HD For Indies

I temporarily pulled my original article. I'm editing it a bit to more accurately qualify my statements. I've tried to reach the folks with the primary beef with what I originally posted, but after numerous attempts to contact via email and phone with no response, after over 72 hours I'm now putting this back up, edited a bit. In short, the original version of this article went up after a specific request to get online ASAP, and that created some false impressions in the minds of others, not helped any by some overzealous and/or indistinct phrasing on my part. If you read the original, you'll note this is about 98% of the same copy, just some qualifiers and clarifications thrown in on top. This is the first of a series of articles I'll post about my trip to LA now that I'm back. I've got a busy week coming up, so I'll be transcribing my notes and organizing my thoughts and doing further writeups between some paying work and long overdue deferred life maintenance, but it is all coming. Patience.

First off, let me start by saying that in my past Red coverage, I've often been told that I'd lost my objectivity, accused of drinking the Kool Aid. Well, read this, and then think what you will. Keep in mind I'm attempting to be conservate in my considerations here (but its tough - it's a really exciting product that looks like it is coming along great). So short version - I've seen images off the Mysterium sensor that is to be used in the Red camera, and even it its very first baby steps in a lab type environment, it is AMAZING. Long version...

...in a nondescript office building, in an uninteresting office park in southern California, I Saw Something.

I can't say all that I saw, where I saw it, or what exactly shot it, but suffice it to say, after nearly two weeks in LA, seeing footage at the DGA Digital Day on a Sony 4K projector; visiting Band Pro, Clairmont Camera, Plus 8, Panavision, and Dalsa; talking to reps, techs, VPs, and execs about their latest and greatest; and seeing a LOT of test footage from a slew of cameras....Red is onto something here. Something BIG. Something that if they continue to fulfill their stated specs and objectives, could and probably should substantially alter the landscape, ESPECIALLY for indies.

(Note for clarification - I saw GREAT looking footage from Genesis and Dalsa while I was out in LA visiting their facilities which they were kind enough to take the time to show me, I'll have more to say on them later.)

The kind of thing where if this goes right for Red, future film school nerds sitting in History of Digital Cameras 101 will be shown a graph of historical bang for the buck on one axis and time on the other. There would be a dramatic bend and point of inflection on the chart, and the professor would point at that sharp break in the line with his laser doohickey and say "...and this is when Red came out."

I happened to come visit Red on a perfect day - they were shooting test footage, some of their first from a full sized, "real" Mysterium sensor. Jarred Land of DVXUser.com was also there (pictured above from right to left are me (Mike Curtis), Jarred Land of DVXUser.com, Ted Schilowitz of Red, and the Red surfboard, and no this was NOT taken with the Red camera!.)

The test footage I saw off the Mysterium, which was early, straight-from-the-lab, pre-beta, not optimized, "we just turned it on and there it was" type of stuff, shot on a low end lens (some from a Canon EOS lens) makes me think that Red is truly going to be a force to contend with, that it will compare favorably with existing high end digital cinema cameras...at ANY price. I also think it likely that it will, IF they continue on the path that they've publicly and privately described to me, be considered at least a peer, if not a preferable option, to the existing, shipping, high end digital cinema cameras.

Many have said, based on the specs published at NAB, that "They just can't do it, because it can't be done." Let me tell you what I saw.

I saw images of a man and child side by side with beautiful, GRAIN AND GAIN FREE smooth, clear, clean skin tones. And I mean NO noise/grain/gain, just smooth imagery, looking at it pixel for pixel on a 30" Apple computer LCD display from a foot away. Framed from mid upper arm to over the head, I could make out EVERY hair, not just the ones the light caught just so. I could see the subtlest details in skin texture. I could see the subtle veins in a young child's eyes. I could see the pucker of skin around the man's facial hairs that needed shaving. Limned with a back light, I could almost make out the individual hairs on the edge of his ears, those tiny sub millimeter long ones we all have if you look close enough...and those weren't even necessarily in the focal plane. But I could discern that it was hair, not backlit skin, or highlight glow or anything like that. I could see shadow detail in his salt and pepper hair, and the only thing that looked like it might be blowing out was the highlights on his shiny, unmade up, unwashed forehead (there is a heat wave on, don't forget). There might have been data in that highlight to recover, I don't know for sure (the working bit depth that they are capturing is...quite impressive).

