targetted solution for basic motion graphics needs for video editors...not so much a tool to take on AE, just to do some similar (definitely overlapping) stuff that AE can do in the lower end of it's capabilities scale. So for editors who need a quick title (woops, that in LiveType) excuse me, a quick motion graphic, they can do it here and it hands off to FCP quick and easy. Ideally, over time, these capabilities would be integrated directly into the FCP program and you could work directly in the timeline, even if it were to then be "collapsed" in view. Hmm, thinking about this, maybe not, since I might have dozens of layers in a comp, and you wouldn't want to have all those layers "live" in the FCP timeline for the rest of the project.

It's NOT a (full) After Effects compositor. It can do some of the same things, but is a small subset of AE's capabilities. I didn't ask about subpixel positioning and hyper accurate scaling...I need to go back and ask. I'd expect it to have good scaling, because that is a common need in OpenGL applications, but subpixel positioning? Maybe, since each object/layer is essentially a 2D piece of geometry being positioned. I'll check. So for keying, compositing, any kind of visual effects work, or for really detailed motion graphics work, this is NOT the tool to use. Stick with After Effects for your motion graphics and medium end compositing work (or combustion if that's your preference), and punt to Shake if your needs are that high.

I predict this will eat into the market for After Effects artists. Editors who want to tackle this more approachable (and far better integrated) tool will be able to do certain motion graphics tasks. They'll be encouraged to tackle this because of the tight integration with FCP, the approachaable user interface, and the very friendly price point. So smaller, simpler jobs will stay inhouse rather than going out to motion graphics freelancers (like me). So while some jobs will be lost at the low end, it will encourage further use of motion graphics in the marketplace...which may then prompt editors to punt to the experts when their needs (or deadlines) demand more than they can/want to do themselves. So they'll call one of us...or punt and do it themselves. Remains to be seen.

The inclusion of After Effects plugin support seems to be more of a marketing checkbox for support - AE plugs don't really fit into the workflow you'd want to use in this thing. They claim 90 or so filters...or was that 90 realtime filters? But plenty to get a lot of interesting stuff done. Those realtime filters are going to have to be written as OpenGL filters - in other words, written for the GPU (Graphics Processing Unit). After Effects plug-ins are written exclusively for CPU processing.

This is a clever bit of work, an extension of what was started with Quartz Extreme a year or two ago. There have been applications that do similar functions in the digital lighting world (concert lighting systems that use video/image projectors mounted in an aimable, DMX programmable lighting-style head), but none with this amount of robust compositing capabilities...and certainly none with this slick of a UI.

Best summed up by the demo guy:

"Hit play, mess around, make some really neat stuff."

Livetype - no export and import, it's no making a QT file, so now whenever you save whenever you save in LT, that link between FCP and LT is automatically updates in FCP.

You're get a Flash about the file being offline, that is it breaking the link and relinking to new media

Yes, you have to re-render in non-realtime, isn't like Motion.

Same integration with Motion in terms of live updates.

Within Motion, when you hit save and go back to FCP, it updates that link in realtime. Any effects in Motion translates over to FCP. It's a one way link - you can

You can import projects - make a project within Motion, Save it, import as a regular file into FCP. It's just as if it's any clip. all that stuff still stays as if it was in Motion. When you render it out - you can't go in and change the effects in FCP from that Motion file.

Something changed in FCP or Motion, any time something changes on the Motion side, when you hit side, it has to be updated in FCP it will blow away the rendered file.

in FCP, tell it you want to change your LiveType, it launches LiveType (if not open) and opens the corresponding LiveType file, then ou make your changes, SAVE IT, go back to FCP, it updates the link, and signals LiveType to start rendering it's project....I think I have that right....I might be wrong....but in any case, there is good handoff and communication between the two.

Some Random Notes from NAB Digital Cinema Track 

Notes from Digital Cinema Symposium:

talked to Josh Pines from Technicolor, he said lattitude is still his biggest issue with HD. All the color tones, and blocking, and color gamut is not so much a problem, but that lattitude range is still the single biggest deal, he said cameras can capture extended dynamic range but the manufacturers don't want to tell you about that because you have to get in there and twiddle with it.

Mentioned the Orphanage guys and he said he helped them get started since they'd known each other at ILM.

--------

from a seminar: Rob Hummel from Technicolor on their restoration efforts: technologies that affect the telling of the story, the public isn't even concious of the changes, what we're doing isn't affecting the telling the story, just helping to tell the stories in a more efficient way.

----------

Digital mastering for film and digital cinema release:

THIS ONE IS 2 HOURS 45 MINUTES LONG

------------------

eTreppid is working on hardware based codecs, integrating that into video editors, lossless compression, integrating it into the backs of recorders, use it as a working codec all the way through - both an acquisition and editing format, and you could make the files smaller (still lossless) at the end of the process for archiving

--------------

new stuff at ATI booth - Xileon - HDTV Wonder for PC, ATI THeater something, digital cable ready and ATSC, low cost alternative to cable and sattelite, expandable home theater products, it's where they're going, PVR/DVR networked video recorder

--------------------

Sony has a large HDV section I need to check out

Monday, April 19, 2004

Update from the show floor 

OK, so I popped for internet access from within LVCC for wireless access. It's 4:40 and I'm finally stopping to have lunch.

Spent a bunch of time at the Apple booth seeing XSan, talking geek about uncompressed HD rates for XRAID, talking 1080p24 workflows for HD. The new DV-HD (or DVCPRO HD, same thing) format is really, really nice. Footage looks good, RT looks good, lots of good stuff. The only downside is that, once again, just like HDV, there is no 1080p24 option. 720p24 yes, 1080i59.94 yes, but not 1080p24. Someone said there was a technical reason for this, but it sniffs of market protection...but I don't know.

The best new product I've seen so far for our type folks is Lumiere, a piece of software to let you work with HDV. Based on preliminary demos, I like it better than Heuris' Indie HD Toolkit. I need to check my notes and verify a few things (I've been voice recording all day, 76 files), but it looks like a good solution.

AJA is also in late June going to ship a very nice HD card, the Kona 2. A 4:1 codec called Qrez sounds interesting, it does real time VERY HIGH QUALITY downconversion to SD if you are working at NTSC or PAL framerates, 12 bit analog D/A monitoring, good stuff.

I've seen one in the Apple booth but haven't learned more about a gadget called one2one, it sounds similar to Decklink's HDLink. HD SDI input converted to show on an Apple Cinema Display. Made by eCinema, I think.

OK, that's all for now, more later.



No time for links now, I'll get to'em later.

Sunday, April 18, 2004

Day 2 Panels: My Raw Notes 

Once again, my horribly raw, incomprehensible notes from panels, this time in one miserably long screed:

OK, I came in late today.

Diving right in:

Advantages of Digital:

copying, changing, controlling, delivering

(CONTROLING - DRM!)

Challenges:

Immature, expensive, complex, impermanent

immature - disadvantage now, but that's temporary - Moore's law will help.

in the first 20 years of film, there were all kinds of early film gauges. Eventually, standards work themselves out

first oact of SMPE, standardized 35mm 4 perf

in film there was one worldwide standard...and within that there was room for other formats from silent 1.33 to Super Cinemascope 2.35




PICTURE 1: SCHEDULE FOR THE DAY

SCHEDULE 2: What Do Cinematographers need? panelists

Allen Daviau - Busy, ET, Empire of the Sun, and Van Helsing (bearded guy I saw yesterday?)

Lindenlaub - Made in Manhattan, The Haunting



Daryn Okada (Japanese guy from yesterday") Texas Rangers, Lake Placid

"Digital Cinematography MUST abandon all links to television" was said 6 years ago

Steve Poster said we don't know how we're affecting the audience - In The River of Conciousness, was written in NY Times:

Does our brain parts work like a motion picture camera?

HE'S TALKING ABOUT HOW OUR BRAINS PERCEIVE REALITY...ARE DISCRETE FRAMES GRABBED AND INTERPRETED INTO A COMPLETE FLOW?

Is conciousness tied to the individual perception of still images?

As people in this industry, wse need to find out how we are affecting our audience in this big change of how we're doing motion pictures

He's (Steve Poster) working on that over the next year with SMPTE

Cinematographer s need to follow the process from start to finish as they always have, now it's different

"The Horror of Video Dailies" and why they are so bad, and what can we do to invent a system for digital dailies to serve the purpose of alll the artists and all the people working on the film "Watching day after day on 1/2 inch video is a crime"

need end to end device independent color management, ubiquitous, industry wide

Allen Daviau - What do Cinematographers need? They need an open mind. He prefers film. It will be distributed on a LIMITED basis to be shown digital. Majority of folks will see it on film. They shot on film, evaluated on film (dailiies), but new they were doing a DI process. He's optimistic to it, thinks it looks great and likes the options available.

he's shot 2 shorts and a number of tests with digital cameras, but "I will rejoice in the purity of film"

finished shooting in July of '03, 2 week blue screen shoot, eFilm is where he's been working, with Steve Scott at eFilm, evaluated different sequences shot under different lighting conditions, but all tied together in the DI process

2 solid months in digital intermediate phase on Van Helsing

need in their contract to work all the way through, since sometimes shots are getting completed over a year later in the DI process

Daryn Okada: film is a proven tech, we know what the results will be, predictable. totally open to digital tech, Imperative that we (Cinematographers) follow a film from beginning to end. Every shot rejected by editor, director, & tudio were shots he never shot before they were composited...he didn't see them until they had been approved by others. "What woke my eyes up to participating" Just because it was scanned in at 4K doesn't mean it will emulate film on the way out...bit depth, linear vs. log, etc. affect it all. Just doing a simple dissolve an issue wif there was not someone with a trained eye dissecting that. Midtones blasting out, blacks crushing, taken us out of a dreamworld into CNN.

WAS THE PROBLEM THAT IT WAS WRONG, OR NOT FILM LIKE?

They need a cinematographer's eye to tell what's going on.

He was involved every day (208 hours a day, in the last month spent 48 hours in a row at the lab, NOT doing color timing) a 4 month operation. If he hadn't checked that final print last week, it was not the look they set out for. Need a cinematographer there to complete it, and make sure everything is the way it was supposed to be.

He DID NOT GET PAID for all that work during post, and had to buy his own lunches, too.

moderator sez: The only way to assure a quality finish, gotta have the cinematographer to time the scans before the scans go to the compositor, so they had abenchmark of what the movie should look like. Gave them a benchmark that was close enough even if not perfect final.

You have to fight with the editor to get them to call you to have you look at them.

Frustrating to come to a final timing session, with 10 different houses all doing thier own slightly different thing. Do you keep the effect shot fucked up, or fuck up all the original photography around it to match that digital shot?



IT'S A GOOD IDEA TO HAVE ONE SCANING HOUSE

Need a method to track color and look all the way through. Kodak is working on something.

Needs to be ubiquitous, non-proprietary standard, internationally.

MMichael Goi: Daryn is high tech, Mike isn't, have shot 2 features on high def, shot one on DV, done a feature through DI, shooting a movie tomorrow morning. Until 3 days ago it was 24p High Def. Shot on beaches with African American people. "in my opinion you're making a serious mistake." Now they're shooting 35mm.

Need to guard against hype. Know which tools are best for which job.

every producer has read all the same marketing stuff "cheaper, easier to light, smaller crew"

he educates people as to what works and what doesn't. Each medium has it's own strengths and weaknesses.

Steven was mentioning dailles. He shoots on location in foreign countries. He know has it in his contract he gets to print 3-4000 feet of film so can know what he's doing.

Dailies on 1/2 inch were to bad were going to fire the camera crew, had to get digibeta to see the difference.

Such a gross misrepresentation to see low res format with something that is not standardized.

Sephen: by doing the cheapest daililes you can, you're costing the project money. Everyone gets a sense of what the movie will be TV sized, not MOVIE sized. They lose the sense of the scope of the picture and tell the story in a way that is ineffective for that movie. Closeup head shots are too prevalent in today's movies. Hurts a production to not see dailies in a larger format.

For HBO, he brought in a digital projector. Watched highly compressed VHS dailies, but showed it in trailer 6 feet wide.

Needs more discussion on dailies other than film (fine in some respects), need a way to have hi def digital dailies projected that we know will represent the look of the movie. "I could go on about that for years."

Walter Lindenluab - approachef for a film in Africa about the genocide there, European production, wondering whether to shoot on video, would rather shoot 16mm. Remote location, not much tech support, rather have reliable tech. 16mm stocks are so good, blowups are MUCH beter than they were 10-15 years ago.

Working with HD and cables and remote screens is so foreign compared to .

Finding out that scans were done so differently on monster huge films (3000 prints, big summer release) that they can't be successfully composited.

Thinks it should be in their contract that they are paid for their time in post production.

He just did his first film with digital dailies, talked to them into doing Digital Intermediates, using Kodak Look Manaagement System, enabled him to control the look of the whole movie through all the stages.

color HD master for the test screening, got thorugh the whole thing in 1 day instead of 2 weeks. That Kodak thing helped, presumably.

Need to get everyone in post production used to that process

In the front end, everyone talks about the look they want, but can be lost in production. Kodak system "is definitely the way to go" since everyone in post production knows what you want to do, where you want to go.

If you had a digitla video camera that used your lenses, that was tough and built by cinmetographers for themselves, would you use it? "In a couple of years."

MODERATOR:

Kineticamera designed and built by cinematographers who have an end to end system. Arri has an exciting camera. single chip cameras are pretty exciting.

AT the other end of the spectrum, large capacity data streams,

Gil Hubbs: Cinematographers (CT) listen to the director to get their vision for the movie. Often important to simplify and not get lost in details. Digital production today is not defining a simple process. Easy to get lost in forest dodging trees. Give us something we can see the picture without taking a room around with us.

I'm working with a company that has a camera in the Defense industry that is AMAZING, 3 inch chip, 100TB of storage, The shots they're doing are all defense related that he hasn't seen. (I'm guessing it's high speed imagery)

MODERATOR: How are we going to deal with large amounts of data? Storing 150 TB of data to store a motion picture? To store output of a high end camera? How do we move that stuff around?

panelist leaned up to say: "PAY CINEMATOGRAPHERS FOR POST!"

Mike's comments: roles are changing in the working world now that digital is on the scene, and people need to be reimbursed accordingly.

===========================================================================

next group:

NEW DIGITAL CAMERAS

next 3 pictures are the list of moderators

next 2 pics are of the panelists

ASC has been very proactive about our tools should not be designed in a vacuum

Digital Cinema is being created right now, so it's the most crucial moment, most critical point, is RIGHT NOW.

Michael Bergeron Panasonic - focusing for the next year on getting the systems to be easier to use, something more transparent, something a DP without an engineering degree can walk onto a set and sue the camea and folow the process through.

in name of increased res or whatever, if we through out gains from going digital, we haven't really gotten anywhere

CineGamma - 2 things - emulates a film look, emulates a film processing workflow, makes a tape that resembles an interpositive - you have more dynamic range, to get the most lattitude from the chip, can do gamma correction similar to what you'd get from scanned in film.

On set, can light like you would film in terms of ratios etc.

Mike sez: Does this make sense? Better to emulate like film, or to optimize for the best results of the gear? Learning the new stuff, SPEAKING IT'S LANGUAGE

Gamma Box - a gamma corrector that applies a gamma corretion to what's going on the tape that looks like an interpositive (dark and not pretty but not what you're going to see in the end) Gamma Box lets it correct to look right on the preview monitor THIS SOUNDS SMART

That same gamma corrector can be used in the post process as well, works with HD-SDI from camera into the rest of the stuff in post production. Also has component outs, putting that gamma correction into some of their monitoring products. THERE IS A PAPER ON THAT AVAILABLE ON THE WEB OR EMAIL HIM

Gamma corrector will be replaced by plugins in your non-linear editing system. Ideally, best to go directly from native compression format of camera and work in that form, then apply the gamma in your NLE. They can't do that...YET.

THIS SOUNDS SMART. But we need to be able to work in native codec of camera...not happing yet.

VIPER FILMSTREAM UPDATE

from Thompson Broadcast Jeff Rosica & Mark Choilis

precolor corrected RAW images

they aren't getting into the film vs. digital debate,

bandwidths they are dealing with today

see pics for

get about 2 1/2 extra stops

Viper does 4:4:4 filmstream now

COME TO THE BOOTH

badass 1K res Fincher commercial was shown

NEXT UP: JOHN COGHILL FROM DALSA

Important for CT is to break from TV roots. Dalsa has no TV roots, so easy for them to do. They do 25 years of high speed sensor for military and industrial stuff. They got a team that did high speed digital still stuff.

needed to be compatible with standard lenses, optical reflex viewing system was crucial, " a lot more to this than pixels" dynamic range, etc.

Over in LA earlier this week, showed some rough cuts from a short project that wrapped in February, general impressions were extememly positive, from an image perspective quality is there or almost there, they consider the camera to be pretty much a done deal from the technical perspective, NOW IT'S ALL ABOUT WORKFLOW to handle large amounts of data. In recognition of that, they decided to provide for multiple outputs, 4K to hard disk output, having efficient 4K workflows, they also allow dual and single link HD outputs.

Their camera is different because working with 35mm lenses, wide dynmaic range, so it gives you 4K high dynamic range source that it downsamples to HD...just limited by the gamut of the HD 709 palette.

Collaboration on workflow is differnt

Digital Video is different from digital data workflows...need to adapt to that.

ORIGIN COMMERCIALLY AVAILABLE BY THE END OF THE YEAR.

NEXT UP: MICHAEL KOPPETZ FROM ARRIFLEX

quick to differentiate himself from the video world, to say they are from film world.

"What would happen if we could put together the tried and tested tech of film cameras with modern technology" alternative approach to HD acquisition. It's a tested. Large format CMOS sensor, same image area as 35mm, same optical characteristics for depth of field etc.

has an optical viewfinder

D20 project I think is what he called it

you can see more than the image is being capture, let's you see stuff out of frame through the viewfinder as you shoot

the look and feel is like a 35mm camera

for productions, same economic advantages as far as support from rental houses and on set workflows - same film crews; they are a rental based business, so they are trying to keep economically viable for the rental companies to leverage their stuff.

The sensor has 6M pixels, 3000x2000 pixels approach. Can extract different frame formats from that size, can downsample to 1920x1080 (it has a color filter layer mask over it). IN this model, 3 HD SDI interfaces: single or dual link SDI, can also output raw data straight from image sensor for processing later on.

will see tech demonstrator at show, the testbed to verify their concepts. Working on a prototype (avail for testing 2nd half of this year), looking at first product intro for TV production for beginning/middle of next year

ASC Tech Committee will have agnostic interactions with all of these people

NEXT UP: JEFF Kreines: "I'm a bit of a Luddite, I think SMPTE has been downhill since they let television in"

has been hacking cameras since the 70s to fit their needs

started looking at what was on the market, the existing digital cameras were not something he could imagine working with.

"Perhaps we could design a camera that would be user friendly"

HAS A HAD CRANK OPTION

use of film lenses

sensor agnostic and resolution agnostic

10-bit log UNCOMPRESSED images

Designed to handle up to 35mm aperture size

they are making high def and 2K now, looking for 4K and 35mm next year

making the head small wasn't too hard, making a portable array was hard, it was too big.

a fiber tether was too much.

Decided to attach the array to the camera, 480GB

Wow!

Workflows: can play with anyone's system, has a 4K RT color corrector, also has a scanner designed to work with it out to stuff.

Can get 4:4:4 out of camera, can get data out in Cineon or other choice.

Camera is tiny, DV camera sized, with mag it is size of 80s VHS consumer camcorder. THIS SOUNDS REALLY GOOD.

SEE THIS AT THE CNL GATHERING TOMORROW (DON'T HAVE A BOOTH..SHIT!)

f950 camera shot 4:2:2 HD CAM this must be the sony guy

keys very nicely...

complete RGBB workflow from capture to storage and si display

Cameron decided to shoot next major feature with 3D digital cameras (heard of it elsewhere)

lot of stuff about more affordable

HDV HE'S MENTIONING...going to support 1080 lines...that's way good news...but I'm sure it'll be interlaced, right?

Sony booth on South Floor

NEXT UP IS PNS TECHNIK

a lot of commercials that are out there that are using this

a lot of shows made for television using it

first movies in theaters in the last few months

in last 2 years learned if digital cinema is to be viable, it has to merge with teh realities of the market (can't take too big steps?)

realtime printer - capture 35 digital format images, also able to print out in realtime gain. Workflow will be closed again. From tape to film in RT. That is a goal, not a product. At IBC they'll show it. But he defines that as 24fps with a color matching system

FROM LOCKHEED: 12 megapixel RGB, 14bits effective, 12bit recorded, 49x62mm sensor (more like 65mm film). Right now focused on dynam ic range problem. Until we prove the dynamic range is really there, 10 stops with 90 dbs noise. 2 to 2 1/2 years will have 4TB recorder about the size of what Jeff showed...then he said in that size they'll have 10TB

MIKE SAYS...this proves that "the film mag" problem will be solved.

question on floor - include metadata into datastream, standardized format please! Floor guy wanted the serial number info in the footage

3 chip prism vs. single chip Bayer for large images - a good question for future

Kinetta.com for more info

==========================================================

NEXT PANEL UP:

called Establishing The Look

Leon Silverman is moderator, exec VP at Laser Pacific, dir of strategic biz dev and he works for Kodak. Kodak bought Laser Pacific.

NEXT PIC/S ARE OF PANEL

the look used to be defined by CT, director, dp, lab tech, color timer, et\c.

now any schmoe can establish a look with Photoshop

Wolfgang Lempp - founder Filmlight - did post production for 15 to 20 years - making things happen & look right is his focus.

trying to apply what he learned

further you explore what a look could be, less certain you get about what it could be

other things far less known need to be looked into

the color gamut of film that has been held up as bigger and better....that is not actually the case in the last number of difficult images we see.

flicker plays an important rol in how image are understood

12Hz alpha waves in the brain....this guy is way out on planet

JOSH PINES from TEchicolor Digital Intermediate, member ASC tech committee. From visual effects world...THESE WERE THE GUYS WHO LEARNED IT FIRST

Deal with acquisition and digitally color correct entire films and master to DVD, VHS, HD, etc.

is Digital Acquisition better than film? What are the tradeoffs? a lot of interst from CT as to what they can do on set to set the look that can be translated back to the facility doing the post & FX

Bill Feightnet, from eFilm - now with digital manipulation hype, what's new today is more possibilities to change images, but more important, we've had for years a reasonably consistent display environment in theaters. Now there are a lot of ways of watching it, but have 35mm film, then to home, then to digital display, each with their own unique properties and different display environments.

"The Look" isn't just color spetrum - spatial components like what's sharper, but also viewing environments, but how it's temporally updated. Just need to recognize that. DPs, CTs, art directors, need to establish that look and What That Is so that when we distribute the images, what parameters are important and not - chromatic adaption - incandescent vs. flourescent vs daylight etc.

need to put a stamp on this work, metadata to define how it should look, and for future generations how to archive it, attach that metadata

Daryn Okada - what we need for look definition - a system or language going beyond all the special things we can do, (lense filters etc.), "some set of definitions that are interchangable between facilities and manufacturers that can define what we meant when an image is created or captured." for images to flourish in the digital world, there has to be a common system bewteen images and formats and ideas so the idea can be conveyed right to the very end

KODAK Peter Postma (young guy). how to face all those choices in terms of how it'll look in cinema, DVD, etc. That communication for what from DP got/wanted on set to match what the tech

Kodak Look Management System - enables CT to capture an image (or film scans), emulate them in software with telecine controls, Print Light, etc. so he can email that to lab timer in lab, or other working pros to maintain the film look.

Audiences are used to maintaining the film look, need to support that, but NOT limit the artists to film space

BASICALLY LET YOU MAINTAIN EXISTING LOOKS AND STANDARDS, WITHOUT KEEPING NEW POSSIBILITIES OF THE NEW TECH CONSTRAINED

Walter the ASC guy - 2 schools of thought - capture all info possible so in post people can decide what the movie really look like - just have to have all the info recorded

other school of thought - "this is how our film should look like" and establish that look through the whole program and maintain it consistently.

The tech can change the look for a shot to make it "better" without realizing how that fits into the big picture

the selling point to studio was to save money by establishing the look early on instead of starting from scratch

in DI sessions, it can take 6 weeks to 2 months to do the process, because everyone thinks they are an expert.

Why not establish the look early on, and therefore not get lost in the creative process when everyone else is jumping in?

Why not establish a look and carry that through, communicating throughout?

MODERATOR: How to make sure people are looking at the same picture all the way through on CRT, laptop, film screen, etc.

This is a color matching issue. Apple get involved with ColorSync for film?

What good is it if the audience can't see what you got/want? Need to start knowing where the image is GOING before you start tweaking on it.

Print film is an established look, that is the STANDARD. Not that that is the ultimately desirable thing, but it is a standard to work from.

Most cases, we're trying to match the one standard - print film - for CRT, etc.

That may change over time

We've chased the print film standard with digital projection, DVD, etc.

Have to consider how to standarize all this. There are only a few projectors that can accurately and repeatably applicable for different viewing environments.

COLOR MANAGEMENT AND CALIBRATION WILL BE A BIG DEAL IN THE FUTURE AS PART OF THE WORKFLOW

Digital is here. Multitude of display enviroments is going to get larger. Establishing a system to manage that is going to be really important. We don't know all the issues involved yet.

In the video world, there was a reference. A reference monitor....even though Fry's has all the TVs that never matched.

We're NOT talking about baking in a look on the set that limits dynamic range and choices....we're talking about capturing high dynamic range untouched, but somehow have the ability to set the look on the set, and have that follow back to the post house. How do we know what that monitor will look like? The best thing to do is characterize the thing you're looking at. You have some chance of knowing what the guy was after (hellloooo, ColorSync!!!!).

If we have NON-baked in look definition...we can't always chose the post facility we're going to. And having one manufacturer's on set look definition system and another manufacturer's color correcdtion system could be BAD. Tech incompatibility. Print Lights might not translate literally as a ssystem, we need a minimal set of color correction utilitie that could be shared by all the manufacturers. Sign on to something that definies a minimal set of things so the DP on set to use a box to define a minimal look that can be carried forward.

Should base this on printer lights or something understood.

need a CDL - Color Decision List

but before that can be done, a number of steps to be sorted out that haven't been. Starting with display, it has to be calibrated. Even a calibrated display (according to recommended procedures) it won't be the same as another calibrated the same way. The intent/desire to make those perfectly consistent is a valid one, but it's hard/impossible to get display consistency among the same display types, then amongst different types of displays.

Display environment would have to be calibrated as well - what is your viewing environment? What is peak light, and what color temp is it?

14 foot amberts is right cinema brightness - but on monitors, don't have standard way of doing this.

THIS IS ALL THE SAME STUFF THE PREPRESS INDUSTRY HAS GONE THROUGH. ALL THE SAME COLOR CORRECTION STUFF. You have to have tags for monitor, image, and environment

I SHOULD THINK ON THIS!!!

the challenge of the ubiquity of the standard - visual effects have their own methodologies for their own look calibration technologies. How does PDI and ILM get their shots to match each others?

NEED A SYSTTEM THROUGHOUT THE WHOLE PROCESS to carry stuff all the way through - capture, edit, post, effect, color correct, DI, back to film

If you mandate a standard now, could stifle technological progress. Need to be a goal to work towards.

We all don't know exactly what it means to define a look.

Differentiate look management from color management.

Intent vs. color management.

Background was changing all the time as he shot stuff. There was a bush backlit by sunlight, but sunlight varied. he could have color timed the overcast moments of that scene to match the cloudy bits, and tweaking just a part of the image to get it to match. If you have big epic movies relying on day exteriors, knowing you have digital post production allows to you keep shooting.

What happens if you have a cloudy version and a sunny version, digital helps you fix that more but not perfectily.

There is room for keeping the traditional techniques to keep it right.

moderator: Will look creation tools become so ubiquitous? Will dolts messing with Look Treatment marginalize cinematographers? When someone can change the look too easily?

Daryn sez: he's had it happen to him when unqualified people make comments. When people who shouldn't be playing with the image come in and mess with it when it's not calibrated and they were fooling with it (boss messing with Photoshop in his laptop, emails it out and says "Hey, I like this! Do it like this!"

XYZ will be part of a useful solution, but not the Big Answer. You can't see XYZ.

==============================================================

NEXT PANEL: CASE STUDY, VAN HELSING

Director: Stephen sommers
Alen Daviau DP

Steve is the DI colorist - worked on LadyKillers and others

Van Helsing - EFilm Digital Intermediate Shots Breakdown

gonna drill down into more detail on this got done. Panels have been high level, this'll go deep. YEAH....

Picture comes out May 7th.

Bob - about 2 years ago in early pre-production, we looked at ways to bring color timing a lot earlier in the movie. In FX heavy films, materials keep coming in to the very end. 1200 shots, about half the film are FX. Can't wait until you have the cut to time it all.

Talked to EFilm about moving that phase much earlier into the process of the movie.

They wanted to scan all the Selects.

RAther than film dailies, 2K scans from film.

That wasn't feasible, which was moving previews and interactions with FX companies could be done earlier.

Picture was scanned in November, gave'em 5 months, gave a lot of lattitude to make choices as needed

make color timing process an integral part of the film and not an afterthought.

It worked out well

Allan Daviau knew what he wanted. Do you shoot to get maximum lattitude (a little flat so you can manipulate)? Or do you just trust your gut and shoot what you think it should be beautiful and compelling? Steve says he appreciates the gut of CTs and it's OK to trust them, let'em do what they want, then work with that...IMPLYING IF AND ONLY IF THEY KNOW EXACTLY WHAT THEY'RE DOING do you set them free

DI is really film's best friend he says, since they scan very carefully to capture full dynamic range (that digital acquisition doesn't have yet) to get incredible detail in a sky that's blown out, etc. Rebalancing an image in a way that wasn't possible.

Details in negatives that weren't there before can be rescued sometime.

Ballroom in Dracula's castle shot from the movie:

wow, looks sumptous and gorgeous. Better than most film displays.

Great highlight/shadow detail

So why use DI?

from original scan: full dynamic range, doesn't look like a timed daily

Printer lights only: he wanted it to LOOK like it was candlelight, made it darker up top to feel like lit from candles

they made a soft matte to darken the top. The locked their darkening matte to travel with the image, not the lense/camera frame. So they tracked a darkening matte to the top part of the scene.

they did another mask to brighten the botom half

Wanted to pop certain color ranges -isolated reds, greens, then travelling mattes, to only correct the part of the imge you're interested in.

did LOTS of careful, discrete, selected color correction WITHOUT any paint work - all keys and soft, generalize dmattes

Ballroom 2 shot:

did a big soft matte to accentuate the key actors, the orchestra, etc.

HEY! I CAN DO ALL THIS! HOW DO I GET INTO THIS!!! FOR DI WORK!!!!

They are STILL making changes, made some changes "the other night" even though it hits theaters on May 7th. WOW.

Farewell at Sunset:

a scan with COMPLETED EFFECT - you usually have limited lattitude. But since had DI had choices

made a soft matte to dim down the sunset's contrast, wanted to make the shot flatter, pumped the highlights on the waving grain, etc.

wanted to bring a bit more sun into it

too much flare on mountain range, to did an irregular matte to correct there.

Another travelling interactive matte to fix the grain pump

WHAT SOFTWARE ARE THE USING?

CT still had final say on the way the final looked.

They have hacked the code on their projectors and tweaked their software to emulate film proejction (using digital projection)

JUST THIS WEEK is the FIRST time they are seeing the filmout from DI. 3 weeks before release is the first time.

Film has image at the toe (deep blacks) that comes out AFTER the DI has been done with digital projection

AND MORE...more shots where they were doing AE type stuff to correct little tiny things...it's all been always possible, but these guys had the budget to get in there and tweak it.

Low end effect houses can now call themselves DI assitants if they know color correction.

MIKE'S NOTE: none of this is particularly new to me...but perhaps it is for them.

In one shot, Hugh's nose was too red - fixed it.

Blurring the line between Da Vinci and Inferno. How to differentiate between compositing and DI? As if you had an Inferno with multiple layers...but in their model, each layer is merely another slice of the source treated differently.

DI is (presumably) cheaper than FX shots.

One of the things in the past, because movie isn't timed yet, a little at random to figure out WHAT the FX houses should be matched TO...so in this process it allows the FX to be done "loose" because color and density can be fixed in DI.

NOTE TO SELF - BUY A SIMPLE POLE STAND FOR CAMERA

It's the first time he's been involved in DI where the movie was so in flux. They'd have a previs animatic so they could color time it to fit in to what they had.

Takes judgement and restaint to figure out where to NOT fool with it. Prioritize! The deadline always looming. Can't have reel 1 looking INCREDIBLE the day before it's due and have the rest of the reels looking "not so fresh."

They make multiple passes - like my old boss said "like an oil painting - do the broad swathes, then go back and fill in the detail."

