Capture workflow, format debate
 

Hey guys,

My dilemma: I need to have a small group of semi tech capable people around the country digitally capturing non commercial, “home made” tapes of live concerts into one standard format that will be archived.

File space is not a concern, and cost is secondary to detail. I need every possible bit of luma and chroma to be captured. (I will need need to learn/teach how to use histograms to make sure there’s no clipping, etc.)

For my purposes, I don’t need to know about any downstream restoration or editing, other teams will be working on that. I just want to know I am giving them (and future generations) the most possible image data to restore.

I’m going around and around in my head about the best way to accomplish my goal. I have had lots of conversations and have read everything I can find here and at videohelp, but I still can’t settle on the best workflow.

My project:

- Need to capture historically valuable concert footage from various vhs/s vhs/hi8/beta original master tapes.

- Accomplished by a small team of people (4-8) across the us. Technical ability solid but varies. Video technical experience limited.

- Captures will be archived “as is”, and then made available to other teams of restorers and editors who may choose to use various tools and programs to enhance/edit/fix the capture. (Totally volunteer/ ad hoc)

- Sound is secondary, as audio was taped separately using pro audio gear. As above, the captured video will be made available and other teams will create synched edits. (Totally volunteer/ ad hoc)

Progress so far:

- have been accumulating good gear with help of LS. Have several vcrs & beta machines, tbc 100, abt-8710, Es-15 (2), hi 8 camera, sign video dual process amp, signvideo detailer.

- have a bunch of capture cards/devices - ati600 usb, blackmagic shuttle, pinnacle usb, Aja kona LHi and others to start testing. We will share and or buy additional as needed.

- We have unlimited storage available, size of archive is not a concern.

Format/workflow considerations: Most of the group have recent windows based machines. We don’t have any windows xp machines. We don’t have any experience with avisynth or other command line based editors.

—-

So, what do we decide to make the “standard” in terms of capture and archive?

The simplest thing to me seems to buy several capture devices that support modern drivers and capture everything in a 10-bit lossless 4:2:2 format.

Overkill? Perhaps. But probably more forgiving in terms of histogram clipping and easier to get up and running on pc equipment already owned.

Downsides? I don’t know....

I realize virtualdub matched with an ati card and huffyuv lossless is the way things are done around here, and if that’s the absolutely best way for us to accomplish the task we will learn to do that as well.

But if there’s an alternative workflow that would meet our specific needs better, I’d love to figure it out now, before I go much further down the rabbit hole.

Thanks in advance for the discussion and experience,

Judd

Bit depth beyond 8-bit is wasted, as VHS was 8-bit at best. Add in generational loss, consumer VCRs/cameras, and 8-bit is well beyond the color of the source.

ATI AIW systems are generally superior, but that requires legacy hardware and OS (Windows XP).

Newer HD hardware is inferior, and SD is always afterthought. That Blackmagic shouldn't even be on the list, don't use it for this. You'll be sorry, if you do.

Essentially, you're doing now what I was doing 15+ years ago. And hardware is last decade is optimized for it.

You have the best hardware acquired, and what needs to come next is test capturing.

You also really need to make sure you're using IPS calibrated monitors, as anything else will lie to you and fool you. Histograms are nice, but even those can be misleading. Sometimes "bad values" are just fleeting values, and you can completely screw up a clip by calibrating against the wrong scene. It takes more than a few graphs to effectively color correct footage, and eyeballs are required.

Keep going down that rabbit holes. You're near the bottom. It's not bottomless, though it can seem to be.

The biggest problem with capturing is hardware and using it.

After capture, that's when restoration fun can begin. Now that rabbit hole can be bottomless, and it is easy to get carried away.

I didn’t buy the Blackmagic. One of the guys had it, along with a bunch of other stuff that’s not going to make the cut.

I’m so confused...

How can 8 bit capture be better than 10 bit capture? I just can’t get my head around this...

Other a>d conversion (audio & photos) benefits greatly from increased bit depth.

There’s absolutely no data that’s being wasted or luma gradients that are being clipped?

I am starting to understand rgb 0-255, with white at 16 and black at 235 to leave room “for error”. There no way for 10 bit to make better use of anything?

It's not about better or worse.

The entire 10bit vs. 8bit issue is like buying grandma a $100k Dodge Charger Hellcat to go to church on Sundays. Sure, it will get her there, but it's massive overkill. She's not going to use what's under the hood. Same for the video: 10bit on VHS is wasted bits. She'd be perfectly fine a sensible Honda/etc, and the VHS tape will be fine processed with 8bit.

Furthermore, both the fancier car and fancier video card yield more effort to use properly and efficiently. It's not necessarily better, as each has side effects that must be taken into consideration. For example, I highly doubt your computer monitor will support more than 8 bits, meaning that must also be replaced.

A similar discussion often happens with 8-bit vs. 16-bit color for photography, aka the JPEG vs. RAW debate. You run into lots of biased info with cherry-picked examples to "prove" the author's point. What always amuses me the BS images that "show" 8-bit vs. 10/16-bit, complete with the posterization ... but on a 8-bit display. Total nonsense. :rolleyes:

Understood.

My thinking is basically along these lines;

1) Long term historical archiving. Will using a 10-but capture create any additional data for future generations who might have improved tools to take advantage of them?

2) Ease of deployment - it’s pretty attractive from my standpoint to be able to buy a bunch of Aja LHi or Osprey cards on eBay, which are still supported and have updated drivers and capture software vs trying to get a bunch of people to build windows xp machines or build them myself and try to support them.

3) Integration with editing/restoration software now and in the future. I barely know a thing about any of these programs - especially the command line programs like avisynth.

As we spoke about, I played around a little bit with Mercalli and Pluraleyes and for my purposes they are a godsend!

Beyond that, I don’t know what’s possible in terms of noise filtering or color correction, or what formats work best with which software, so I’m clueless. But I’m pretty certain it would be best to provide archive files that integrate with the latest versions of the most popular software titles - Davinci, Premiere and Vegas. I just don’t see too many people wanting to learn avisynth.

There is no additional data on VHS. As I've mentioned before, even 6-bit dithered would probably be adequate for the format. But no such card exists, just 8-bit.

Capturing is never fun to setup, regardless of OS.

Avisynth has continually progressed over the years, both in core functions are the filters made by indy devs. It's quite nifty, though it still does not correct anything that ails a video. Just a lot.

Mercalli is a perfect example of something that Avisynth, VirtualDub, Premiere, After Effects, and many others fail at. Very useful tool. As a tip, you may need to split up many scenes and process separately.

- NR best in VirtualDub and Avisythn
- Color NR also best in VirtualDub and Avisynth.
- Some color correction can be done in both VirtualDub (Colormill) and Avisynth, but you may like an NLE like Premiere far better. And then definitely take a look at DaVinci (which was bought by Blackmagic, then made free).

The tbc 100 and avt-8710 both digitize to 8-bit internally, so on the capture end you won't get any more than if they are in the chain.

I didn't even think of that. Nice catch. :salute: