Dynamic range in analog sources (e.g. VHS/SVHS) and digital formats

[swiego]

I've been thinking a lot about proc amps and whether I should use them when my goal is simply a "digital negative." I don't have much experience here compared to other fields, but I know that the concept of a "proc amp" device isn't very interesting to me when I'm scanning film negatives because with a good scanner (something like my Nikon 9000ED) I can pretty much capture the full dynamic range that's present in the original negative. Moreover, since I am capturing to formats with high color bit depth (say 16 bits/channel) I am capturing to a format that will preserve that dynamic range. Therefore for digital archival purposes, I don't worry much about tweaking contrast, brightness, color balance, etc prior to digital encoding.

Ditto these days to my digital photography..... exposure still matters, but by shooting RAW I capture enough dynamic range in the source to be able to deal with a lot of blown highlights and the like, should my exposure be off.

SO... I've been wondering about video capture. What are the parallels here? How is dynamic range measured for a typical analog tape (be it 8mm or VHS or BetaSP) and what are the typical outer limits? How do these compare to the the ability of the ADCs in typical capture cards to "see" this dynamic range and to the file formats we use (e.g. RGB32) to encode it post-capture? I get the rough impression that the answers are "not very good" across the board, which is why proc amps are of such value in this place but largely unnecessary in every other archival project I've done (film, LP, flat art). However I don't know enough about this stuff to really say that for sure...

Thoughts? Are there capture solutions out there that capture enough range such that I needn't worry about having to tweak the source prior to capture? I'd love one even if it meant a lot more storage required. By having to adjust the source with my eyes prior to capture, I'm stuck dealing with my poor eyesight and color blindness, the limits of my computer moniter's ability to display accurate color and range, and more, when all I want is the digital negative

Note: I am not referring to TBC-type intermediaries... only to contrast/brightness/color/hue/etc/etc type functions that proc amps have.

[Cornucopia]

I would say that:

1, Having extra/excess DynRange that you can waste isn't an excuse to not set up a shot, printing, scan, take, etc. You can't really get it back once it's lost.

2. You're right that Film (still or motion) is still usually much better with latitude than video. There are some HD/DCI cameras that can capture log, cineon, 12bit4:4:4 and the like. And those can get quite close. Check out what you can do with the Panny DVX100 mod called "Andromeda" http://www.reel-stream.com/

3. Otherwise, with standard digital cameras (still or motion) that compress, color subsample, colorspace convert and/or apply color bitdepth reduction is going to disappoint one who's used to larger latitude.

4. Because of their older technology (lenses, sensors, processing, etc) and non-bypassable packaging for video standards (NTSC, PAL) which use similar/equivalent color subsampling, color conversion and all (based on bandwidth limiting), ANALOG video cameras are probably going to be MORE of a disappointment than good, new professional or prosumer DIGITAL cameras.

5. With Stills you can bracket exposures and then combine for a beautiful HDR master. Can't really do that with something in motion (unless you're talking about automated, reproduceable, computer controlled machine motion capture of a still).

Can't give you "Numbers" right now, but I'll get back to you. Just a guess though--If you're not wasting any DynRange in your exposure and scan, you could possibly have full 36-48bit color saved in a file. How many video formats routinely give you 48bit color?
(***Note: 32bit color usually reserves 8 of those bits for the Alpha/Transparency channel, so you're really talking about 24bit color or 8bit per color primary)

Scott

[lordsmurf]

There is a lot of quality in VHS, for example, that only better S-VHS players can extract. Otherwise you get "Polaroid quality" (instant snapshots instead of prints/scans from film negatives).

Proc amps are not so much about extracting inherent quality as much as it is about fixing flaws (usually flaws that supercede the media itself, but go back to the creation steps involved in how it got onto that older media in the first place).

There is something to be said of uncompressed digital video, but few people ever actually see it, and it tends to be wasted if the source is not super-clean itself. An uncompressed capture of a VHS tape won't be very spectacular if the VHS was grainy and full of chroma noise. Filtering it an applying compression (which can hide flaws, such as the temporal nature of MPEG-2 DVD-Video) may look more appealing anyway.

Stuff like this has to be taken on a case-by-case basis. There is no single answer and there never will be.

[sphinx99]

I guess what I'm noticing is that our ability to correct video defects is improving at an ever-accelerating rate. It's amazing the kind of noise reduction available today (albiet at 1-2fps sometimes!) that wouldn't have been imaginable five years ago. Storage capacity has made similar leaps. For someone like me, then, with lots of "stuff" to digitize and no immediate need to convert it to a finished product, I'm inclined to simply capture the highest possible digital copy of the analog original and save the restoration for the digital domain on my 6 GHz 128-core laptop ten years from now. I just don't have confidence that huffyuv, MPEG2, WMV1, etc., have the latitude to carry the amount of information needed to do so, and even if they did I don't have the confidence that a mediocre capture card can sample the original analog signal with that level of precision. (I can sample audio all day long at 24-bits 192kHz with a $200 sound card... how come I can't even come close to that kind of fidelity with a video capture card? Or can I, with sufficient investment?)

[Cornucopia]

Well, you could pay $$$ and get a cap card that does 4:4:4, RGB or YUV, 8bpc or better. But the limiting factor will already have been the originating tape format, so why bother unless your tape format is at the top tier (in which case you don't need to be asking this question, because all peripherals and setup costs will be way high (in the $150k+ range). Video is ALWAYS more complicated and expensive and bandwidth/storage-hogging than audio.

BTW, a $200 audio card is fine for much consumer use, but often has CRAP S/N ratio, random non-linearities in quantization level and samplerate ("judder"). Hiqh quality A/D converters are sometimes ~$1-5k (and are external to a PC, to alleviate EMI/RFI problems).

Scott