Da Vinci will need to gain these skills and capabilities over time is my (MIke's) opinion.

Could ID early on if their were problems with the way the FX were fitting in. Were able to interact EARLIER with the FX shots.

They'll give them the timed file to work work, to improve the quality of the compositing or whatever the issue is.

An early "day for dusk" shot was done to test something, EFilm developed a look PRIOR to shooting the full sequence. In the middle of SHOOTING, DI was part of the process to help define what the film would look like.

They were able to DI a shot heavily to get it "just so" and hand off that SAMPLE look to the FX house to say "do the rest of these/the rest of the scene like this one." And that made it all go more smoothly.

Allan is very clear about what he wants, shot what he wants. So much EASIER to shoot something that looks fantastic and then improve it. "You can make bad stuff look better, but you can't make it look great."j

So that kills the thought of shoot everything super wide lattitude and you can fix it all later. Not only would that not work, it also would take too damn long, and wouldn't get the great results. Shoot great stuff first, then make it look better.

********

The demands of the iindustry is changing: used to be edit the whole film, THEN time it. It's not going to work that way any more in the future, especially for big stuff.

Since they knew that VFX wouldn't be finished until VERY late in the process, and would have to be timed, better to have this going on throughout the process.

"We have no intention of trying to kill film. The DM process have some serious strengths over conventional film timing" -Bob

"It's a great hubrid

They want traditional timers to come and see the images that they are working with.

Proprietary tools, nothing new compared to combustion and AE...but they can handle the WHOLE MOVIE..and do things like "take this whole scene, and pull it back a point of yellow...and animate it across the whole thing"

It isn't any new speicific effects capabilities, they are creating tools to MANAGE the color corrections of the whole thing.

GOT A CARD FROM WHITE MOUNTAIN.

====================================


=========================================================================

SESSION 2: walked in late

nonlinear, overlapping workflow now - can keep editing up to the end

Digital Workflow

takes 3 weeks to make release prints, so that is the REAL deadline for projects

-15 years ago did a TV show Michael Mann was editing, he found out he could keep doing edits and cuts down to satellite broadcast.

His predictions is that you'll see that 3 weeks dissapear. DI will go up until several days before final. It depends on what the final digital distribution methodollogy will be.

There will be more versions of the final project - more like what is done with TV.

More complexity with the digital realm:

FILM vs.
scanned elements, HD, digital cameras, CFG, different res, different file formats, different color spaces for digital

More creative choice: Acquisition choice: ability to mix media, seamlessly

Digital Intermediate: can do things not otherwise possible

Digital Projection

THE ROADBLOCKS:

conforming and editorial in DI: no standard for workflw, need to go beyond CMX 3600 (a 20 yr old format)

need a standard to describe color correction - SOME WAY TO CAPTURE AND SAVE THE METADATA

NEED A PIPELINE ARCHITECTED BY A VENDOR THAT CAN CONTROL ALL THE DIFFERENT PIECES

need better support for higher bit depths, higher resolution - need to be resolution independent, support open ended bit depths. Negatives have MUCH MUCH WAY MORE range for than a final print. Needs to be supported in software


if you have a 2K master and a 4K projector comes along, how far back do you go to recreate it?

there will be new software tools for DI on the show floor
hardware incompatibilities

Lustre, BaseLight, Da Vinci, etc.

new hardware, too.


hardware incompatibilities

-petabytes and yata (yadda?) bytes are next after terabytes

more changes in the last 15 years than in the prior 50 years: more changes will happen in the next 3 years (TREMENDOUS MOMENT OF OPPORTUNITY HERE)

The New Workflow

Data management slide from DALSA website

Possible to have 193 TB of footage from a movie.

...and that's before you start making new versions for effects, color correction, DI, etc. 1 petabyte could be possible on a big movie

So there is no way to define and keep this stuff. What happens when you get all this massive amounts of data.

These new cameras as data hoses - what happens when you turn them on?

with a million feet of OCN you could end up with than many TB of data


===============================

next panel - cutting room panel

picture for next panel members

Digital Cinema magazine guy on the end

The Role of the Cutting Room in the new Digital Post Production Process

Sheigh was an editor for cretiveplanet websites

Cinematographers have done a good job in the first wave of transition

Apple showing 8-bit Final Cut Pro HD

Norm - doesn't see editing as a defense against trouble...he sees editorial as the hub of the deal. Pixar has editorial as the hub of the wheel - the idea people come into the edit suite to pull pieces from other movies to edit to get the feel right

development sees that

it comes back to editorial and they edit some PShop stuff

The editing servers are the hub of the wheel where everything works.

"The real reason people shoot movies is so that I can cut them."

Bruce - create in editing room, treat it as a hub, a technological hub of teh productin and post production

very large computer server networks that cna ingest dailies, info from set (color management stills from set), c

that data could be coming into these servers in all kinds of formats

start smaller and be ready to expand

train them on how this process will work

you have to lunge forward with tech and deal with what you've got, can't wait for it to all be there.

Editing in High Def is a big leap formward to see what you'll have to work with.

Work with small digital projectors, looking at high def on a 6-8 foot screen. They aren't editing color, they're editng for content.

Anton - editing community is evolving into something new. At the cusp of using digital for editing, but haven't used to it's full extent. We're starting to see the promise of new techniques to shrink the distance and time between people and task completion.

Telecine done in NY, Avid data goes over internet so they can edit about 3 hours later. Faster than FedEX!!! Do more faster.

How to properly work with high def and move away from down conversion tapes. Get it on hard drive and start editing with it. Be it with Avid or Final Cut or whatever.



Brian McKernan - Digital Cinema Magazine - for indie production (TALK TO THIS GUY, GET HIM MY CARD!!!).

Is a revolution for indie.

XPri can edit native IMX (Sony, right?)

Digital touches every aspect of production these days (duh), but indies are most impacted by these new available tools.

reducing the barriers to entry, mostly cost based

hopefully the talent will evidence themselves

At NAB this year the affordability of HD editing is the big thing - Adobe, Apple, Pinnacle, Sony, etc.

Codecs that get better and better for acquisition.

Gotta make sure solutions aren't developed in a vacuum. Has to make sense for the market.

things to consider:

HDV - a lot about this - how good is it? It's creating a new level of filmmaker. It's long GOP. Does anybody know how to edit in in a usable format? IF I shoot in one format, can I edit in another and output in another? Will it all play nicely with each other codecs? Other kinds of compressed HD? Will it work OK out of the box> How does a film lab print from an HD source tape? How to get to 35mm? Will it cost $100K to do it? Or spend that much at a facility to get them to try to fix it? What gotchas are down the road? Can I work at a low data rate and hand off to a full fledged facility? How do you interop with 3rd party companies? Whhich ones that you cooperate and compete with? Beware companies that don't know the whole workflow. Beware singular widget makers that don't think about beyond their one widget?

TALK TO NORM! HE'S ON IT.

One of the tings in the last several years is that the editors are being asked to handle more things. The ability to manipulate images within frames, create title sequences, to burn the DVD, etc. are all tools he likes, but the next guy might not want to.

There's editing and what's been known as assistant editing. In indies, that was one guy. It's all getting mixed up these days.

is there too much "job stuffing" going on? Sometimes yes, sometimes no...depends

new editors coming in from MGM or wherever: some editors are more tech savvy, some "need more training" systems become more complex

Trying to make a better end product, not necessarily save money. To get to that goal, to use tech across different budget ranges, everybody needs to become more adept and get used to where each piece of tech makes sense.

Tech may be similar across budget ranges, but the workflow can be quite, quite different to an amazing extent. R&D can go into the high end to define the tech, but workflow improvements come from the low/indie level, since they HAVE to get stuff done.

Everyone wants high def cutting and previews, the indies are finding the ways to get it done.

Anton developed an HD workflow for around the bend. It was a lot to get started. A lot of time you have set workflows that have worked for 15 years and they don't think to go beyond that.

They took off the shelf software and used it in a different way. Worked everything at 23.98 fps. When they came to do their HD preview screenings, the online went extremely smoothly. he did my "duh" way.

What resolution is good enough for editor's desktop, and what previewing size.

At MGM, wanting to use compressed hi def for editing, 25MBit. For integrating for timing, VFX etc. you need higher. compressed HD is plenty to work with for creative decisions. Managable from a storage standpoint, and pipe-able around the facility without making a tape. Putting small projectors in the editing suites. All that is way better than what they are currently doing.

Stream dailies around the studio, or somebody's home for review, etc. All based at the studio is a security advantage, hoping to find ways to watermark everything either visibly or invisibly.

As an editor, I don't care if it's sucky/chunky it's OK to make creative decisions. But to work with others the tech is being pushed. Multiple formats for different uses. For studio execs you use a higher quality for them fo rD-VHS or a hi-def server. Online dailies are compressed to QT or WMA to stream. Lots of requirements to find the proper medium for each. Even with hi-def screens - 1080? 24P? How to preview?

How much is studio fear of piracy use of latest available digital tools for editing or dailies? Dailies aren't a concern, just assembled sequences. Digital tech will offer higher security. It's the subcontractors that are their concern.
As they realize they "can get caught, and will get caught, and will go to jail, and loose everything they ever had."

Redefining how you put something on a server or work in a shared environment, learning about streaming technology, redefines what the editor needs to know.

Digital dailies over the Internet, in those apps can leave comments for others in the production chain

How long until dailies are DVDs with alternative audio with audio comments for production purposes.

Dailies were for CT and director to judge technical quality and performance, and also often became what you edited with. Now they serve different purposes and come in different forms.

==============================================================================

NEXT UP: Digital Restoration of Classic Films

Rob Hummel, Senior VP Production Technologies of Warner

Jan Yarbrough, TD, Warner video operations

Digital Restoration of Technical Three-Strip Motion Pictures

mining 60-65 years old to a level of quality not seen in decades

Technicolor 3 strip camera weighed 175 lbs...soundproof weighed 550 pounds

ran 3 separate strips of B&W film for R, G, & B. that is stayed in sync.

TEchnicolor was 4:4:2 by default since the red was packed in behind the blue

From RGB negatives, they made CMY tinted film, they printed those together.

To the best of their ability, they couldn't get color registration.

Classic film don't live up to the quality we expect. Classic titles didn't look good on DVD.

Elitism interfered...archivists think it's "bad" when you restore them

Nitrate negs warp physically over time as well, one per color channel

new software - re-registers stuff

32 blade processors can do 16 frames a minute

He's doing cool but totally believable stuff - scan for edge detection, realign and kick out color channel aligned

Showed some before/afters of some classics...the difference was amazing!

Gone With The Wind jus got finished 3 weeks ago.

Wizard of Oz is being worked on right now

FOR REMASTERING OF STUFF: TECHNICOLOR 3 PERF is way too strong/saturated. They pulled it BACK for the 2K by 1024 master we just watched. Going forward, in the future, they plan on scanning at 4K with their new Spirit scanner.

this is cool, and good, and I'm glad they're doing it, and the results are striking in their improvement...I just see it as "OK, nice, great...why didn't you do this before?"

I do like the blade server application, though...thinking at a proper scale.

================================================

NEXT UP:

DIGITAL MASTERING FOR FILM AND DIGITAL RELEASE

Debra - West Coast Editor, Film & Video Magazine

next pictures - the panel list and panelists

TALK TO DEBRA

Executive summary -

more people can fuck up, and un-fuckup, the movie with DI

Some key components in solution:
device independent refernce color space
accurate translations to/from real physical devices
observed standards for color mamagement and other metadata, they don't exist yet
WAY more bandwidth and capacity - Moore's Law, keep going!

NEED HEADROOM - allows for deferred decisions, film neg captures more than you can see, need this principle to be maintained in a digital process.

gotta have enough bits in res, bit cdepth, color range, reserve image area for framing decisions (overshoot)

digital must not slow us down! - gotta be faster than traditional solutions

"No longer the blue screen problem, it's the blue screen of death"

pinch points - film in and film out

jbetter calibration means you can run film records in parallel, but depends on very accurate alibration

basically anything is a pinch point in workflow

some things MUST be realtime:

with 4K data, can't be realtime on everything

Gotta be RT when human intervention/adjustment:
color correction
image positioning in compositing

others things just need ASAP:
one-light transfers
import/export

GOTTA BE RIGOUROUS about your analysis of what has to be RT and what DOESN'T

virutal DataCine approach:

rRT random access to all shots
RT in pipleline - image sizine and positions, color correctin, grain managemnt, etc.
PICTURE

in a SAN setup, be sure QOS (Quality of Service) is ASSURED by the vendors. Get it in writing before you buy.

Digital must be trustworthy WYSIWYG.

gotta know the ugly truth of doesn't work, model of what

Better color Mgt:

formailization of mltiple/differnt color spaces
spearation of "color correction" into domains:
scene
look
display environment

better use of metadata for session exchange along and across workflow

should be in a color space appropriate to the SCENE, not just appropriate to some device you're working with.

need fewer optical print generations, lower lens distoritions, better dust busting, etc.

want all deliverables to come from the one digital master

THE DI MUST

be completely ftrustwowrthy WYSIWYG

inpose no speed, quality, dollar or other compromises

provide bonuses and bennies to make it worthwhile

must become a vialble economic thing.

ROB HUMMEL ON 4K SCANNING FROM A BIZ PERSPECTIVE

Catwoman in production, were asked if they would scan everyting (600K feet of film) wanted to be able to auto assemble off the whole thing. Using same system as Lord of the Rings,

At 100TB they ran out of room after 380K feet of film.

ADIC based RAID solution, they had hard drive failures.

Next DI films are planned at 4K, they have 160TB

tried ATA for neartime storage, about 20TB for RT footage across network.

ATA is more efficient for smaller and cheaper.

doing stuff where even Thompson says "We didn't know THAT would happen..."

4K is startling better than 2K back to back...2K is 75% LESS data

BE CAREFUL..don't assume all will work out of the gate

they had same ADIC system as a gov't nuclear explosion testing simulation guys uise....and Technicolor has more data!!!

he wanted more tools, was under illusion that there was more time in the process than there really was

guy from EFILM: Wants ability to FILTER ... "be able to search & handle all the left side of a conversation, for instance" all again is a search & batch type stuff

They archive final digital intermediate...don't do and wish they could - be able to store the original scans...and even that doesn't do it...he needs a Quark "Collect for Output" type thing...that is the REAL way to archive, along with the final DM.

SHIT, MAN, THIS IS WHAT I'VE BEEN DOING ALL ALONG....
MIKE'S THOUGHTS:
There needs to be OTS (off the shelf) DI software...it's NOT an NLE...it's more...something else...certainly want to have tight integration and a handoff with

STEPHEN WOULD WANT: Stephen, a colorist from Technicolor - if you sit with your client and you color correct, everything has to have a LUT to look like film. he's working with stuff that doesn't quite perfectly nail it.

But the lab also has some variability....calibration with the lab is ALSO a crucial issue....again this is a pre-press like issue...just cuz your screen and matchprint match, doesn't mean the press operator will generate an exact match.

The LUT stuff doesn't sound like it's quite doing everything it is needed

LOU; WHAT YOU WANT IN YOUR SUITE: "Tools that work" he asks "why are we doing this" on some projects. On some projects it is because there is no other way. Sometimes it's because they think it would be a good idea, not because of a particular NEED. How are you improving your storytelling with this? It's just really not easy yet, you can kinda get done what you need to get done. "Color correction gets consumed in real time in a linear fashion, you need to be able to see & feel the flow, otherwise you can't really tell what's going on."
Be clear about what & wh you're doing as opposed to just doing it "cuz"

then I asked a question, feeling kinda stoopid....


======================================================

NEXT UP: DIGITAL ARCHIVING:

question to ask: for archiving purposes, I've noticed that hard drives are now cheaper than tape. How are we gonna back up the digital stuff in a digitally retrievable format?

Talking about maintaining the fullest capture of the source, and being able to access it in the future. OR, to be able to re-create from the files.

Role of archivist is to preserve and protect the remaining record of that content for future folks to access and distribute it.

preserve, protect, manage, and distribute that content

must be as close as possible to the source, and widely acceptable, so a widely available standard to archive in

Digital info can be useful only if standardized and universalized - international (think ISO, ANSI, etc.)

Inherent Danger of electronic Byte System:

more info stored digitally than on all other information systems combined

currently climate controlled film vaults is the way it's done - source film, InterPositive, etc.

Film archiving is transitioning to digital through a hybrid right now. Camera neg and release neg are the analog. Digital will be camera acquisition, digital intermediate, release DM.

Film archive advantage: 35mm is universal, stable when climate controlled, a learned process. Drawbacks are expensive (bulky to store, climate control), gen loss when duped, when pulled from storage, your first step is likely to be do digitize it.

Digital is easy to copy, exploit (work with it), move around, access, etc. They are expensive, cheaper in real estate and expensive in equipment. Major drawback is that media is not as durable as film, nor as likely to be stable. There is no universal standard. Digital Cinema archiving is going to be expensive. uncompressed, a 12 bit 4K 3 hour feature is 50 to 100 TB.

30 petabytes a year of Hollywood files. 15 petabytes with 2:1 compression. Scanned at 6K, gonna be even larger. Archived 16 bits per channel, bigger still. What to archive, and how? How to access? What is the metadata? He suggest MXF and UID are good starts for this.

Cinema industry doesn't need to solve the problem by itself. Three pillars of high end visualization: entertainment, military, industrial/medical.

Cinema is late to the game. USGS maintains Earth Observation Archive in South Dakota - they have film archives to 1937 - 80,000 rolls of film. All kinds of formats, rolls and sheets, B&W vs. infrared, etc.

Stored in underground vaults.

Good for at least 100 years, maybe longer.

Higher and higher % are digital. Digital imaging of earth goes back to 70s. they have 2 peetabytes, expect to double in 2 years. Only use lossless compression. Their mission is to hold it "FOREVER"

they assume media life isn't long enough. They assume hardware & software won't work with each other.

Data integrity - pick standard formats (largely set by imagers). Lots of error correction and checksumming etc. Stringent audit trails.

They do backups and and offsites on the data, metadata, and databases storing the metadata on how to find stuff

data migration - whatever they are on today, they'll have to move to new media every 5 or so years. new media is usually smaller, greater data density, etc. With automation, they can keep the labor costs down.

Layered storage architectures - online (fast disk), nearline (slow disk), offline (tape or optical), offsite (offsite tape & optical)

Sarbannes Oxley has to be kept for life of account plus 6 years for secirities firms.

HIPAA says medical data has to be kept for life of patient plus N years

Library of Congress is going digital

etc. etc. etc.

Others have dealt with this.

Disk sales in world - 1995 - less than 200 petabytes
in 2000 was 2000 petabytes of disks

storage sales are up, and a commensurate economy of scale.

tape densities have grown 1000 fold in 30 years, speed is only up 24 fold.

New media choices - not tape or disk. Bluray is coming, UDO is coming at 50GB per 5.25 " disk. Holographic optical - 200GB on a platter. 1.6 TB in a few years. Optical thin wave holographic storage - 1 inch square, 1mm thick, 1GB today, 10GB in 2006

tape backups - 200GB per cartridge - 1TB by 2008

standards are coming from the storage vendors - April 6 announced an interoperability - SMIS standard (certified interoperable)

Digital archiving is possible but a challenge

Just need to ride the trrends that military, science/medicine are already solving.

Library of Congress in text is 5TB

TAPE IS TOO SLOW, TOO EXPENSIVE in some market spaces

Truly lossless compression for archiving imagery from eTreppid Technologies

Lossless multipass digital data compression

true archival storage - mathematically lossless compression, not visually lossless compression.

why lossless compression:

The Need for Lossless Compression
retains all the original content
allows for multiple edit sessions ithout encode/decode generational loss
compresses all the "stuff" - video, audio, close captions, project files etc

It is only useful if there is significant savings in storage space!!!

(That's why disk driver level compression went away)

about 2:1 compression is the usual deal...which is marginally useful...they are about 8:1 to 10:1 for LOSSLESS compression. WinZip was about 17MB from the source 20MB.

Lempel, Ziv, Welch (LZW), Huffman was a biggie before them

4096x1714 is the StEM footage

For final archive, can get better than 9:1 compression. They got 22:1 if they let it run "a long time" to further analyze for smaller stuff. So it is VERY assymetrical. But what is the compresion/decompression time?

Took a couple of minutes to compress one frame on her laptop.

They are coming out of R&D, hope they to have standards bodies pick their tech for new standards

Grover Crisp VP, Asset Management and Film Preservation for Sony Pictures

"if you want to retrieve your film in 40 years, PUT IT ON FILM now...."

a reporter asked about film restoration -

the lack of proper preservation strategies - the key is proper handling and storage of source

controlled access for future use, protect against damage

film has to be cold and dry environment

studios have historically treated their film elements, many studios used originals as a printing negative, which can be very punishing. It's the reason many negs no longer exist. Ironically, the more successful the film was, therefore the more damaged it gets!

It was then stored improperly for decades, sometimes they get lost completely.

No longer print off of originals, most studios have built proper storage vaults that are not local (disaster proof).

ProTek from Kodak is the standard bearer of commmercial

Protection Elements - stuff to help recreate the original.

The holy grail is the cut OCN (Original Camera Negative).

35mm color separations on B&W stock and put'em in a vault.

Then make another answer print from that to make sure that the process worked right.

That testing is the most critical component - hate to THINK you made a correct copy when you just stashed a messed up one

All these elements get stored geographically separate.

Sony has 3 facilities in different parts of country. If any TWO of those got nuked, could still to a great extent recover every title. (IT'S LIKE A RAID 3 OR 5 FOR ANALOG!!!)

There are no real standard in place for DI. Many processes are unique to each facility, software and workflow are different, unique, and proprietary.

What do they get from production? Data files that theoretically include all the data to recreate. It can't be tested (DO THEY HAVE APPLICATIONS TO WORK WITH?). There is no way to test that it is archived and retrievable without replicating the DI process again.

Long term - file format obsolescence

they also store the final negative. A format that can easily be scanned again. They are trying to think 50 years ahead.

negative is scanned in at 2K. Most DI in last 5 years is at 2K. Bigger problems in the future.

Resultant DI negative is the most original content - again, duh, just like I've been doing - source material almost doesn't matter in terms of it's link to final - it's nice to have if I need to recreate or change content, but it is in no way indicative of what my final was.

blah blah blah...this guy is going way over....

ASC Tech Committee has a subcommittee on archiving


Josh Pines is working on using color seps for preserving color sep metadata in the DI process

Q1: cost of tape, when cost factored including the tape drive
Q2: for your digital archives, I'm sure the apps aren't included

Backup copies: one online, one offline, one backup if you're worried

Origianl material, 2 protection copies used at Sony

Q&A:

Migration is NOT preservation - Migration is carrying forward your best approximation of film -

Scanning is only sampling of a continuous image. 6K sampling to downsample to 4K, it's at least as good as the eye can tell.

Some followup afterwards:

Glenn Kelley from TI says they did look to some prepress models of color matching systems, and tried to just use cut down versions of that for their stuff since they wanted to define a reference projector.

Etreppid is talking to vendors about RT hardware for using their stuff not just for archiving compression, but also for a working codec. The interesting thing about it is that the longer you let it cook, the more it can compress it down. So the same alghorithm can be used for Real Time (RT) compression for a camera acquisition format (they're talking to manufacturers), for editing (talking to them too), and for full workflow from capture, through editing, into color correction, and ultimately for archiving....but the archived copy can "cook longer" and obtain much better compression...then it's just a time/file room tradeoff.

THE BIG NEWS: Final Cut Pro HD 

Apple has launched a ton of new products today (updates to Shake, Logic, DVD Studio Pro, a new motion graphics application called, well, Motion; updated PowerBook and iBooks, a new 15GB iPod, etc.)

But the big one as far as I'm concerned is the new Final Cut Pro HD.

Main features:

-Capture DVCPro HD over FireWire
-edit in that native format
-output over FireWire with no generational loss
-RT Extreme (realtime dissolves, color correction, etc.) to an Apple Cinema Display. (yeah, I gotta get me one-a those now)
-fullscreen playback on Apple Cinema Display

Mike's Comments:

This is a big move. While it's been possible to do HD over PCI cards, this is a big move up.

Apple and Panasonic's agreement is a big deal. This is a significant push forward for Apple in the low cost, high quality department.

More on this later..

Apple rolls out Motion, their own motion graphics package 

Apple now has their own motion graphics application, Motion. I'm surprised, almost shocked that Apple is doing this. Forget firing across the bow of Adobe, this is shooting at the bridge. They are claiming "realtime previews, procedural animation, and Final Cut Pro HD integration."

Other features:
-natural simulations - simulate gravity, wind, etc.
-particle system
-Photoshop integration with layer and blend mode support
-Bezier keyframe editor - but of course...
-bunch of templates included
-90 realtime filters
-Primatte RT - a nice keyer with realtime HD (yes, HD) chroma keying. This is huge.
-and of course, support for After Effects plugins

Mike's Commentary:This sounds BIG.

Clearly this is their integrated response to Adobe's tight integration of Premiere Pro, After Effects, Photoshop and Encore (their DVD authoring application).

This should be very interesting to compare to the Adobe lineup, which is Windows only. The only piece missing from Apple's version is Photoshop or a similar application. But since Photoshop has been a verb for about 10 years, I don't see them tackling Adobe there. (Nobody says "Just Premiere Pro that and it'll fix it right up." but "just Photoshop that" has been in my argot for a long, long time.)

I'll be checking this out at the show (NAB) this week.

Saturday, April 17, 2004

With the 2004 National Association of Broadcasters (NAB) trade show set to happen April 17-22, UK developer The Pixel Farm has announced six new tools aimed at digital video professionals, as well as an upgrades to the company's PFTrack application, which tracks computer-generated video output against the original footage. The Pixel Farm will display all their new and updated applications during NAB. [MacCentral]
Red Giant Software today announced Primatte Keyer for Avid AVX, which brings the power of the Primat... [MacNN | The Macintosh News Network]

Some Summary/Commentary on Day 1 of Digital Cinema Track at NAB 

OK, got a little time, here's what I think was coolest today:

DALSA Origin camera (link elsewhere in the blog): AMAZING image quality. Has something like the dynamic range of film (approximately), but WITHOUT apparent grain. I'm going to be curious to find out more about how it is compressing the images, how much, etc. If I recall correctly, it is 4:4:4 color. Interesting.

Christie digital projector was interesting, but way expensive (I'm assuming).

Most useful conclusion: Digital projection in theaters is going to be held up for awhile.

From my notes:

"Exhibitors are hosed by current model of digital vs. film projection - a $100M movie is still going to be a $100M movie (in sales, not prod costs!) however it is projected - what incentive is there to go digital for the theater owners? The manufacturers have an advantage in that they get to sell new gear. The studios have an advantage because THEY get to save money. The theater owner has to spend money to buy a digital projection device, which costs more, amortizes faster, risks technological obsolesence, requires more expensive and frequent maintenance from higher skilled (more expensive) employees, gets restricted by DRM sometimes (some movie key licenses lock to a particular projector, so they can't run it on another screen), forces them to share info they don't want to volunteer necessarily (such as how many times the movie was shown, (including after horus for staff? -mike), etc. And the theater owner, under current biz setup, doesn't get any more money. "

and also,

"MIKE'S COMMENT: Now I understand why D Cinema is experiencing a slow rollout. Equipment manufacturers, movie makers (Lucas), studios all want D Cinema because they'll make money on the deal. Until a convincing business case can be made as to why they should switch over, it would be financial suicide for exhibitors to switch over now with the current deal. Now, if they paid less for the digital film, or got to keep more of the money, perhaps that would be a way to make it work (Mike's thought there). There needs to be a financial incentive and to date there isn't one. people aren't flocking to see it projected digitally, at this point it would be for the novelty value, not for a quality value.

UNTIL D CINEMA IS BETTER THAN FILM PRESENTATION - UNTIL IT HAS HIGHER RES, BETTER CONTRAST, BETTER GAMUT, SOMETHING, there's no reason to want to go see something digitally shown. $5 ticket maybe? I don't see a reason. the image is more stable, but nobody is jumping up and down for that. 2K vs. 4K? Most audiences would be hard pressed to see the difference, and you'd need a Lawrence of Arabia or LotR: The Return of the King to make it visibly different to an audience. It needs a "Wow!" factor that is lacking. So until it gets WAY better quality, noticably better than film when projecting (I'm not talking acquisition AT ALL here), AND it is price viable for the theater owners, AND they feel safe they won't be obsoleted next Thursday and have to buy again, AND there is a profit/business model for them, it's not going to happen.

The way to make and project better than film: Shoot with DALSA, project with Quvis on Christie. And how much more would that cost than film at present? EXACTLY."

The discussion on color space was interesting. With traditional film projector, you have to worry at playback time about whether the bulb is bright enough and if it's in focus. With digital, there is a HUGE other list of things to worry about. Glenn, the guy from TI, as a part of a working group, has a lovely plan all worked out about how to handle these things. Whether the industry can be convinced to go to that level of complexity in their distro models will have to be seen. My gut level instinct says it's going to be tough to get the full spec as he's suggesting implemented. I'll need to look over my notes, and think about workflows, to do form my own opinion as to how viable the plan is...or at least how accomodatable from an indy's point of view. It could potentially become another barrier to entry for indies, which would be ironic since digital was supposed to make it easier for them to get in. Then again, it may just end up being a plug-in at the end of the chain working with metadata from your shoot. Ultimately, your camera should include that info with every frame it shoots.

Hey, there's a thought to ask the DALSA folks: if you're shooting digitally anyway, why NOT encode every bit of info you can about the shooting conditions? For each frame, what was the frame rate, the aperture setting, the focus setting, the rate of change in those at this frame, etc. etc. etc. Why not? You're already capturing huge honkin' ridiculous amounts of data, so why not a little more.

D-Cinema in Asia - interesting to see them way up the chain on this. 2.5 inch drive based data cartridges as distro physical format - very interesting. They are compressing to 1280x1024 anamorphic for some of their stuff, and 1920x1080 for some others. But the data rates are comfortably within the specs of those laptop drives (the 2.5" drives are the same as laptop drives). It is unclear whether they are fuly enclosed drives, or something like a Jaz cart. I'll need to follow up with them...

The Quvis media server - the whole point is to have an inexpensive distribution format, to have something cheaper than film to duplicate and ship. What about re-usable drives, as Asia is doing? Or hell, just mail a FireWire drive...

Digital Projection is getting good, but needs to be better than film to cost justify...or enable some business model that allows for a business case to make it worth doing. Otherwise it simply won't happen. Yield management could be one way to do it - easy to scale how many screens you are showing your feature on. 10 on opening night, 2 by week 3....but bump it back up on the weekends. Whatever.

And finally, ENCRYPTION. All through this stuff today, and especially when they were talking about the packaging of the data, heavy DRM throughout the chain. Hollywood really, really, REALLY wants content distribution control.

As it was so eloquently put by Michael from NATO ("the one without the guns" as he put it):

Security is the art of abstracting trust - take it out of the hands of those you don't trust and put it in the hands of those you do.

So that says a lot about the studios and how they feel about exhibitors.

OK, outta time. More later.

With a pounding headache, this is Mike in Vegas, signing off...

NAB Digital Cinema Track: Raw Notes on D Cinema in Asia 

NAB Digital Cinema Track: Raw Notes on D Cinema in Asia

D Cinema in Asia:

Dr. Man Nang Chong

using a pure software approach for their servers, using Motion JPEG, MPEG-2, JPEG 2K, etc.

200 theaters worldwide are running on DSR Servers

5 in US, 14 in China, 22 in Singapore

500 digital prints prepared

small hard drives as data cartridges

Singapore D-Cinema Multiplex

Nov 2003 announced to convert 20 screens

fiber and/or satellite connectivity to be installed for digital reception

2TB per server, 20-40 titles stored on each

expect to finish by 3rde quarter this year

not store and forward, real time streaming jukebox. content loaded through central server, can stream to all the screens, different content, at the same time, remote control from a central office. Can adjust the number of screens you are showing a movie on to do yield management.

Digital screeings of 3 movies and 1 Hong Kong movie since Nov 03 (Master & Commander, Brother Bear, etc.)

Singapore is a digital hub, good models of distribution of content for cable and satelite.

They want to position themselves as a value add center (encoding, subtitles, distribution hub, etc.) for Asia at large

2.5" hard disk based cartridges. Can use USB keys for subtitles, could add security keys for USB. Laptop drives are basically what they are using.

Film, digitzed & color corrected to a 2K color correction system, put on D-5, compressed & encoded using MPEG-2 AES encryption 128-bit encryption to DLP or other projectors

They do some live events as well.

DVD-IP non-realtime delivery. Final Fantasy was delivered to Shanghai as a test. Delivery of coentent via satellite as tests.

cable delivers via ATM was tried from Hong Kong to Singapore on a SingTel. Sent in less than 6 hours, 15 (50?) megabit per second. Couldn't tell which he said.

