HD For Indies

Another day, another post-midnight note taking session.

short version - LOVED color on HVX200, the JVC looked better than my first impressions indicated, setting up gear is Fraught With Horror when you haven't done proper legwork. We got LUCKY we got everything to work. Woops, lost like three paragraphs here. This'll have to do.

The facts: Zane met me a little after 8am and we planned on building a couple of RAIDs with new drives that had arrived priority Fed Ex the day before, and using a couple of enclosures I brought from home I'd dissassembled and removed the drives from. We sit down with a couple of screwdrivers and 16 little bracket parts that require 4 screws each, and discovered that the new drives were...wrong. ATA not Serial ATA. Sigh. Another priority overnight FedEx wasted. Ugh.

So we bailed on that and set out to get all the capture stations up and rolling smoothly, and started assigning computers to cameras. I'd asked a bunch of folks to show up and help capture, and in the end, I got 5 volunteers:

-Rita Sanders, editor on the doc Slam Planet that debuted at SXSW that I did some post work on
-Neil Halloran, local filmmaker and client of mine, who was INCREDIBLY gracious and generous to let me borrow not just his person for most of Saturday, but also his COMPLETE G5 based editing system, including his 19" JVC studio HD monitor.
-Lary Cotten, who deep geeks deeper than anyone I know (and I mean that in the best way), CTO of OpenLabs, makers of bogglingly phenomenal pro audio gear that mashes a computer with touchscreen into an musical keyboard
-Craig Negoescu, co-founder of OpenLabs, and like Lary and myself, an ex-frogdesign employee (he was one of the handful who started the Austin new media office where I worked)
-Jenn White, a friend and local DoP who ended up running the Varicam a good portion of the day while I captured the output from it

The plan was to capture the uncompressed output of the cameras, either via HD-SDI or through a converter attached to the HD analog component outputs of the cameras. Either AJA HD10A converters were used, or the analog HD inputs on my BlackMagic Multibridge Extreme or through an AJA HD10A converter (HD analog component to HD-SDI) to an HD-SDI capture card, either BMD or AJA.

Why? Becaue ZERO compression artifacts since all these sources ARE before compression, beause you get full raster (1080i HDV is only 1440 pixels wide, not the 1920 most assume, for instance), and is 10, not 8 bits of color depth (more subtle color choices available - can your camera reproduce them?). Of course, uncompressed capture is only viable under limited circumstances, such as on a studio set, greenscreen set, or BIG set that already had a video village type of need.

Please keep in mind these are the random floaty brain bits of an exhausted guy after another 14-15 hour day on set. These are preliminary notes and observations, and don't have the qualifying wrapper of serious analystical comparisons and analysis - these are my gut reactions at this late hour.

Duly disclaimered, read on:

NOTES ON SHOOT;

Adam Wilt, Camera King on this shoot, who has a better trained and and more experience in these things than I, commented on some footage we viewed (and I missed a bunch, but here is some):

-ADAM on 18 vs 35 mbit on the F350 - "18 looked like a bit more degradation when sitting 2 feet away from screen, 18 looks better than I would have expected" - (I'm curious about 18 here vs. JVC 19 mbit?). In motion they both looked good when viewed from 15 feet away, but looking at stills, BIIIIIIIIIG difference. (I thought 25mbit HDV from XL H1 held up surprisingly well against 35mbit F350. But F350 is a COMPLICATED camera, and it just started shipping a few weeks ago, and none of our crew had spent huge gobs of time with it).

-F350 harder to focus, something strange about pulling focus with this - a lot closer to the 1/3" focus than 2/3" in the focusing ability - is it viewfinder or depth of field? Viewfinder was hard to use - contrast was all the way down in viewfinder, made it tougher to focus

The thing that panasonic does better is naturalistic color and good gamma, and skin tones (going into overexposure it doesn't blow straight out like the Sony (even F350), skin tones going to overexposure is NICE

----------

-F350 reportedly has only 4:2:0 amount of information in the 4:2:2 stream coming out of the HD-SDI, when asked to explain at 10pm after an 8am crew call, I blurted "It's like saving a GIF as a TIFF." to which I'd now add "...and then smiling like you did a great thing."

-the question that popped into my head - after a CURSORY examination of some of the footage (I ended up running the Varicam capture all day today, so I thought all the footage looked great since that was what was in front of me), I went and looked at some Z1U footage. Comparatively, eww. But hey, whaddaya expect when one camera costs 15-20 times more than the other?