...and that was shot with a Canon EOS lens. I was at lunch describing the detail with someone, and realized I could, when peering close at that footage on screen, make out about as much detail as I could sitting across the small table at lunch, right there in person.

(I followed up with Ted to double check - that image WAS shot with the sensor rolling at 24fps - so NOT "just a still")

Now, I've recently seen footage from Sony's F350, F900, and F950; Panavision's Genesis, the Dalsa Origin, and the ARRI D-20. I saw that footage either on high quality monitors, or projected digitally or from film prints, either in well set up screening rooms or in everyday movie theaters, so it isn't quite a comparable viewing environment to what I'm describing here. I haven't had the chance to analyze footage as closely and consistently from the other cameras as I have here, nor side by side. But I was very, very impressed with the Red images I saw this way, and I've zoomed in tight and analyzed a lot of footage lately from some other HD cameras.

I watched the Red team, in a Batcave like warehouse setting shoot FOOTAGE (clarification - I'd said images before) of an Oakley watch on a turntable (this was shot with a Zeiss lens) - the buffed metal organically rounded links in the band showed no blown highlights, just verrrrrry smooth continuous tone, without the usually harsh video clip and falloff. At first I saw in the RAW Bayer image what looked like noise...but it was actually striations in the finely buffed metal - wow. The kind of detail you wouldn't even know existed on SD or maybe even a 720p camera, and it was abundantly clear here. Very clear subtle shadow detail in the patterned black watch face, fine white lines of the minute markers on the face crisply defined, and a bright red sweep hand...shot at 1/24th shutter while the watch was on a turntable IN MOTION. And it was amazingly sharp....even only using a portion of the imager's surface area. (At this point, the problem wasn't the sensor, but rigging up a disk array fast enough to capture all that data from full area of the sensor on set...so they had to window the sensor.)

The test footage was a bit dark, but that is easily attributable to the fact that the imager is not yet characterized, this is the rawest of possible output off the sensor that has just barely been manipulated, and they are just starting to get into their tests. This was not a staged demo of what they'd done before for my benefit, I just happened to show up on The Right Day and got to watch as they got it all going and they themselves gathered around to eagerly watch the results.

To my eye (and I admit I'm not cinematographer, but I've been working with digitized images from film and video for a living since the earliest 1990s *** see Footnote at end of article to qualify this a bit), this looks like they may well have one of the cleanest imaging video/d-cinema sensors I've ever seen pictures from. (That line is what I orginally wrote Wednesday night. Friday morning I saw footage from the Dalsa Origin - more on that later.) As well as a just INCREDIBLY sharp picture. As for exposure lattitude, required lighting levels to get these kinds of images, what aperture required etc., I don't know, I wasn't privvy to all the details involved, so that is an area that creates some wiggle room in terms of just how rockin' this camera is and just how significant that demo footage was. (After writing THAT line, two days later I saw an extremely impressive demo of lattitude from the Origin camera that I'd like to see anyone top.) As I learned at the Texas HD Shootout, nothing counts like controlled, well documented circumstances and side by side comparisons. But the Red footage (and FOOTAGE, not just still frames) looked damn, DAMN good, shooting a watch with metal links and a black face, and maintaining good tonal range throughout.

What they DID say was that they rated the camera ISO 320. I heard Jim mentioning F8 to get the depth of field he wanted in the shot.