In India they partnered with Essel Shyam Comm LTD. secure vpn, encrypted content delivery.

India has been building a very efficient digital print production

HD Telecine Thompson, Da Vinci 2K color correctors

in last 10 months, 34 feature films released in India, a world record

while the negative is being used for optical dupes, it was being scanned frame by frame for digital distribution

Parallel distribution - some 300 to 400 Analong 35mm prints for A Clas Centers

Digital satellite feed enables an X number of digital prints to the existing B & C class centers, the # of digital cinemas can scale up easily

Piracy markey is same size as legit theater market. The organized market value almost equals unorganized film piracy market value

Pirate video parlors in the suburbs.

Big revenue loss to gov't as well from taxes

cost of 35mm print: $1300US

transport cost: $220

satelite delivery is good because:

no physical media to steal

reliable and secured delivery, no secrity guard is needed

centralized control for mngmt of content delivery

etc.

www.gdc-tech (google it!)

private screens in China: 60,000 plus

but commercial, US style, running movie movies, only a few thousand

What they are actually presenting: 2 options they use: in their workflow

In India right now, they can feed dual link HD-SDI 1920x1080 with their Da Vinci color correctors, resize it to 1280x1024 with a baked in anamorphic stretch

bandwidth for 1.3K res is 65Mbit/sec

for 1920 is about 100 Mbit/sec

VBR for both of those (variable bit rate)

NAB Digital Cinema Track: Raw Notes on Fiber Distribution of Live D Cinema Events  

NAB Digital Cinema Track: Raw Notes on Fiber Distribution of Live D Cinema Events

Fiber distribution of d cinema
This makes more sense to me - except for the lack of

alternative content is one way to help pay for the transition to D Cinema

20% occupancy overall in theaters

big problem with current alternative content is that it is usually DVD bandwidth, not "visually lossless" as he calls it

CINEvents delivers live events to movie theaters

live over fiber optic to theaters in Dallas, Detroit and San Diego

underlying network technologywas ATM over a MPLS backbone, which is currently state of the art but they are looking for what's next & better

QOS guarantees were noegotiated with the carriers for total dedicated circuits, not 99.999%, but 100%

They did band interviews etc. with Lone Star (the band), they tried to involve those involved and NOT watching a TV show. Gotta make the remote audience feel like they are really there.

Was It A Success?

Failure is not an option in ShowBiz

had major technical hurdles, 80Mbit video, full Dolby AC-3 sound
BIG PROBLEMS:

don't want ot dsend a live event while Bank of America is making a funds transfer

LOCAL LOOP TECH IS STILL LIMITED BY TELCOS - tarriffs, etc.

Network management and control belongs in the hands of the customer
DON'T DO THIS OVER IP, since there is no guarantee of service

feels LAME to have the interview with the guys trying to address the remote audience...feels cheap and gimmicky - audiences can smell that. But if they did something during the LIVE show to address the remotes, that would be more viable I would think...

They did it in HD

will this work in general? I don't think so - it's missing the emotional appeal of a lie show. There isn't that focus on interest, and it can be replicated by rebroadcasting...you'd have to be really into whatever the content was in order to make it fly...

If it ever does catch on, I think it would kinda suck..it's sorta like applauding a character in a movie...they don't here you...they say something about a town and I'm wondering wheter that is the local audience of the remote one. The local audience stands up and cheers, if the remote one had, wouldn't that be lame? Since they can't see them? How different would teh epxperience be if there were a camera in teh remotre audience feeding a signal back to a monitor the band could see?


A-HA! DURING Q&A, he said they fed the audio from the remote theaters back to the live event location...good...

He talks about selling hats, t-shirts, beer, wine, etc. at the theaters...viable? Lame? What do you think?

Technically I think it is an interesting idea...

80Mbit is the technical limit...4:2:2

11% of DVDs in Blockbuster are music

How do you advertise and market for this kind of one-time event? Trying to get a regular schedule of events...let band websites know. They filled up 30 theaters for Prince.

What about other events in superb HD? Like SuperBowl (at Alamo Drafthouse in HD with an awesome projector...YEAH!), or Indy 500, or whatever other sporting events. Again, it has to have a value add...

There are about 1.1 billion tickets a year to be sold. (stat from a speaker) Profits have gone up because of rising prices of tickets, not due to more people seeing movies or those same people seeing more movies.

D Cinema makes a good match for Alamo type venues.

NAB Digital Cinema Track: Raw Notes on D Cinema in Europe 

NAB Digital Cinema Track: Raw Notes on D Cinema in Europe

D Cinema in Europe: there's a book coming out from them on pros and cons of digital moviemaking focusing on HD (duh). But watch for it, could be useful.

My god, this guy is putting me to sleep (watch me need a job from this guy later and get Career Death By Google). So I come out to the hall and plug in and start typing this up.

Projectionists come in and see the test bed for the UK digital cinema stuff, it looks like a TV station or networking/computing rack...because that's what it is.

25 million pound project, 250 screens ($40M us) in 150 cinemas nationwide

all types of cinema

largest d cinema push in the western world

www.edcf.net

he's doing great stuff, but man he's dry.

NAB Digital Cinema Track: Raw Notes on Michael Karagosian talking about NATO, National Association of Theater Owners 

NAB Digital Cinema Track: Raw Notes on Michael Karagosian talking about NATO, National Association of Theater Owners

Speaker from NATO, the National Association of Theater Owners

he put up a slide titled Ways To Fail with 3 main points:

Proprietary end to end systems

Private interoperability instead of true standards

Not understanding the market

There has been discussion between big port vs. little port. I wasn't sure whether he meant big screen (cinema) vs. little screen (computer/TV), or image capturing methodology, big port (35mm sized capture chips) or small port (CCDs are their present size)

Business requirements come first, and should drive standards, not vice versa

Exhibitors are hosed by current model of digital vs. film projection - a $100M movie is still going to be a $100M movie (in sales, not prod costs!) however it is projected - what incentive is there to go digital for the theater owners? The manufacturers have an advantage in that they get to sell new gear. The studios have an advantage because THEY get to save money. The theater owner has to spend money to buy a digital projection device, which costs more, amortizes faster, risks technological obsolesence, requires more expensive and frequent maintenance from higher skilled (more expensive) employees, gets restricted by DRM sometimes (some movie key licenses lock to a particular projector, so they can't run it on another screen), forces them to share info they don't want to volunteer necessarily (such as how many times the movie was shown, (including after horus for staff? -mike), etc. And the theater owner, under current biz setup, doesn't get any more money.

Security is the art of abstracting trust - take it out of the hands of those you don't trust and put it in the hands of those you do.

Security can mess up the biz model - locking movies to projectors, etc.

DRM - not conducive to practicalities of the business as it stands. Exhibitors have to give up data they don't necessarily want to give up - like screenings info, etc.

MIKE'S COMMENT: Now I understand why D Cinema is experiencing a slow rollout. Equipment manufacturers, movie makers (Lucas), studios all want D Cinema because they'll make money on the deal. Until a convincing business case can be made as to why they should switch over, it would be financial suicide for exhibitors to switch over now with the current deal. Now, if they paid less for the digital film, or got to keep more of the money, perhaps that would be a way to make it work (Mike's thought there). There needs to be a financial incentive and to date there isn't one. people aren't flocking to see it projected digitally, at this point it would be for the novelty value, not for a quality value.

UNTIL D CINEMA IS BETTER THAN FILM PRESENTATION - UNTIL IT HAS HIGHER RES, BETTER CONTRAST, BETTER GAMUT, SOMETHING, there's no reason to want to go see something digitally shown. $5 ticket maybe? I don't see a reason. the image is more stable, but nobody is jumping up and down for that. 2K vs. 4K? Most audiences would be hard pressed to see the difference, and you'd need a Lawrence of Arabia or LotR: The Return of the King to make it visibly different to an audience. It needs a "Wow!" factor that is lacking. So until it gets WAY better quality, noticably better than film when projecting (I'm not talking acquisition AT ALL here), AND it is price viable for the theater owners, AND they feel safe they won't be obsoleted next Thursday and have to buy again, AND there is a profit/business model for them, it's not going to happen.

The way to make and project better than film: Shoot with DALSA, project with Quvis on Christie. And how much more would that cost than film at present? EXACTLY.



The general tone of his presentation was defensive, frustrated, almost hostile (not in his voice, but in the content of his presentation). Certainly ticked off.

NAB Digital Cinema Track, Raw Notes from Christie Digital on their 2K projector. 

NAB Digital Cinema Track, Raw Notes from Christie Digital on their 2K projector.

DIGITAL CINEMA SYSTEMS DIPLAYS

Brian Claypool from Christie Digital

They make this huge projector I took pictures of

they make the 2K CP2000 projector - what does it cost?

TI 2K DLP chip

Digital Micro Mirror Device 2048x1080 2.39 aspect ratio

DLP can do almost all the same colors available in film

40% greater color gamut than HDTV
16 bit linear prodessing (15 linear bits to screen per RGB) 45 trillion colors

CineBlack - ighest available dynmaic range for a 3chip DLP

specific DMD level enhancements "black chip" helps the contrast

stable dynamic range with DLP over time

Cinecanvas - SEE THE SITE, READ THE PAMPHLETS FOR THIS STUFF

It can handle keystone cropping with anti-aliased edges, etc.

immediate suport for subtitles and stuff

CineLink - provides a secure encrypted connection between the projector and server encrypts the signal

40-75 foot wide screens for the CP2000 product line 4.5 kilowatt or 6 kilowatt lamp

they make another model for smaller screens using 2-3 kW bulbs

has an 8RU empty rack for whatever you want to do

various keyable access

matches film - 14foot Lambert at up to 75 feet away

2000:1 contrast ratio

ANSI Contrast: +500:1
Brightness uniformity: greater than 90% when showing full white

18000 Lumens

in lab up to 2800:1, but 2000:1 in normal theater environments

effectively 15 bits per channel per pixel range

supports 1920x1080@24p or 23.98

uses DVI input (ha!)

essentially as bright (or better) 35mm film projection setups

DON'T FORGET WHAT I LEARNED FROM SXSW GUY THAT MADE THE 50S GAY CHARACTER FILM - calibration is a pain - the tech would calibrate for the first show, but then wouldn't calibrate for the other shows...so everybody had to look the way the first film shown did. Not an issue in normal movie showing, but it is a big problem for film festival showings...when you're trying to sell the film...

LOW BATTERY - GOING TO ANALOG (PEN AND PAPER)...


-30-

here's a transcription of what I wrote down pen and paper:

Christie Projectors:

It is HUGE (see pics)

note to self - bring damn camera connector cable next time!!!!

can be remotely administered via Internet (extra cost)

somebody asked about cost in Q&A, the pres of the company deferred and said come to the booth. The crowd laughed and went "ohhhh..."

Must be really, REALLY expensive then.

the touchscreen control is a WinCE device

they have 60 locations with 68 screens in the US

10 in Europe

131 locations worldwide, 141 screens

They put up a list of post production facilities that have them, about 15-20 locations, and some were in bold, those were the ones that had the newer CP2000 units, not the M15 first generation product.

ILM has CP2000 installed. Two of them. Damn.

NAB Digital Cinema Track: raw notes on the Dalsa Presentation 

DALSA PRESENTATION:

They do the digital cinema camera

they wanted to test with actual, real, directors of photography of digital vs. film, so they made a custom side by side rig that shot simultaneously with BOTH - so they BOTH had to shoot the EXACT same lighting conditions. This is the most viable test I've heard of so far.

demo shot with alpha version camera, they have a beta camera on the floor

it's a rotating shutter camera like a film camera


John Coghill from Dalsa

they set up 5212 100T Kodak in a film camera, there's a side-by-side rig shooting film and Dalsa

take all the source, both, output to both digital an film is the ultimate goal

large aperature CCD with a image processing computer on the back of it.

These are my notes, typed in the dark, while they ran their demo

WOW - great coor! and no grain...image compression is good - WE'RE SEEING RAW FOOTAGE AND HASN'T BEEN POSTED

DEPTH OF FIELD FOCUS IS LIKE A FILM CAMERA

4K downresed to 2K is what we are looking for

HOLY SHIT it is clean and detailed

black jackets in foreground, sunlight background 13 STOPS OF DYNMAMIC RANGE

DAMN, THIS IS THE )_SHIT)_))

it's like film in dynamic range, but not..SUPER CLEAN no grain!!!

some strobing due to HMI light timing issues

guys in black waking on snow -

HMI strobing is a bad thing -

poorly lit /low light flesh tones - really well done!

OPTICALLY COMPATIBLE WITH NORMAL CINE-LENSES

THESE GUYS ARE DOING IT RIGHT - SKIP THE VIDEO, GO STRAIGHT TO DIGITAL IMAGE ACQUISITION

They are running off of a BOXX system to show their footage - is it 4:4:4?

flesh tones are good

depth of Field is really consistently good and film like

Effective ASA of the camera: in July they will have the next generation acquisition chip

Tungsten 64 and Tungsten 100 is the effective ASA

daylight rating is 200

STORAGE as you shoot - come to the booth on Monday

NAB Digital Cinema Track: Raw notes on Quvis media server 

They were playing back some demos from the Quvis media server. I have pictures I'll post later.

QUVIS SERVER:

23-28 megabytes per second, spatial only (no frame to frame intercompression), realtime 4K to 2K downsampling (since there is no 4K standard to project as yet)

30:1 compression

their codec gets MORE efficient with large file sizes

18.17GB for 12 or so minutes compressed, 680GB or so uncompressed

Digtal cinema goals:

distribution cost savings and flexibility
improved consistency of and quality of presentation over film
additional revenue from certain events alternative content (prize fights, concerts, sporting events, etc.)

improved security, transactin processing and DRM

enhanced and unique theatricla experience

production costs and time saver

content encapsulation for use and archive-reels don't get all scattered all over the place - film is very stable, but since it is bulky, has to be mechanically organized

technology headroom for long amortization - exhibition industry is used to 20 years before you have to oil them....can last damn near forever. 35mm sprocket spacing was defined in 1909.

Electronic is rarely long term amortization, so have to design realistic amortization into the spec...

as well it has to be at least as good and at least have the potential to be better

how many audio channels, how many controls, etc. is in the future.

are we trying to enable electronic theatrical distribution? Or is it something special, unique, new & different? Is this something that will be available to the consumer in some way?

Now with HDTV, we finally get to the point where we'll have something BETTER than theatrical standards of the day. Will home theater compete, or will theatrial projection be a viable social gathering and event?

ENABLE ELECTRONIC THEATRICAL DISTRIBUTION
-only provide something that is "good enough" or "nearly as good as" people are used to today has been the metric

-adds an authoring process to production that requires QA. As a process it is presently slow.

-if it isn't better than the current standard, it is subject to early technological obsolescence if it isn't better and gives some headroom

IF WE AIMI TO TRANSCEND FILM:

-enable the delivery of something that exceeds current capabilities. Need better resolution, dynamic range and signal to noise ratio...and needs to do it for a better price

-initially provide an electronic equivalent to the print process but transform the entire process over time

-long effective life- give the tech a lifespan to stay viable and make sure exhibitor can amortize the investment so he'll buy in on the next round

-enhance the creative process with integrted real time tools and a richer pallette

ELECTRONIC ACQUISITION will fuel that process he thinks

-Film infrastructure is pretty stable - 120 years old, 35mm is the standard for image quality vs. cost, and camera/lense weight

film tech is mature and improving slowly
enhanced film performance is EXPENSIVE

bigger film to get better image quality make it 4 times bigger, but that leads to a 10 times heavier lense.

IMAX cameras way a TON for this reason

ELECTRONIC CAMERAS

-potential exists for improvement in quantum efficiency (photons to signal) CAN ALLOW MAJOR IMPROVEMENTS IN RESOLUTION AND SNR WITHOUT REQUIRING LARGER AND HEAVIER EQUIPMENT.

electronic cameras have a bad taste for people right now due to limited dynamic range of current tech

immediate access to the image - film gotta wait, video can flat or black field it, correct for all kinds of things, use multiple shutter exposures during exposire, etc. HUGE POTENTIAL that is presently not fully tapped. Film has some limits that can't be improved as much.

Theatrical electronic production and ost

combine the good characteristsics of film and video
Integrated tools that enable:
cost savings
time savings
enhanced creative contrrol

an efficent digital master that can be easily and cost effectively stored and transported

ideally it can be delivered transparently to multiple display technologies (tv, projection, broadcast SD, broadcast HD, etc.)

HEY! THE DEMO TODAY WAS DONE BY SATTELITE LINK!!!

DM to a wide range of distro formats quickly will be GOOD thing to have

SOME REQUIREMENTS

If Digital Cinema is to achieve the goal of transecending film, the enabling tech and systems must:


Symmetric real time encoding and decoding
to assure a unique theatrical experience in terms of resolution, dynamic range and signal fidelity shoudl be deterined by limitations of physics, not film

efficent quality, not rate priority encoding for acquisition, transport and archive - playback can lose data, acquisition should not

-fast and cost effective production process similar to video today 0- 4K easy as DV

integrated security, DRM, transaction management...but DON'T HAVE A MIDDLEMAN or gatekeeper. Helllllo, MicroSoft! So how to structure the business to have a franchisable, distributable DRM and security that doesn't have a gate keeper charging you money

technological headroom to ensure a long life to the tech that will amortize in a reasonable timeframe

where are we going to proceed soonest? More res isn't necessary, greater dynamic range. 8K is the limit, 4K for the workhorse res.

4K cameras are possible. betwen 2K and 4K will be the standard in teh next few years according to the Quvis president

4K from a recording reason is a "good" number, 2K is popular because of the closeness to HD

digital cameras have great capabilities that are as yet untapped...he's talking about the long term future, not the current state of the art

NAB Digital Cinema Track: Raw Unedited notes on Digital Testing 

DIGITAL TESTING:

The ASC/DCI StEM Test Scheme anbd Production Requirements

projection tests etc.

Walt Ordway, DCI
Allen Daviau, ASC
Daryn Okada, ASC
Howard Lukk, DCI

realized that needed a real film to test! Actual film type footage, realized they needed it to be avaialble to anybody in the industry to play with it - HEY I WANT TO GET A COPY OF THIS STUFF!!!! Process used to create and post this stuff

For years wanted to see same images on film and HD

came up with images that would challenge the system technically, but also that would engender emotion

shot it anamorphic Super 35mm, and selected scenes in 65mm

what do we compare it do? Digital Projection as good as film from a lab? Better than that?

compare it as to how it would compare to the original negative - that was the intent to compare it to the original creative intent of the director of photography, etc.

MASSIVE CHRSTIE PROJECTOR

www.dcimovies.com has info on how to contact etc.

shot it over 3 days on the Universal Studios lot

negative taken to PhotoChem and Technicolor and did some film dailies off of that

had a general idea - telecine to HD with keycode
downresed to Digibeta and input into an FCP setup

two hours of totla moaterial
make 2 test things

a display reel an answer print from the cut negative vs. an answer copy (is that right)

other is a mini-movie, selected from a creative side , it's a mix and match

that madea scan list

Pacific Title scanned the 35mm stuff at 6K with 14 bit linear Northlight

4K 10 bit for 65mm stuff

35mm dowrezzed to 4K in square pixels, 16-bit tiff files

taken over to Pacific Theater on a Lustre at the Pacific Theater through the Christie Proejctor using dual link 4:4:4 on a large screen

Therer was a color timed 35mm answer print that they compared the digital stuff against this to match it

Lustre did the conforms, 8 effects shots, down at 4K, conformed it all together, came up with the final Mini Movie and Display Reel

Created some music since no dialog was created

It was intentionally made to stress the system.

Doing high-quality 2K and HD versions of this, they are playing with various downsample methods to see what produces the best results. COMMENT: I CAN'T BELIEVE THEY ARE STRUGGLING WITH DOWNSAMPLE METHODOLOGIES! IT SEEMS EASY ENOUGH TO ME...

the D.P. on Eastman negative 5218 - well designed to interace with digital "Van Helsing was shot on 5218"

we shot on film, a feature film concept, not to look for a test

5245 was used for the shot tilting down from the sky

Howard (tech guy) says: Cubis compressed it and delivered to Microspace at 2K..this footage has been downsampled from 4K to 2K and THEN UPSAMPLED TO 4K AGAIN!!! Why?



StEM - Standard Evaluation Material

confetti - don't see artifcacts

looks good and sharp, could be a tiny bit sharper

shot on film and transferred to digital for test purposes

rich vibrant colors!

no reason for this not to look great - it's film scanned and digitally projected

I'm seeing some tiny little motion tearing or sometihg during fast transients - need to see an action scene to be able to really tell

what is the frame rate of teh projectior? 24p I was assume

dWHAT DAMN COMPRESSION WAS USED!!! THEY AREN'T TELLING US! WHAT WAS THE COMPRESSION AMOUNT AND DATA RATE AND CODC?

yes, I'm definitely seeing some motion wierdness in this as compared to film.

small text is good but not great in


Cool night test sequence:

shadow detail in night scene is god but not awesone a but not awesome

but hey it's film!



text white on black is good

whites/highlight detail is CRISP and looks good



red street light is RED!!!

titles -

title crawl - slight vertial blurryness - why? Did they render with or without motion blur for that? Probably not with any motion blur



in titles a few tiny shakes and flickers, almost like dropous - or is that just an artifact of pixel placement, like subpixel placement issue aka After Effects



DCI IS DIGITAL CINEMA INITIATIVES, Inc.

asked the question about compression -


The final in post was 680 or so gigabytes of data, made an 18 GB compressed version

HEY! THIS WAS A DI - DIGITAL INTERMEDIATE - PROCESS, but it is delivered digitally, not back to film

TALK TO THE CUBIS (QUBIS?) guys about commpression

the question was does digital cinema take anything away from the storytelling?

OH, SHIT! THIS DIGITAL CINEMA TRACK IS ENTIRELY ABOUT DIGITAL CINEMA FROM A PROJECTION STANDPOINT, NOT NOT NOT FROM A DIGITAL FILMMAKING THING

So this StEM footage is just test footage to torture compression stuff.

The digital images were put together in Shake and Lustre. Effects were in Shake at Pacific Title.

worked with 16 bit TIFFs, reduced to 12 bit because that's the spec, we saw 10 bit dual link 1080 HD

going through DI - when going to film out - he's noticed if you want extra saturation etc., can be done in DI, and define the look of your film that way. DI opens up a greater range of control and choices than has ever been available before.

blue shirt, grey suit, glasses, short

60-80 Mbits per seond for the HD stuff said Qubis QUVIS is the company

confettiworked well, African guy looked good

low saturation black and white (actually slightlycolor) is a good compression test

torches are hard to compress, get blocky easily

14foot Lamberts for the peak color correction, they found that was the limit for film with the deep ends(whatever that means)

fog transitions makes it harder


the rain sho

the rain shot had high contrast

NAB Digital Cinema Track - Raw Notes on Packaging Tutorial 

SCIENCE AND PERCEPTION TUTORIALS - PACAKAGING TUTORIAL (WHAT'S THIS?) KEVIN WINES, WINESCO, Secretary, DC 28.20

how to put audio and video together and get it to the cinemas.

Digital Cinema Packaging Primer

what is DC 28? SMPTE committee organized to create standards for digital cinema

28.10 is mastering - image, audio, sub picture (subtitles), compression

28.20 is distribution (packaging, encription,

28.30 is exhibition (interface, and all of the in theater interfaces, performance)

28.20's goal is to create a series of specs and docs designed to handle Packaging requirements specific to Digital Cinema

Doc specs:
operational constraints
packing list
composition playlist
track files (the movie itself)
and some other stuff

Encryption is NOT part of this discussion

how to divide the media up? How to segment the physical media? 150-300 Gigabytes (he's worried about it, why not ship a FW drive?)

REQUIREEMENTS

Provide a standardized, common method for exchanging Digital Cinema Essence and associated fiel and dat

essence type aware, but essence agnostic

where possible, build from exisitgn work and/or standards

Philosophy:

The D-Cinema Packing is afile interchange form/spec 0-it is NOT necessrilya storage format, so you could copy the files off to your hard drive, not DVD, or hard drive, or floppy, or SyQuest - be agnostic

Segmentation for physical media - what's small, cheap, dense data that writes reasonably fast for distribution?

Digtal Cinema Distribtion package contents

it will always have one and only one Packing List and at least one of the following:

zero or more Sound or Picture track files
zero or more subtitle track files
zero or more compositoin Playlists
zero or other files such as fonts, rendered images, etc.

Operational contraints - for constraints needed across spec boundaries (????)

must span 2 or more spec docs
such as frame rate & audio sample rate
A valid DCP has one and only 1 Packing List

specific constraints only apply to THAT package

Packing List spec - list of the files and info about those files includede in a DCP (Digital Cinema Package) using XML

XML can define it's own tags

Packing List is a list of all assets included in a given Digital Cinema Distribution Package
inclues more info: is it encrypted? how big are the files? a hash of each file (checksum), MIME type of the asset (.tif, .mov, .jpg, etc.)

packing list is used to verify all files were received and are intact. Make sense if it's a 7 disc set.

in general it wants to give lots of info, supports public and private key encryption, let's you track it back to EXACT source, which will help identify possible copyright violation, can also tell if the files were altered. But there doesn't seem to be ANY link between this and the playback stuff...so altering the contents doesn't mess up stuff

COMPOSITION PLAY LIST

defintes in XML how to play back the track files

specs the manner in which each MXF track fiels are rendered within the comp playlist there is an ordered spequence of list of track files (reels)

comp list is a dself contained repreentation of a complete WORK, ad vs. movie etc.

integrity is Proected via certificate and digital signature - is this file, and the file it references, intact? hacker protection

see photo for example

SECURITY-SEE PHOTO

building this so that there is a lot of connectivity between packing list and composition playlist - do they refer to each other?

Use by Exhibitor to "play" a version of the movie

theere is a separate Compositoin playlist for each language dub - a FIGS release would need 4 composition playlists (French, Italian, German, Spanish)

There is a separate comp playlist for each SubTitle release and for all combos of the above

The assets remain the same, you just have a different playlist to play those assets in a different order - like a DVD authoring structure type thing

Reel Structure:

ID, annotationtext, AssetList, mainMarkers, MainPictgure, MainSound

filling in from earlier - comp playlist can be for JUST an ad, a trailer, a movie - they are SEPARATE files very much on purpose, thus you can mix and match, you'd queue up a series of comp playlists for ad, ad, ad, trailer,trailer,trailer,trailer, movie

everything is defined down to the granular level - SMART!!!

MainMarkers - used in theater automation...lights up/down, etc.

MainPicture-framerate, aspect ratio, size (in pixels? dunno), anamorphic squeeze amount or ratio, etc.

Sound and Picture track files (I'll take pictures here)

MXF is Material eXchange Format - standards created in SMPTE W25

picture 137-3766
TRANSCRIBE!!!

next couple of pictures are screens as well

KLV Encoding is
K=Key: a unique identifier
L=Length: how long is the field
V=value: what is the value of the field - this is the content - the picture track, the audio track, etc.

TOOK A BUNCH OF PICTURES FOR THE STUFF I HAVEN'T JUST DESCRIBED

NOW ON TO SUBTITLING:

2 possible ways to do it - PNG files vs. the TI approach which is a text file and a font file (better!)

should it be wrapped into XML? OR LEAVE AS STANDALONE?

work in progress - determine preferred method of packaging

HOW TO BREAK THE DOCUMENT INTO MANAGEABLE PIECES?

The goal is to bring down COST!

DVD-ROM? Hard drives not as big as the whole thing? What?

Digital inema Certificate: (pics)

encryption as-dcp

one key per track file - didn't want to go nuts
first part of each frame may be unencrypted`

NAB Digital Cinema track: Raw Notes on Science and Perception Tutorials - Color Encoding 

THIS IS FROM DAY ONE OF THE DIGITAL CINEMA SESSIONS-THESE ARE TOTALLY RAW NOTES, NO FILTERING WHATSOEVER

Missed the black & white panel - talked about defining black and white points in digital systems in digital projection & image acquisition

color pallette for digital - gotta be careful when defining pallette when talking about decisions made today don't encumber the future image acquisition AND projection systems

Glenn Kennell

Science and Perception Tutorials - Color Encoding DC 28.10 AHG Color

"gotta be careful we don't restrict the future"

report from color group within SMPTE

Criteria Digital Cinema Distribution Master

CHOOSEING A COLOR SPACE FOR THE DCDM

ONE MASTER SHALL PLAY ON ALL PROJECTORS and look the same on all projectors (device independent)

open standard w/no licensing fees

Extensible to wider color gamut and wider contrast in future projectors

future masters shall be backward compatible

has to work with today's tech

today's stuff has about a 2000:1 contrast ratio

future stuff may have more than 3 primaries and may use laser projection

DCDM in the chain

want to include Reference projector gamut metadata attached to the DCDM(XYZ)

COLOR SPACE OPTIONS CONSIDERED:

RGB - using extended damut DLP primaries with Xenon light source

WIDE GAMUT RGB - USEING CIE 1931 XYZ COLOR PRIMARIES - this includes the FULL color space, this is the classic

PARAMETRIC RGB - RGB with metadata to define color primaries ro Refernc eProjector

NOTE Gamut mapping requirements are indempent of this choice

XYZ made the most sense (XYZ PRIME just means it's gamma corrected) for a bunch of reasons

XYZ FUNDAMENTALS

it's 3 channel like RGB

to define XYZ, use 3 sensors, 3 spectral characteriestics (CMFs), X, Y, and Z are outputs

XYZ is a mathematical description of the color

x, y, Y is how it's notated

there's a simple match transcription to convert

XYZ is a superset of what we can see and what we can produce- so it's a safe superset of whatever we might capture, project, or just plain see with our eyeballs.

neutral white may change as tastes and tech change
XYZ CONCEPTS:

primaries of X, Y, and Z are NOT humanly visible in their extreme

easy math to convert to RGB or other color spaces, and it supports ANY white point

How to make color in XYZ? color in display is created by mixing the RGB values from math transforms




COLOR SPACE VISUALIZATION

NOTE: WOW, THIS SHIT IS DEEP GEEK!!! I HOPE IT GETS MORE PRACTICAL, AND YOU, EARTH BASED TECHNOLOGY. LOTS OF DISCUSSION ABOUT COLOR PRIMARIES THAT AREN'T VISIBLE TO, YOU KNOW, ANYTHING BUT BES AND SHIT.

dXenon DLP can produce more colors than the TV gamut

whatwe can see is quite a bit larger than what XeXenon can do

today's tech can't produce what we can see by a WIDE margin


and XYZ can do an even bigger range of stuff. so XYZ is a good chi choice, because it can reproduce stf stuff well beyone beyond what we can see (it's dark, can't see what I'm typing)




REFERENCE PROJECTOR

projector used in the mastering environment, when you're doing your color critical work

today's are DLP Cinema with a color gamut expanded beyond CCIR 709 to more acurrately represent cinema gamut

Refernce Projector Peak white - 12 foot Lamberts 42 cd/m 2 (squared)

we verified that film produces 14 ftL for peak whites (with 16 ftL open gate (raw no film)

we recommend a peka white level for 14FtL 48 cd/M2





COLOR ENCODING AND DECODING

Gamma 1/2.6 Transfer Function

12 bits per channel (see Appendix B) they verified that the models published for eyeball sensitivity make sense in the context of how you'll watch in a theater)

need at least 10 bits, 11 would work, 12 gave good headroom

metadata required for gamut mapping in teh future - isn't required today but will be necessary when you get better display capabilities

equal code value White Point (1/1/1), but supports whatever

luminance is isolated to one component (nice to isolate)

4096 levels

WHY GAMMA 2.6?

it's a good fit for the contraxt sensitivity of the human eye

in a dark surround theatrical watching environment


COLOR MANAGEMENT

metadata is required for gamut mapping

defines the colorimetric calibration of the Ref Projectotor
color Primaries, peak white luminance, contrast ratio (gamma? don't know) is necessary to KNOW to accurately represent

Gamut Mapping in the Cinema Projector

cinema projector interprets this metadata and reproduces the color of the ref projector


in the future, if the color gamut of the master exceeds that of the Cinema Projector, the Cinema Projector is required to map the out of gamut colors into it's gamut

they see how in prepress how color matching systems are implemented, they want to do something better than that. They want that kind of mapping to work in the future, even though it isn't required today

CMS in the Mastering Process
proprietary color processing (3D LUTs or otehrwise) used to match print film or create a look may be included inteh Reference Projection, but

This proprietary color proecessing must be rendered into the DCDM 0 duh - cook your look into your final, don't rely on having tweaked your monitor. Duh. (hey dude- ya gotta tweak the reds on yor monitor for this to look right! NO DON'T DO THAT!!!)

so XYZ is robust, device independent encoding defintion

indepedent of display primaries and white point
supports fut future improvements in color gamut and contrast ratio
DCI StEM test provides proof of concept

SO HOW WILL IT WORK - COLOR CORRECTION ON THE FLY? OR RENDER ONCE ON SITE FOR YOUR PROJECTOR? I DON'T SEE COMPANIES BUYING GEAR THAT HARD?

for their tests, scanned on a Northlight 6K 14 bit, did color grading on a Discreet Lustre with 4K res, 16 bit TIF files

WHAT IS DUAL 292 IN PROJECTION? FIND OUT MIKEY!