-I then thought about this - esp. on the cameras that don't produce color well at first blush (and based on EXTREMELY PRELIMINARY GLANCING WITHOUT EXTENDED TESTING, EXTENDED "I RESERVE THE RIGHT TO CHANGE MY MIND DISCLAIMER DISCLAIMER), I'd put category of in the "not so great/not so pleasing color rendition" the GY-HD100U and the Z1U.

ANYWAY, for the cameras that don't do nice pleasing color with low noise, even capturing uncompressed (and I gotta look at the uncompressed footage), my gut says it is of lesser value to go to the trouble of capturing uncompressed (see about troubles below) from these cameras, better off going compressed (or uncompressed) from a better camera. that's my gut vibe now, I'll have to see how it turns out later. Because...

...recording uncompressed to disk is fairly complicated. You can say "Well hell, you just buy a G5, a Sonnet E4P card, and 4 or more drive RAID, a BMD or AJA card and away you go!" To which I say, "Kinda sorta." On they fly, later in the day, we decided to shoot some tethered (record uncompressed to disk via Mac with HD-SDI capture card) footage from outside the building. Problem - it's about 175 foot cable run from where the Macs were set up to where we wanted to shoot. We hooked up two HD-SDI cables and used a barrel adaptor to link'em together. We took the BlackMagic Multibridge Extreme box (remember all the brains 'n guts are inthe box, the card is barely anything, just a connector) outside, and run that long HD-SDI back to another capture Mac in the studio. Hooked it up, fired it up, and....nothing. Changed to an HD-SDI camera (not component analog) and....something, but jibberish. I then realized the problem could lie in:
1.) the long cable run, esp. with a barrel adaptor
2.) something misconfigured in the Multibridge
3.) If something were misconfigured in the Multibridge, I'd have to troubleshoot it, and possibly take my laptop out and hook it up via USB to configure it until I got it working...with the G5 200 feet away.
4.) Or it could be a problem with the capturing Mac

...so I bailed on tethered, uncompressed recording, since in addition to the trouble I was having at the moment, I'd be having to switch the HD-SDI cable from capture Mac to capture Mac to put in the right capture bins, or straighten it out later, or possibly troubleshoot any new difficulties.

Now, on a "normal" (if there is such a scenario) shoot wanting to shoot uncompressed, there would largely be one chunk of technical troubleshooting and then you'd be good until something went wrong or broke.

So I bailed on it. Tethered capture to uncompressed is a hassle - gotta have all that extra gear out on set, it's noisy, sucks a ton of power, requires another person to operate, whine whine bitch moan and YEAH, it is a pain. So think twice, nay nine times before going to the trouble.

And practive BEFOREHAND. I could have easily spent 2-5 more days prepping, testing, etc. Only becauase I've been playing with this stuff for 2+ years is it even sane for me to have tried this. To have only failed on a 1080i50 vs 1080i60 issue for a few shots, and had to cancel one test, ain't bad at all I figure, and I was LUCKY. There were many, many possible points of failure in this ordeal today.

FYI, unlike last year, switching frame rates on Varicam didn't crash Macs attached to it at the time. Either an AJA vs. BMD thing (BMD trouble last year), or

-when I lamented to Nate that I'd had a few problems today, Nate said don't sweat it, we'd done better than he'd expected when he heard what we were trying to do. Boy, that helped!

-along those lines, I hoserated myself by NOT Sticking To The Plan. I'd sat down with a spreadsheet and carefully calculated out which Macs, with which capture cards, with which RAIDs, with which RAID cards, would be cobbled together and record from which camera. What size computer monitor, whether it could run out a computer LCD with the video image on it, the RAID speed, the card's capture capabilities, all factored in. It was very specific and limiting. At one point, I'd plotted out that the Z1U should go to the Quad G5 with the Multibridge Extreme. I figured heck with it and used a different Mac running through an AJA HD10A converter - big mistake, the HD10A doesn't do 50i, so I had to have that Mac sit out of some capturing.