Now, I saw shots from some of the other existing D-cinema cameras that were blowing out, but I don't know (and don't feel I have the judgement to know) whether that was the limits of the camera or a choice from the DoP on set. (This was demo footage at other facilities whilst in LA, shot by clients, not by vendor staff.) To really make a truly fair and accurate comparision, I personally feel I would need to shoot the same subject, from the same location, with the same lens (if possible), and screen in it in the same fashion at the same time, preferably with butterfly comparisons. (Or, obviously, shoot some charts and over/under subjects in a controlled environment for some of that info.) Since that can't be done yet, I need to withhold final judgement. All of the cameras I've seen this week (D-20, Genesis, Dalsa, Viper, Red) can make lovely, sharp, wide lattitude shots. I don't feel comfortable or confident to say which were better at what - the Dalsa, for instance, records 4K but they could only screen 2K with their Christie projector - so who knows if it was noticeably sharper than 2K or not. But I do feel confident in saying that if you were to plot out image sharpness and dynamic range, Red would be in the same area as the others mentioned. I'm not saying it would do worse, I'm saying it is competitive, but I don't feel comfortable declaring which falls where at this time with the knowledge I have. Plus, Red has only spent a very limited time, not years, optimizing their imagery off of this sensor.

What I saw from Red looked like a very nice professional digital SLR still camera image - and these were frames shot at 24 fps with 1/24th of a second exposure of a moving subject- so no closing of a gate or shutter, no black level resetting or flushing of an imager's CCD/CMOS buffers (that's one reason digital SLR images look better than video) - just REALLY nice, clean, smooth images.

I tend to look at a source image not so much for what it IS, but what it can BE - as someone familiar with how far you can (or can't) push a source image in color correction in post. While it is nice to have pleasing color "off the truck," beyond that I look for good shadow and highlight detail so I can push an image around in post to arrive at an ultimately pleasing image. I saw extremely good things in the footage from that regard.

Based on what I saw, I'm pretty confident that they should have some 4K footage to show at IBC (as they've stated was their intent), and that it'll make jaws drop. It sounds like they'll have something NICE planned to show it off.

So what do I think? Based on what I saw that day, I think Red is well on their way towards having an industry altering product here if they continue to deliver on their stated goals. A product that will force the other manufacturers to either respond by nuking their existing pricing structures, or as the market catches on to what Red offers (again, if they deliver as stated), the competitors will lose a significant number of sales to Red once the market realizes what Red is all about.

Along those lines, note that Panasonic has been lopping off prices on cameras lately, notice the $10K from of the SDX900 from about $27K to about $17K, to make room for the new high def HVX900 at the old price point of $27Kish. What will the other industry players do if Red arrives and delivers as promised?

Note that I don't mention how it compares to any camera in its price range, like the SDX900, the Sony XDCAM HD cameras, the specs for the unreleased HVX900, or the higher priced Panasonic Varicam nor the Sony F900. I don't think those will even be in the same league. I think both in overall net benefits of the package and individually category by category, Red is going to be either an outright winner or in the race with the higher end d-cinema cameras.

Comparing to what is known or expected to be on the market by the time Red is expected to ship in volume, Red wins in terms of:

-maximum frame rate capabilities - from low to high in fine increments (except for the high end specialty cameras that are impractical to do all your shooting with, although somebody recently said Phantom could be practical to shoot long form with - so I need to investigate that).

-resolution flexibilty (720p, 1080p, 1080i, 2Kx1080, 4K, 4.5K)

-shooting/recording mode options - from relatively low cost on board compressed recording for 720p/1080p/1080i/2K in 4:2:2 or 4:4:4 (or for that matter out the HD-SDI ports for traditional tape or data recording), and uncompressed RAW recording for up to 4.5K to the as yet unknown REDRAID device (that I know nothing more than the stated intent of having a data recorder that'll do 4.5K @ 60fps....Codex is a NICE HD/2K/4K recorder, Dalsa relies on that, more on those later)

price - the most competitive camera shipping at Red's stated price that I can think of is the Sony F330 XDCAM HD...and that's not even a fair fight between those two (granted, Sony's 2/3" 4:2:2 model due next year will be closer to Red's stated capabilities, but likely pricier....but may ship around the same time. And Sony may drop the F330/F350 prices by the time Red ships in any case. Oh, and the also unreleased Silicon Imaging camera could well be a competitor as well, with a plethora of similar features...also more on that later)

-accessibility - note that the Dalsa Origin, Panavision Genesis, and ARRI D-20 are ONLY available as rentals, not for sale at any price. Some of these rent for about $3K/day on a 3 day week - at that rate, it takes two weeks to pay for your Red body. A further update - a full package from another vendor, with camera, recorder, and lenses runs about $20K a week - enough to buy a Red body and some amount of recording stuff (not all clear yet).