OH! OK! I GET IT! THEY ARE STORING THE FINAL FILES AS XYZ COLOR SPACE FILES - NOT AN RGB QUICKTIME! IT IS DECODED ON THE FLY AS 2K Digital Cinema Display Master 12 bit (presently limited to 10 bit in their playback scenario) with a gamma of 1/2.6 in their 1080 server. So some math is required ot playback.

StEM Color Correction

working at 4K is tough today, esp. moving data around

Hierarchical color correction does work well (working with 1K or 2K proxies, 4K batch render)

with files in density space, must apply a film "print curve", film look LUTs were applied in color corrector and "baked into" the DCDM

incorporating XYZ into the color corrector is a future goal

Mike's comments: so it sounds like they want to have XYZ be the native color space in the future

today's color corection methodologies are used to working in RGB since that's how we think. Perhaps over time it might change. Might maybe be XYZ behind the scenes in the future.

They see it as a last step in the process - work in RGB to a certain stage (is this right? Not sure if you've already converted to RGB). -- in any case, convert at the last minute, and put the onus of responsibility

commentary - it's a great, thorough plan, but will the industry be smart enough to adopt it in it's entirety?

Notes on arrival in Vegas 

Vegas!

or rather, ahem...

"Vegas, Baby! Yeah!"

Flew in. As usual, I got selected for the "random search" at security. let's see, large mid-thirties male travelling alone with a lot of technical wiry stuff in a backpack. Strip search! Ok, not that bad, but I had to take off my shoes, empty my pockets, take out my laptop, fire it up, they swabbed it for explosives/cocaine/pixie dust, I had to undo my belt buckle and let an old man feel around the top of my pants (I hope that's a TSA procedure and not just his own little whim.)

Talked the whole flight to Drew, a 22 year old college student who's flown out to see the Phish show.

Longest cab line at the airport I've ever seen in my life - 500 to 1000 people waiting for a cab. I joked with my fellow line waiters that it was like a Disney ride...the only significant advancement they've had in 20 years is what they call "line management" where they keep sending you around turns to hide how long the line really is.

Got to the hotel, Circus Circus. I've been in Vegas once before, during a road trip last summer. I knew it sucked from when I'd walked through before. The Luxor, where I stayed in July, had low end Mercedes up on pedestals you could win by gambling. Circus Circus? P.T. Cruisers. I kid you not. I wouldn't WANT to win one of those.

The staff behind the counter could easily have been interchanged with the graveyard shift at any Walmart in Kansas.

This place is distinctly NOT "so money" in Vince Vaughn in Swingers parlance.

It's going to be fun to see how lame this place really is.

Got up to the room, peeled back the bedcover (you know they aren't required to EVER, EVER, EVER wash those, so they don't. Think about how many gross, nasty, (and we're talking VEGAS gross, here, folks!) depraved, messy acts of, um, congress have taken place on that. Now kick it off into the corner with your hiking boots, not touching it with your hands.

I pulled back the blinds by the little pole, which immediately fell to the floor, blinds still attached. Excellent.

I have an excellent view from my 28th floor room....of the massive RV parking lot out behind the casino. Clue, anyone? Bueller? Bueller....???

On the plus side, I CAN open the window about 10 inches...

(see picture) the lamp is fixed to the table. BOLTED. So you can't move it over. And while there are two electrical outlets, 2 of the 3 lights in the room are plugged into one. The other is 10 feet away...so the laptop sits at af funny angle for me to try to type. Note to self: when travelling tech, bring an extension cord and a power strip...

BONUS ROUND - so I have two plugs usable in the room - and one is wired badly, so I have to get the battery charger "just right" for it's light to come on. Whee!

This should be an interesting week...

Woo Hoo! I'm in Vegas Baby! Yeah! VEGAS!  

OK, I'm not that excited anymore, but that was the general jist of my arrival. I'm stoked to be where NAB is, and that happens to be Vegas. Today was long and very technical. I'm paying for internet access by the minute, so I am making this quick. As soon as I get this up, I'm going to post my raw notes from each session today. And I mean RAW, as I typed them, typos, missed keys, and all. Tonight if I have the energy (mental and literal PowerBook battery power) I'll clean up and add commentary. But for now I'm just going to get a serious case of blogorhea and just dump all my info into the blog, if only as a backup. YES, I will come back and clean it up, but can't do so immediately and wanted to get some data up today.

First I'm going to post some notes I typed up upon arrival, then I'll get into the notes themselves.

I was a doofus and left the cable to download pictures into the laptop back at the hotel, I won't make that mistake tonight. I'm going to cab it to Fry's and buy a voice recorder doohickey for interviews that will download to Mac in a readable format. This may mean an iPod with voice record module. If I had one, it wouldn't be the end of the world, you know.

Friday, April 16, 2004

barefeats.com (Go Robert!) Does a Very Useful set of tests on hard drives, plus some analysis 

Robert over at barefeats.com does a lot of excellent drive testing and we compare notes from time to time. He delves into all the drive throughput issues that are crucial for HD type stuff.

Today he has an article on what are the fastest drives to use.

Mike's Comments: As expected, the small (74GB) Western Digital Raptor 10K (10,000 rpm) SATA drives were faster than the Hitachi SATA 7200 rpm drives at 250 and 400 GB by a decent margin (read the article, I won't spoil the surprise). The good news is that the new Hitachi 7K400 drive is as fast as it's smaller 250 GB cousin, so that bodes well for large arrays. Of interest was also that the Diamond Max Plus 9 (160 GB tested but available at up to 250 GB) performed similarly to the Hitachis in RAID configurations but not in single configurations. It would be worth checking out it's price/performance in comparison.

Rob also points out that with the new Swift kit (read his review, you can put 3 more (for a total of 5) SATA drives INSIDE a G5.

This would allow for a standard SATA boot drive and a 1.6TB RAID 0 internal to a G5 using the Hitachi disks, which would give enough throughput for 10 bit 1080p24 work, but somewhere along the way as the array got full it might not be fast enough. So partition it with SoftRAID.

Or you could boot from FireWire drive and use all 5 internals and be fast enough (somewhere around 130-150 MB/sec at the end of the array) for single channel work with a total capacity of 2GB.

Hmm, this may be my new bang-for-buck champion solution.

-mike

Thursday, April 15, 2004

ProMax Systems introduced the ProMedia Converter for analog and digital audio/video formats, with uncompressed video over FireWire, Genlock, and SDI with 4-channels of embedded audio. [MacInTouch]



Mike's first thoughts: After reading about the DeckLink Pro HD for only a few hundred dollars more, this product doesn't sound so awe inspiring as it might once have. It does have the significant advantage of being able to handle DV, SD SDI, and BetaSP uncompressed footage over FireWire as a 2RU rack mountable device, but with so many HD options, the price point seems questionable. Then again, standard definition video is the VAST majority of the market



This seems to be their answer to AJA's I/O unit, which has very similar specifications and features.



For standard definition editors, this is definitely a good thing, and choice is always nice in the marketplace. I think they may have taken a bit too long to get it to market, however. Three to six months ago this would have been substantially bigger news.



I'm getting the impression this is going to be the Year of Inexpensive HD at NAB this year, and this product just doesn't quite fit the marketplace pricing of the moment.



I'll be checking this out, and open to re-interpretation once I see it.



But to be sure, rackmountable FireWire is MUCH MUCH MUCH more convenient than a PCI card, and I laud their accomplishment in that regard - infinitely easier to deal with in a small studio environment, and let's face it - that's where the market is going.



-mike

Technology-focused private equity firm Accel-KKR will acquire Alias from SGI for $57.5 million as pa... [MacNN | The Macintosh News Network]



Mike's Comments: Not of explicit interest for HD types, but big news for anyone in content creation, especially 3D artists. This is one of the two biggest high end 3D packages. It makes sense that it is moving from SGI, which is struggling to find a new business case (aka a reason to exist) and probably needs the cash.

Digital Heaven today announced eight plug-ins for Apple's Final Cut Pro v4 and Final Cut Express v2,... [MacNN | The Macintosh News Network]



Mike's Comments: Of small interest. Some dropout and dead/empty pixel correctors, some grid generators, simple stuff. Might come in handy, but nothing earth shattering here. But the plugins are available for $10 each, which is a nice way to sell these things. Need a quick tool along the lines of the right sized screwdriver? Just download and buy and you have it. I like that business model, even if the tools here aren't terribly exciting. I'm being overly cruel. They could prove themselves quite useful.

Wednesday, April 14, 2004

Did I mention I have an Atom feed? ...and I'll still be posting live from NAB? 

Atom feed here.

In plain English (OK, HTML) it is:

http://www.hdforindies.com/atom.xml

I'll be posting as I learn cool goodness at NAB, also.

So get your Atom browser set up, and start checking Saturday, as I'll be at the Digital Cinema track on Saturday and Sunday, then posting from the trade show floor on Monday through Thursday (assuming good wi-fi access, that is).

So what is Atom? It's like RSS. What's RSS? RSS stands for Real Simple Syndication. It puts out a blurb whenever you update a site with an RSS feed. So if you subscribe to a feed, you know when a site has been updated. Atom is like RSS, but is Google's preferred format, apparently in the Microsoft "Embrace, Extend, and Crush Like A Grape" style. There are talks to merge the two formats.

My friend, Charlie Wood, who set up the Atom to RSS converter wrote an article about all this at on his site.

-mike

Did I mention I have an RSS feed? And I'll be updating live at NAB? 

RSS feed here.

In plain English (OK, HTML) it is:

http://www.hdforindies.com/rss.xml

I'll be posting as I learn cool goodness at NAB, also.

So get your RSS browser set up, and start checking Saturday, as I'll be at the Digital Cinema track on Saturday and Sunday, then posting from the trade show floor on Monday through Thursday (assuming good wi-fi access, that is).

-mike

Backup strategies & risk analysis for HD quantities of data 

So after my last post, and thinking about the risk of losing 3 or 4 terabytes (3000 to 4000 gigs of data), backup is an issue. I did a bunch of in depth analysis last year on the cost of various backup systems, and came to the conclusion that tape (the data backup kind) was dead as a cost effective backup solution. Granted, there are some arguments to be made for using tape (I still use it to archive all my installers for everything), but I no longer archive projects to digital data backup tape.

Over a year ago, buying an external FireWire drive (even if running out to Fry's and buying an external case and whatever deal they had going for a large ATA drive) is BETTER. Because:

1.) It's massively faster - the most cost effective tape solution available when I bought was AIT 1. Now there is AIT3, and it is good for 12 MB/sec maximum, and that is a virtual rate if it's pulling compressed data. Digital video doesn't compress any further worth a damn. But being generous, 12 MB/sec tops. Any decent FireWire drive is going to be at least twice as fast as that.

While I was looking, I noted S-AIT tape drives available that can hold 500GB of data on one tape. Wow! Cool! How much!? $12000 for the drive. Never mind. I won't even ask how much the tapes are.

2.) Less infrastructure necessary - AIT requires a SCSI interface, $250 and up. You already have a FireWire port, and if you don't, they're about $35-50 at Fry's or anywhere.

3.) It's STORAGE, not STORED. So if you do have 500 gigabytes of data on tape, you then have to have 500 GB of SPACE on some drives to put it onto. More cost if you don't have that capacity lying around. Tape is STORED data, drives are STORAGE. So if you need to look at it, mess with it, it is THERE on the drive to be read, not locked into the tape unreadable until it is put somewhere...a time consuming task.

4.) It's just plain cheaper - when I ran the math over a year ago, the breakeven point was around 20 terabytes of data for tape to be cheaper than drives. Drives have continued to get cheaper since then, frequently 80¢ a gigabyte or less.

5.) Drives are easier - pull it off the shelf, plug it in. Find your file. Look at it. Copy it over to your high speed storage if you need to. DONE. On tape? Fire up the SCSI tape drive. Fire up Retrospect (or whatever software you use). Do a search, flag the files you want, start retrieving them. Wait and swap out tapes. Wait and swap out tapes. Wait and swap out tapes....

So FireWire drives are better. How does that help me?

Let's say you're working on your HD masterpiece. You don't want to run the risk of losing all your valuable edit decisions. If you store your NON-media assets on another FireWire drive (your FCP project file, your Photoshop art, your After Effects files, anything else that is NOT digitized footage), it is safe (relatively). Still do regular routine incremental backups, even if just to another FireWire drive. It's easy to configure to back up at night unattended.

Meanwhile, you have your gigantic, stupendously fast POD (Pile of Drives) that some lunatic on a blog convinced you to buy. It's RAID 0, so if any one drive keels, U R FUCT. Since it is up to 10 drives, the likelihood of the array going down is 10 times more than a single drive. Making up numbers here based on my own experiences, if a drive has a 50% chance of hosing once every couple of years, that means there is a 50% chance of your array hosing every 2.4 months. Egads! Is that math right? I hope not. I don't know, I need to research that a bit. Don't take that math as bible truth, please.

So let's assume it is. If you ONLY have digitized footage on your array - in other words, just stuff you've digitized off a tape - none of that is information losable forever. The timecode to recapture it is in your FCP project, which is NOT on the array, and that data gets backed up regularly (even if just to a CD). So if the array does hose, once you have it up and running you'd need to redigitize all that footage as a batch operation in Final Cut Pro.

Would it suck?

Yes.

Would it take days?

Yes.

What if I had to rent an HDCAM deck again to capture all that stuff?

That would both suck AND cost money, and cost me a couple of days.

OK, what would it cost to NOT have to worry about that?

Solution 1: You'd have to have a RAID, at least Level 5, more likely Level 50. That's going to cost 2 1/2 times more than what I've suggested, and probably not be as fast.

Hmm. _I_ don't have that kind of money for my system. That pushes our $18K system to $25K.


Solution 2: You'd have to have a means of backing up all that data. A big pile of FireWire drives and Retrospect (or other software) could solve that. They don't have to be fast, just cheap in terms of cost per gigabyte of storage. They do need to be fast enough to back up any given day's worth of work overnight while you sleep...but that's not an enormous requirement. Any decent FireWire 400 drive can do 20-30 MB/sec if it has an Oxford 911 bridge chipset and a decent drive.

Solution 3: Risk it. The cost/benefit analysis scenario: If the odds are, say, 1 in 10 (or 5, or 3, or whatever) that my array will hose during my post production scenario, what is the incremental cost of fixing that problem? If I have to rent a deck again and spend a couple of days recapturing all my stuff, that is expensive and time consuming. Balance that against the 1 in 10 (or 1 in 5, or whatever) odds that it will happen. Multiply cost times risk ratio. If deck rental is $2K (I have no idea) and the risk factor is 1 in 3, then it's a $666 risk for my production, plus a couple (or 4) days of my time. But I saved $8000. Maybe it's worth doing if I don't have the money. Now if you were foolish and saved your FCP project, all the backups of it, all your original art you created, all your digital notes etc. on the array, U R FUCT.

But if you didn't, you say dammit, rent the deck, replace the hosed drive (cursing it liberally), and off you go again.

I'm not saying this is a pretty solution, nor one I'd choose first. But it is a solution that brings the cost into the realm of the plausible.

Okay, somebody email me at mike@hdforindies.com and ream me out for how foolish this is...but explain why in detail...

-mike

Pre-emptive guess at current best bang-for-buck HD system. 

OK, after reading and writing about the new offerings coming up at NAB, here's what I'd say is the most powerful bang-for-buck HD system:

PowerMac Dual 2.0 GHz G5: $3000
Apple Cinema Display 23HD: $1500 after $500 back for buying with G5
2nd Apple Cinema Display 23HD: $2000
more RAM: $300
VGA display for palettes, etc.: $300
External 250GB FireWire 800 drive to boot from and store non-video assets on: $400
10 Hitachi 400 GB drives (these are the new, not quite out yet ones): est. price each: $500*10=$5000
8 cases to put 8 drives in (or get a RAID chassis): 8*$50 each =$400 from PPA, Inc. (get at Fry's)
SoftRAID RAID software if not using Apple's (is faster): $100
2 (none shipping for Mac as of yet) PCI-X 4 external port SATA cards:2 X $250 (est) = $500
DeckLink HD Pro: $2500
HDLink: $1300
cables and whatnot: $200
Final Cut Pro: $1000

So this gives a 2 monitor, 1 HD display editing system capable of at least 10 bit * editing and color correction, with offline HD resolution capabilities, 4TB of online storage, enough storage to work with a 2 hour feature at 4:4:4 12 bit, throughput capable of handling 12 bit 4:4:4 footage (software compatibility willing, don't think it is yet), blah blah blah blah. Super high end, rockingly cool, film color correction capable. Oh, and it includes pixel for pixel monitoring.

