Vet new VHS-MP4 workflow; Hybrid stumping me?
Hey, gang.
I've been ramping up my VHS capture workflow in anticipation of hundreds of new captures in the coming weeks and months. I went on a bit of a buying spree after years of using consumer VCRs, a TBC and capture solutions through the years ranging from the AIW9800 to the Hauppauge HD-PVR 1212. My current input stack:
Now, my first venture into capturing uncompressed video in a long, long time. I went with Virtualdub2, got all the drivers geared up, and managed to capture a tape to HuffYUV at 720x480. It's from the late 80s but appears to be first- or second-generation, and captured beautifully. I watched the histogram and didn't see clipping, but there was definitely signal being recorded below 16 and above 235, so I expected to have to adjust that from PC to TV levels later. After reading lordsmurf's praise of Hybrid, I grabbed that, and was excited to finally have a full front-end to Avisynth stuff, which has always been an area that I've been reluctant to dive into. My goal, as it seems to be with some others who've posted here recently, is to output a x264 progressive ~60fps file for viewing online, on phones, and so on. I'm well aware of the "keep interlaced as interlaced" sentiment in the community, and I intend to make a separate interlaced file for archival and future purposes. Right now I'm trying to nail down the easy peasy delivery format. And Hybrid, bless its ffmpeg heart, is making me crazy. I'm on my eighth iteration of trying to get what I suspect I want, and please tell me if parts of this are bonkers:
Here's where I'm getting stuck. Hybrid is analyzing my capture as progressive. I can see with my eyes that it's interlaced. So, I've tried assorted checkbox and value change adjustments to tell Hybrid to ignore its better judgment and treat this as interlaced. I've tried top field first and bottom field first - neither yields acceptable video. Perhaps I'm misreading what the many options do, but I've done what my brain thinks SHOULD work, and it's not, so I'm asking for help. Ordinarily I'd handle most of this in Vegas, as I've used that NLE for a decade or more. But I've not been able to load the HuffYUV files into Vegas, despite the usual suspects (VLC, MPCHC, Virtualdub) having no problem with them. So here I am. Thoughts, gang? |
Replying as I read...
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
Now, my first venture into capturing uncompressed video in a long, long time. Quote:
Sometimes, rarely, 1.10.x or the FilterMod (pre-2 version). VirtualDub2 is terrible at capturing. Quote:
Quote:
Quote:
Quote:
29.97i is 720x240 alternating data. To create 59.94p at 720x480, new data must be created. 29.97p does discard some motion data, it's realistically about 25%. But while 59.94 retains all motion data, artifacts are created, and the file bloats. There's no correct answer, and must be case-by-case. 59.94 can having nasty side effects on home movies. (The same is true of 25/50 for PAL frame rates) Quote:
Do not use the QTGMC slower preset. Use faster. Slower blurs, takes needless longer to process. Quote:
Quote:
Quote:
Quote:
Quote:
Filtering tab > Deinterlace > Override* > TFF * Currently says "overwrite", but that's no correct, need to submit a bugfix to selur here. Quote:
Quote:
|
Thanks for the reply!
Quote:
Quote:
Quote:
Quote:
Quote:
I'm still kind of jumpy when it comes to capturing video. When I started years ago with the AIW, we had to get RAIDs set up to accommodate the data flow, turn off everything in Windows except for the capture program, make sure the screen saver was disabled, and avoid even wiggling the mouse. I'm scared right now, writing this reply on the same machine that's capturing. Quote:
Quote:
The resizing. Hmm. I'd LIKE to deliver 640x480 but haven't been able to pin down which combination of settings in Hybrid actually gives me a correct aspect ratio on output. Perhaps I need to override the default PAR/DAR settings, but whatever resize method I've tried in the final step has been off enough to be distracting. So, for the current batch, I'm trimming to 704x480 on the way to final compression. Quote:
Other loose ends ... I don't remember when I picked up the TBC-1000, but it had to have been in the 1990s. It works just great, all but one of the outputs IIRC. I've always had a hard time trusting a black box with no knobs, buttons or switches. With the leaps in technology we've seen in the past 20 years, and the current ease of crafting custom parts and projects, I'd be delighted if someone somewhere would design a basic TBC and proc amp with a little screen in it so we could make wfm/histogram/vectorscope adjustments. Or get ballsy and put a DSP chip in there with hardware versions of QTGMC etc. Yeah, it'd be a very niche thing and would be a short production run, only of interest to a few. But if it were good, wouldn't you buy a case of them? |
VirtualDub2 is very different capture interface, approaches drivers differently. It's more glitchy, more unstable, causes dropped frames at times. It can sometimes make cards work, that had previously NOT worked in VirtualDub 1.9/10/FM. But it's not overall best. 1.9.x is best, then FM (pre-2), and 2 is last. Even 1.5 can be more stable that 2 at times. 2 was not intended for capture, but better input/output (codecs). The dev did work on capture years later, but it's still beta quality.
You can reorder Hybrid filters. I often just write an Avisynth script manually, and load the .avs in Hybrid. NeatVideo is a hatchet job, Avisynth is a scalpel. It was good 15 years ago, not so much now. The problem with making TBCs is an utter lack of chips to do it. That's why Cypress and DataVideo exited the TBC markets, no more supply of chips, and unwilling to do new R&D with the sparse chips available. |
Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.