Put all those together, I don't think anybody is going to be able to touch this overall combo, ESPECIALLY for the price.

About the only thing I can think of that folks could bitch about would be the lack of an onboard tape recording option...and I'm ready to be post-tape, data driven anyway. You'll have single/dual HD-SDI for recording to outboard tape if you want to do that....and the F950, Viper and Dalsa don't have ANY provision for onboard tape, the Genesis does, and the D-20 MIGHT in the future.

As I saw other cameras during the week, I kept trying to think back to which ones did how well in what situation. But I kept doubting myself about how good the Red was. But then I kept thinking of that father/daughter shot, and the smooth, smooth, clean, noise/grain/gain free tonality of it, and the incredible sharpness of the image, both on the 2560x1600 Apple monitor for that imagery and the watch footage on a 3840x2400 Viewsonic LCD panel as well. And that restored my faith.

CAVEATS AND POTENTIAL ISSUES:

I wrote a big thing a few months ago (also includes links to all my Red coverage up to that point) about how the biggest challenges I saw for Red were (after the price was set) how soon it would ship and how good the images would be. Based on the sample images and footage I saw, and a reasonable level of projection of what can readily be done to improve upon what I saw, the issue of sensor viability I think is now or will be soon enough put to bed. The Red team was very pleased at the quality of what they were getting, and I almost felt the emotional cheapskate for not being as excited or impressed seeing the same results....but I've trusted they were going to get there, I didn't have serious doubts. They said they'd do it, and they did it, so I wasn't too surprised.

Now that I've seen them accomplish this, and the sensor is working they way they wanted it to, I started thinking about what the next challenges are once the sensor (the core of this camera, and what really makes it significant) could be checked off the Worry List. Other things to be worried about now are:

1.) The codec - I don't know the full story on current development, haven't seen it in action, so that's a potential challenge for them to resolve on the timetable they'll need. A great image off the sensor is meaningless if it gets chewed to dogfood getting recorded. Can they get fast enough compression onboard for the frame sizes and frame rates they are discussing? And if they can generate that much data, can they record it all fast enough with the options they are discussing?

2.) The NLE support - getting an AVI or QT codec done after the core codec code is complete shouldn't be too hard, but getting Apple, Avid, Adobe, etc. (OK forget etc., those three are the VAST majority of the market, esp. for users of this product) on board to hook in under the skin, to do realtime performance, will be another significant challenge. I don't want just realtime playback, I want realtime 3-way color corrector, cross dissolves, etc., same as I'd get for any other HD/2K/similar footage. Plus, the NLE guys are ALWAYS slow to respond - I'd be very, VERY surprised if a new version of FCP or Avid or Premiere shipped with native support for Red next NAB. Look at how long it took to get HDV support after the format was announced and then after cameras shipped (and it isn't complete - hullo, HDV 24p? None yet as of this writing).

Also, since it will be competitive at the high end beyond the indie crowd, figuring out workflows for integration with high end post will be another challenge as well, but it should be entirely possible to convert to high bit depth TIFF or DPX or whatever.

3.) Typical manufacturing issues - can they get decent yields on the CMOS chips? Will there be any bottlenecks on part supplies, etc.? Not saying that I heard any such problems stated, just the same new product manufacturing concerns ANY new product has.

4.) Typical new business logistics - service and support, replacement parts, and a good organization to deal with in case something goes wrong, gets dropped or broken, can't be figured out, etc.

...and others I'm sure that might be more or less significant than those, but that is what popped into my head first.