How much?

$18,500

Of course, this would be a RAID 0, so if ANY one of those 10 striped drives goes down, your data is HOSED.

And it would be an ungodly cabling mess, AND it would be big loose stacks of drives in cheap cases getting very hot. OK, this isn't such a great idea. So I'm working on finding a better casing/housing solution for the drives, because this isn't very realistic in a professional environment.

So what to do about that? Either you risk it and only have digitized or rendered footage on those drives, recoverable in the sense that you can redigitize (risky move), or consider a mighty stack of FireWire drives to back all that stuff up on a regular basis. Eeek, that sounds awfully messy and non-high end professional. That would cost another $5000 or so. Hmmm...then it's almost up to the cost of a 3.5TB X-RAID from Apple, at around $12500 properly configured with interface card. But Apple's wouldn't be as fast...somebody told me they got consistently over 190 MB/sec with an X-RAID. The system I'm talking about with the 10 SATA drives should ALWAYS be capable of at least 270 MB/sec, going up to 500 MB/sec peak transfer rate in theory. I've only striped up to 6 drives at a time with SATA on a G5, so I don't have enough data yet to say for sure. But something I'd like to test.

But as for the RAID issue, I'm reluctant to spend nearly $13000 for a 3.5 TB array that does 200+MB/sec. Granted, that is one of the best deals going as far as I can tell - rack mount, RAID 50, all kinds of management stuff, yada yada yada.

So I'm planning on going for a homebrew solution after I get back from NAB.

I have a 14 bay SCSI-1 RAID chassis I picked up for $15 for (literally) a brother-in-law deal. His business was closing it's Austin office and needed to get rid of stuff, they didn't know what it was worth. It's great - 14 bays, dual power supplies, big fans, little shelves for the drives (not enough though!), keylock for the glass front and the sides.

So I'm thinking of trying to find SATA power & data connections to mount in there after gutting out the SCSI stuff. The problem is it is SATA, so it can't be more than about 3 feet from the computer. That many drives with fans are going to be LOUD. Hmm. Definite Fry's research trip to be made...but when I do it, I'll document and photograph it all as I go.

* (waiting on better codec I think, not sure all that's worked out, even though there is the MicroCosm codec to render to)

More Great Decklink news - HD SDI to Apple Cinema23HD display adaptor, only $1300! 

Updated: here's a link to their page on it, and here's a link to the PDF data sheet.

Also in an email from Decklink today:

HDLink

This is a new and exciting product. Ever since I saw the quality of full
motion video displayed on the first Apple flat panel LCD Cinema Displays
displays back when we started doing uncompressed SDI 10 bit back in 2000, I
thought they would make nice video monitors for television.

We have now completed a new converter called HDLink that features Dual Link
4:4:4 input, and displays the video full screen on any supported LCD DVI-D
based computer monitor. A 1920 x 1200 LCD display such as the Apple HD
Cinema Display is perfect for displaying 1080 HDTV images.

HDLink auto switches to 4:2:2 HD-SDI or standard definition SDI. It makes a
perfect companion to DeckLink HD Pro allowing a complete 4:4:4 based
production station at very low cost. Every HD pixel is digitally mapped to
the LCD pixels, so it's a perfect HDTV digital monitor.

One of the problems with displaying video on LCD monitors is not all frame
rates we really need when working with all the frame rates of HDTV are
supported on DVI-D displays. The Apple HD Cinema Display is a good example
of this, and it only supports about 60 fps.

So HDLink builds in a PowerPC based pulldown processor for adapting frame
rates to the supported rates of the display. Unlike other products it also
includes audio de-embedding with variable audio delay to match the
pulldown processor delay. This eliminates AV sync issues.

This is very exciting product, as HDLink makes a complete audio and video
monitoring solution perfect quality, and low cost. HDLink's retail price is
US$1295, and it will be available within a week or so after NAB. We will be
showing HDLink on our booth at NAB.

All the info on these new products are on our web site in about an hour's
time, so please take a look for more info. Also, if your traveling to NAB
this year, please drop by the Blackmagic Design booth for a demo!

Mike's comments -this looks like a great product - the only other competitor I've heard of costs $7000. Yet more support for the argument of Don't Buy It Till You Need It. Or at least rent if you aren't going to use it constantly...

From their web page on the product:

For matching LCD display colorimetry, independent 10 bit RGB gamma tables are fully adjustable via a high speed USB 2.0 host computer connection. Customizable gamma tables also allow film industry log video to be converted to linear for monitoring when used for feature film work.

Control software is available for both Mac OS X™ and Windows XP™, with USB 2.0 connectivity allowing firmware upgrades via internet downloadable updates.


...so that's good news to handle the CLUT (Color Look Up Table) issues.

Wow - HD 4:4:4 12-bit card from Decklink, the Decklink HD Pro for $2500, plus analysis 

UPDATED twice on Wednesday: see analysis at bottom, and here's link to their page on the product, and a link to their PDF data sheet.

Got an email this morning from Decklink:

I wanted to update you today on our new products for NAB. We have a new
HDTV 4:4:4 capture card called DeckLink Pro HD, and a new dual link SDI to
DVI converter called HDLink we are announcing.

DeckLink HD Pro

This is our new high end DeckLink card, and it includes the most amazing
quality possible. We have taken the basic DeckLink HD card, and massively
increased the connectivity and processing power on board.

We have dual channel SDI connections for 4:4:4 capture and playback, and
with PCI-X 133 MHz speed, we have designed this card to support up to 12
bit RGB 4:4:4 video quality. This is the maximum the SDI television
standard can support.

The card can also be switched to standard definition, or normal 4:2:2
HD-SDI modes.

The analog monitoring added to this card is also very exciting. It includes
14 bit digital to analog conversion, and this is designed to handle clean
monitoring of 12 bit log material that will be used by the film industry.
Also, at the quality of HD 4:4:4 used on this card, a lot of people will
use video projectors with very high contrast ratio's, so having the deepest
bit depth for monitoring is important.

Of course if DeckLink HD Pro is used for standard definition work, the
quality is way over the top, however at the price we think a lot of
standard definition work will be completed on DeckLink HD Pro cards!

The analog monitoring also includes full 4:4:4 support when working in that
mode, and when running in standard definition the video is 5 x over-sampled
for incredible quality.

We really have gone all out on DeckLink HD Pro to design the best quality
card that could be made and it's technically way ahead of any other product
on the market today. It's very exciting!

DeckLink HD Pro has 4 BNC connectors and a short break out cable with the
analog monitoring, AES/SPDIF input and output, audio work clock out,
genlock/trisync input, and RS-422 deck control.

The card will retail at US$2,495 and we will be showing it running on our
booth at NAB. It will be shipping in about 3-4 weeks.

Mike's comments - damn - this looks like an awfully good deal. Also of note is the fact that it does analog monitoring, so a $2000 adaptor isn't required any longer.

Several good things in this:

1.) It's 4:4:4, so no color downsampling. Your pixels are your pixels, no mushy colors (unless you shot it that way). Yes, it requires a fairly stupendous setup, but it gives you the possibility of doing highest possible quality work.

2.) It supports up to 12 bits per pixel. This means you have a lot of lattitude to do color correction for digital cinema type uses. I'll be talking at length at the booth with them about this stuff.

They mention the card supports 12 bits per channel per pixel...which means 36 bits per pixel...which is pretty non-standard, and more than the typical 32 bit size...and they DON'T mention a codec to support that as yet.

Assuming one did exist, the data rates on this sucker would be enormous - their existing 4:2:2 codec at 10 bits takes up a lot of space already -

10 bit 4:2:2 @ 1920 x 1080 @ 24fps (cinema frame rate) = 109.43 MB per/sec, or 394 GB per/hr.

Multiply that by 1/2 again more to account for the jump from 4:2:2 to 4:4:4 and that is:

10 bit 4:4:4 @ 1920 x 1080 @ 24fps (cinema frame rate) = 164.145 MB/sec, or 591 GB/hr.

Now add 20% to account for the jump from 10 bit to 12 bit and you get:

12 bit 4:4:4 @ 1920 x 1080 @ 24fps (cinema frame rate) = 196.974 MB/sec, or 709.2 GB/hr

...so your 2 hour movie takes up 1.42 GB of space. Eeeyowch.

And at about 200 MB/sec, that requires one colossal drive system...and a lot of them.

Also of note for the film crowd is the line:

14 bit digital to analog conversion, and this is designed to handle clean monitoring of 12 bit log material that will be used by the film industry.

...sounds like these guys are right on target.

and a nice quote from their website:

If your in a smaller facility and you hire in a deck when it's required, or your in a large post or broadcast facility where you often lose deck access when it's routed to another suite, then you will really appreciate independent monitoring. DeckLink is the only card that lets you keep working when the deck is unavailable.

...yep. What we need. Sounds like they really took a good look at what has been holding back HD for small shops. Now if they only came out with their own affordable RAID solution....

NAB 2004 looks like it's going to be a HUGE year for low cost HD post. I'll be checking it out and giving a full report once I get to NAB next week...

Monday, April 12, 2004

5 SATA drives in a G5 - woops, missed posting this one 

This is a couple of weeks old but still of interest. Robert (he rocks) over at barefeats.com posted this article about Trans Intl's new adaptor to fit 5 SATA drives in a G5. That gives you one boot volume and 4 drives to stripe, a nice setup for high data rate capture needs. He also has some commentary about similar products.

Read the article here.

And his article from 1/30/04 on FireWire vs. SATA vs. SCSI for 4 drive arrays is still a valid assessment and of interest.

More products to watch at NAB... 

From CreativeMac.com:

Aurora rolls out a new $500 standard definition card, Pipe SDI.

Not HD, but good news - as HD becomes the new "gotta have" technology, the price of standard definition gear, even the high end SDI stuff, is falling FAST.

Canon rolls out 6 new HD lenses and anamorphic converter.

...and yet more lenses from Canon for ENG/EFP.

More choices in lenses is good.

Saturday, April 10, 2004

interesting thread on Mac OS X RAID setups on macintouch.com 

Update: With another interesting update today.

Macintouch has a user thread on Mac OS X RAID setups. It starts getting interesting around the April 9th post.

Discusses RAID 5 OS X compatible setups, what companies are using the same physical guts to set up their hardware controlled RAIDs, and other very interesting and useful tidbits.

Review of color correction training tool for Final Cut Pro 

DigitalPostProduction.com has a review of Practical Color Correction, training tools to use color correction in Final Cut Pro 4. Only $35.

Friday, April 09, 2004

DALSA's Digital Origin camera to be demo'd at NAB Digital Cinema Conference 

Dalsa's Origin digital cinema camera will be demonstrated at the NABannual convention during the Digital Cinema sub-conference.

The Origin camera shoots 4K and 2K images, very high quality, very not cheap. But a technology to watch - this is the way things are going to go in time in terms of digital cinema cameras.

The unit will also be demonstrated on the show floor during the trade show.

Click here for the info from Digital Producer Magazine.

This is interesting news, also because last year they showed a demo prototype unit, and now they are advancing towards shippable products.

Wednesday, April 07, 2004

Way cool - Magic Bullet for Editors (runs on FCP, not AE) 

UPDATE:Here's a direct link to the Red Giant's page on FCP for Editors

Magic Bullet, the suite of tools for handling a variety of tasks for taking video to film, now has a plugin for Final Cut Pro, Magic Bullet for Editors. Their previous plugin was only for After Effects. From creativemac.com's website:

"Magic Bullet for Editors is a plugin for Final Cut Pro designed to emulate the look of film processes and treatments. It includes stock emulation and diffusion and offers 50 pre-designed film looks and 14 plugins for adding film damage characteristics. It also offers Look Suite for creating or manipulating film looks. And it includes Misfire, a tool for creating custom film damage effects, such as grain, scratches, splotches, flicker and gate weave. Misfire includes 13 parameter categories for controlling film damage."

New import plugin from Automatic Duck for combustion 3 

From their email:

With Pro Import C3, you will be able to import OMF® and AAF files from Avid® and Final Cut Pro® editing systems into combustion.

Media 100 will show Media 100 HD at NAB (& other stuff, too) 

Media 100 will demonstrate the 844/X Version 3 and Media 100 HD at the National Association of Broadcasters (NAB) conference, which runs April 17-22 at the Las Vegas Convention Center... [MacMinute]

New software tools to help with show production (of minor interest) 

Intelligent Media Consultants -- a.k.a. Todd Fantz -- on Tuesday released VidiSuite, which is described as a "work in progress" application suite aimed at television broadcasters. The software consists of several freeware applications and a commercial product. [MacCentral]

SERIOUS RAID bug with 10.3.3 - some RAIDs won't mount 

There seems to be a serious RAID bug in Mac OS X 10.3.3 Update. [MacInTouch]. Looks like a SCSI related problem, but can occur with SATA & FireWire as well. Hasn't happened to me yet, booting from a FireWire 400 drive with OS 10.3.3 on it and mounting a 4 drive SATA RAID 0 array.

I'm going to NAB, will be posting from the show. 

I will be going to NAB to check out all the new stuff and posting reports from the show on all the interesting new tidbits & toys I see. Heard of something cool I should be checking out? Email me at mike@hdforindies.com to fill me in.

Interesting new RAID 5 chassis with SATA, USB 2.0, and FireWire connectivity for $1500 

Update 4/14/04: Re-reading their website, I noticed that you can connect this to your computer via Serial ATA (SATA). The website claims:

USB2.0, FireWire/1394, FireWire 800/1394b and Serial ATA; transfer up to 150MB/s

...so I emailed them and asked about some stuff.

Things I learned:

RAID 5 cannot exceed 5 drives

Multiples of these things can't be daisy chained to make a bigger, faster array. So 150 MB/sec is max...if it can even do that. From their email, "RAID5 is a tricky beast and its hard to put a set benchmark number to it. SATA is capable of that xfer rate."

Update: Just emailed with their sales staff, even though the website says Mac OS 9.2, it supports 9.2 AND UP. So 10.3.x support is present. Website to be fixed directly they tell me.

FireWire Direct today announced a new RAID chassis capable of RAID levels 0, 0+1, 1, 3, 5, and JBOD. It holds up to 5 ATA-133 drives (no capacity limit on the drives), has a dedicated controller, and a bunch of other interesting features. It's $1500 for the chassis alone. I'd be interested to see the performance of this one once loaded up with fast ATA drives. The big deal is that it supports RAID Levels 3 & 5 which handle data redundancy in case of a drive failure. RAID 5 lets you do distributed parity, such that if ANY one drive dies, the data can be recompiled from the other disks (the 4 others with this particular unit).

This gets especially significant once you start getting serious amounts of storage together. If you have, for instance, a 10 drive array, and the odds of a single drive failing are once every 10 months (not a realistic statistic), then you have multiplied your risks by 10 fold...or once a month. RAID levels 3 & 5 give you a safety net, so that even if s drive fails, you're covered.

Final Cut Pro user group meeting during NAB 

The Final Cut Pro Users Group will be gathering at the Stardust Hotel in Las Vegas on April 21 during NAB in Las Vegas... [MacNN | The Macintosh News Network]

Linux based HD DDR/editor - not something I'd use, but interesting nonetheless 

Slashdot has a link today to an interesting article about a Linux based DDR/editor currently being used on Starship Troopers 2. A virtual deck, an editor, capable of RS-422 control (deck emulation) and other interesting goodies. Based around a dual 2.4 GHz PC and the AJA video card with custom drivers, an interesting device.

Friday, April 02, 2004

Re:Vision Effects has released Effections, a bundle of RE:Vision Effects' products for use within Ad... [MacNN | The Macintosh News Network]

Thursday, April 01, 2004

Storage peripheral maker WiebeTech LLC on Tuesday introduced its Quad BayDock 800. The device is a four-drive FireWire 800 enclosure suitable for RAID applications on both Mac and Windows computers. [MacCentral]
Here's a warning about issues involved in expanding Xserve RAID storage. [MacInTouch]

This page is powered by Blogger. Isn't yours?

Listed on BlogShares