GEAR: Oh, MAN, I had a ton of different stuff! A sample listing of gear used:

-Quad G5 2.5 GHz
-Dual 2.3 GHz PCI-X G5
-Dual 2.0 GHz PCI-X G5 (multiple)
-Dual 2.5 GHz PCI-X G5 (multiple)
-AJA Kona2 (multiple)
-AJA Kona3 (and why does the Kona LH have analog HD input and this doesn't?)
-AJA HDP
-AJA HD10A (LOVE these! Had two on set)
-BMD Multibridge Extreme card/breakout box
-BMD DeckLink HD Pro Single Link PCI-X
-BMD DeckLink HD Pro Dual Link PCI-X
-BMD DeckLink HD Pro Dual Link PCIe
-MacGurus 5 bay eSATA port multiplying enclosure with five Seagate NL35 400GB drives
-MacGurus 5 bay eSATA port multiplying enclosure with five Hitachi 400 GB drives
-MacGurus Burly Box enclosures (two 4 bay enclosures) with 8 unknown drives (Seagate 7200.8's if he got what I recommended)
-two arrays comprised of Firmtek Seritek 2 bay enclosures with Seagate 7200.8 drives (set up two of'em, for 4 drive array, partitioned at 560GB to be sure it's fast enough)
-LaCie Biggest S2S 5 bay enclosure (went unused, didn't need after all)
-Trans International Mini-G 4 bay SATA enclosure (one I did a review of)
-huge, dim, blurry, 80-90 pound Power Computer PowerTron 24" CRTs that I never, ever, EVER want to lift again
-Apple 15 or 17" LCD, old plastic style (went unused)
-Apple 23" LCDs (multiple)
-Dell 2405 24" LCDs (multiple)
-Apple 30" LCD (ooooooooooh, I want one - it's so big you could crawl into it through the screen.)
-Sonnet Tempo 4+4 SATA cards
-Sonnet Tempo 8 port eSATA card
-Soonet E4P cards (in both Quad G5s)
-JVC 19" broadcast HD CRTs
-Pansonic 17 or 18" broadcast CRTs with HD-SDI and component inputs, and built in waveform - LOVED these, but Jordan said they cost $-$5K
-then of course the cameras:
JVC GY-HD100U
Sony HVR-Z1U
Panasonic HVX200
Panasonic Varicam
Canon XL H1
Sony F350 (the new XDCAM HD)

HD10A thoughts -
GOOD: rockingly small, lightweight, easy to use, has HD-SDI passthrough (smart!)
BAD: no 50i support (I got bit by that), requires use of dip switches on bottom (doesn't auto-detect in some cases), a touch pricey these days

Multibridge Extreme - a bit complicated since it does so much. Took me about 15 minutes of troubleshooting, even after working extensively with BMD products for a couple of years, to figure out that selecting analog inputs had to be done in the Preferences Pane, not in FCP (or if it is in FCP, I didn't know where to look. Of COURSE, I didn't read the instructions, why do you ask?)

-AJA HDP, BlackMagic Multibridge Extreme's DVI output, and BMD's HDLink - great for checking focus on set, not so great for interlaced or true 24p footage, not sufficient for very fine/finicky color correction work - poor black levels, impossible to calibrate by standard video means, JVC 17" CRT too close to price competitive - just get the real thing rather than a simulator for COLOR, but for DETAIL, they can't be beat for color accuracy, fidelity, black levels, etc.

Is it worth capturing lower level cameras uncompressed? I'm kinda thinking not, just seeing their compressed output. Even without compression artifacts, there are still artifacts.

The HVX200, which was ranking low in resolution tests, and was described as noisier than many if not all others (I'll have to prove that, may well be wrong), looked GREAT as raw footage. Colors were just a bit off from other cameras. I'd think about capturing his uncmopressed. Or at least, the Z1U I probalby wouldn't, dunno on the GY-HD100U. XL H1 looks pretty darned good with HDV footage.

BUT is Varicam worth capturing uncompressed, ever? In theory the answer is a clar yes - you gain:

10 vs 8 bits/pixel color depth
none vs. moderate compression
1280 vs.. 960 recording pixels

Also, specs aren't everything. While the HVX200 kept sinking in the ratings in the objective tests (Chroma DuMonde, etc.), when I saw footage I LOVED it - colors look vivid and real in CineLink-D. Now, liking the color reproduction of a camera CAN be a subjetive thing. But nobody chimed in saying they liked the HD100U, Z1U, or XL H1 color reproduction better. I likes it.

But I'll have to refer to footage and guage for myself.

OK, I just hit the point of about to fall asleep on keyboard.

this should keep you busy for awhile.

No pics tonight, too much work to do now.

All this is totally random, incompete, and inconclusive.

Thunk.

Labels: acquisition, Production, TX HD Shootout