OTHER STUFF I SAW:

-New form factor options that I'm betting they'll show at IBC - this will put a LOT of concerns to bed. They've been listening, and they're The Right Stuff.

-More granular detail on the camera itself - where stuff is and goes, so folks will have an idea of where the controls are, the ports are, stuff like that.

-More evolved ideas on the accessories....Ooooh! Good stuff, you'll just have to wait.

-I brought up some UI issues, and got great answers back.

-I brought up some workflow issues, and got good answers back - it's a bit early to tell some things.

-and more I can't discuss, but Good Encouraging Stuff.

So what does this imply?

If you were thinking about getting a reservation for one, I'd say Get It Now. During and immediately after IBC, the majority of fence sitters that were waiting to see footage quality will jump on in. If the image quality was the reason you were holding out, I can't imagine anyone seeing what I expect they'll have ready in a month and deciding "Nah. Don't want it."

The single thing that most impressed me today was the smooth, noise free, even tone of the images, with good dynamic range, even though it was in the controlled lighting environment of these tests.

Last I heard, they were around reservation # 340 or so. If they show 4K footageat IBC, and it is as good as I think it will be, I'd bet they take hundreds more reservations at IBC. And once the cameras do start shipping, the last of the fence sitters will probably jump in. So now's your chance - if you were thinking you might want one, put down $1000. It is fully refundable at any time, no questions asked. The company is founded and backed by Jim Jannard, a billionaire. I PROMISE, he is not going to take your money and run for Mexico in a live TV car chase. So it is financially risk free - put your order in, there will be over 300 cameras in front of yours so that you'd get a chance to see and hear about 300 units in the field before yours was ready. And even then, cancel if it didn't feel right. No way to lose money on this deal.

So, if you think my judgement is worth something, and were holding back to wait and see image quality, so far it looks incredibly promising - get in line is my advice.

Now, if you don't need one, don't want one, don't want to be an early adopter, don't want a 1.0 product....great! Then don't buy it. That is an entirely valid and justifiable position. I'm just saying if you were leaning towards wanting one, this would be a good time to act.

And another thing - talking to Jarred from DVX User, he said he's already seeing people on his Red discusion thread snapping up used lenses. The early 90s Nikons are apparently a great low cost way to go (according to Jim Jannard, himself a huge camera fanatic, which is why he started Red). So if you ARE planning on getting a Red, either now or in the future, it wouldn't be a bad idea to start picking up used lenses on the cheap before the rest of the market catches up and snaps up what is out there.

As I said in the first article, "So if you can, come to IBC to "say hello to my leetel frien'"....the dragonslayer. The industry changer. The Point Of Inflection."

You'll note I haven't changed a word of that sentence.

-mike

----
Footnote:

*** So what is my experience to say these things? I made my living doing Photoshop retouching in the early 1990s, when we'd go into ad agencies and they'd tell us that it wasn't possible to do print grade retouching on Macs...and we'd be standing in front of them with the digitally scanned and retouched 4x5 and 8x10 transparencies to prove them wrong (1992ish). I was doing motion capture driven 3d animation composited photorealistically on top of live action plates all done on desktop computers...in 1996 (we won multiple awards for that one). I was doing 2K resolution motion capture driven animation (40-50 toon shaded characters) for B2B marcomm projects in 1998 (pan & scan over the final render). I was doing 300 element visual effect composites in 1080p HD on Macs in 2003, with multiply nested composites, time remapping, greenscreen keys, digital set extension, 2 1/2 D composites, etc.. I spent 6 months developing my own digital uncompressed 1080p 10 bit RGB 4:4:4 HD field recorder, and another 6 months working with powerful desktop based realtime color correction software (Final Touch), using it and working with the developer to iron out the bugs. For over three years I've been focused exclusively on researching how to use digital technology for moviemaking, doing R&D, consulting, and the occassional post job just to make sure my theories aren't crap. I've made my living for the past 15 or so years manipulating professionally shot still and motion images, working on multiple international award winning projects.