Capturing video: How does this plan look?
 

I am getting decent captures now with virtualdub and huffyuv.

Some of my captures are tapes with multiple recordings. Here is my plan... please tell me if you would do it differently.

1. Trim and edit captures into individual recordings using Virtualdub. (is a different editor better for doing simple edits without re-encoding?)

2. Apply filters as needed via AviSynth and VirtualDub for fixing quality issues.

3. Deinterlace with QTGMC one copy intended for youtube upload.

4. With Handbrake, re-encode deinterlaced copy with H.264 to reasonable file size, and re-encode interlaced copy to mpeg for DVD authoring.

Lossles capture and lossless processing are the right idea for your source, but why all the complications concerni9ng interlace? For web mounting you need a deinterlaced encode. Obviuously you aren't going o mount everything on the web, so you deinterlace only what you ned for the web.

The rest of your interlaced videos should remain interlaced. You'll be surprised to learn that DVD and standard-def BluRay are interlaced. 1920x1080 and 1440x1080 BluRay are interlaced formats unless they're encoded at slower film-source speeds; DVD is encoded as MPEG, and BluRay is encoded as either MPEG, h.264, or VC-1. Just because a video is encoded as h.264 is no reason to deinterlace.

Deinterlacing has a quality cost, even with QTGMC. Always. At times it can't be avoided, such as for web mounting (whose quality isn't all that great to begin with) and sometimes for certain kinds of denoising or other post-processing. I see you also referred to "re-encode the deinterlaced copy". No, please. Never re-encode lossy encoded video if you still have the lossless original. No sense shooting yourself in the foot after all that work.

Another consideration is your source material. If your tapes are Hollywood movies or other film-based material such as TV shows, you can't deinterlace telecined material without serious damage. Those taps require inverse telecine (IVTC), not QTGMC, for progressive ouput.

I don';t understand this obsession with deinterlacing.

I just thought I had to deinterlace prior to upload to Vimeo, and that I should deinterlace before any other kind of processing or encoding into an uploadable file. Or is there a way to do that all in one step? Or am I just understanding that incorrectly?

I've spent several hours at this point reading other threads on this board, so I understand what you're saying about leaving videos interlaced if my goal is putting them on DVD or otherwise playing them over a television as a digital file. In fact, I will only be making DVDs for a small percentage of the videos, and my primary use for files will be playing digital files via WDTV or similar player on TV. (Or on Vimeo over computer.)

I guess one thing I'm not certain about is the best way to cut apart a capture before doing any processing.

For example if a two-hour tape includes a birthday party, a t-ball game and a TV show taped over the air, I would want to separate that into three different files that would each be processed differently. What's the most efficient way to do that?

Or again if I'm saying things that are nuts please let me know!

WDTV and other external players can play interlaced video, I don't know why people think they can't. They can also play telecined video as well as progressive. Obviously you have to deinterlace for web mounting. Otherwise there's nothing to gain by deinterlacing, regardless of the encoding codec.

I regularly split captures into smaller segments, sometimes very small ones since VHS requires different filters and processing on a constant basis, sometimes from scene to scene. You can make an edit in VirtualDub by opening an AVI capture, making your start and end cuts, and saving the video using "Video" -> "direct stream copy" to prevent a colorspace conversion. This saves your edit in its original losslessly compressed colorspace. You can do this in other editors as well, but you'll have to manually set output colorspace and compression in most NLE's.

For many post-process operations, deinterlacing isn't required. You can crop off bottom-border head switching noise and reinstate black pixel borders without deinteracing. Some denoising operations require deinterlaced video, some can be processed using SeparateFields() alone. You mention TV shows, many of which are produced on film, and as I said earlier these are usually hard-coded telecine, not interlace. Deinterlacing telecined video results in many problems, not to mention duplicated frames. Submit an unfiltered and unencoded capture sample if you need help on that. If you don't know how to make and post an unaltered short sample of just several seconds, just ask.

What I do for web targeted videos is to process the video normally with the original interlaced structure in mind. Those segments that require web mounting are deinterlaced an encoded separately.

I suppose you're aware that deinterlacing video will double the frame rate and double the frame count. For most final formats these are reinterlaced after processing. For telecined video, telecine (pulldown) is replaced during encoding. This can be done whether you use h.264 or not. There is no requirement anywhere in the world that says h.264 always has to be de-interlaced. Perhaps you're conditioned to think that way because of the internet.

You don't deinterlace every TV show that you watch over cable and every DVD or BluRay that you own, do you? Does your equipment have trouble playing those sources? Why do all that work when players deinterlace for you during play?
:wink2:

THANK YOU! This is just the tip i was looking for.

Your point is well taken on don't deinterlace unless I need to... I guess the only things I'm unclear about is in those cases, let's say i'm about to put a video on Vimeo -- should I shrink it down to H.264 15mbps then deinterlace then upload? Or do the deinterlacing before that?


And if I do processing that requires deinterlacing, are you saying I should reinterlace at the end?

Any change in the imaging or frame structure must be done before encoding, and this includes deinterlaing and re-interlacing. Encoding is never an intermediate step. it's always dead last.

Pardoin, I didn't answer completely in my last post. I powered down early because of a thunderstorm and then the power went off for a while, so I couldn['t get back on to edit my last post.

If you've deintelacd in normal processing, leave the segment deinterlaced and use it to encode your web copy. Then take the deintelaced, unencoded version, re-interlace it, and encode for your DVD or system playback.

If you change your mind later and need something for web posting, remember that you have your archived original that you can always turn to for making a web segment. Even if you don't have the archive, you can get a slightly lower quality web copy by using your interlaced high-bitrate DVD or system play copy. You can decode, de-interlace and carefully re-encode an h.264 web version for posting. This will be slightly lower quality than if you had made it from a lossless original, but you can do all this carefully and still get good results for the web, better than the sloppy jobs you usually see on the 'net, and web standards are lower to begin with -- and by the time Youtube gets finished screwing it up, no one will know the difference.

Just checking if it's normal for this process to still take a looooong time (rendering at 30-40fps so almost realtime). I was expecting a fast saving process like in Videoredo when trimming an mpeg stream. But of course I very easily could have a bad setting saved.

Something wrong there. Direct stream copy is not a recompress, it's a simple copy. I just used the above method to edit a Lagarith AVI and copy the edit to a new file. The new file was 430mb and copied at 165 fps with the same colorspace.

If you opened that edit with an Avisynth script, the video was decompressed and copied as uncompressed. Open the fie directly in Virtualdub without an avs script, make your edit, and use "direct stream copy" to create the new file. Also, if you're making the new copy onto an external drive or USB drive, it will be slower writing.

Alas, VirtualDub had reverted back to full processing mode and I didn't catch it.

Opening file directly, making edit, selecting direct stream copy, then save as AVI, and the Video Rendering rate is averaging 108 fps. I put Processing Threat Priority at Highest, and put speed limit as fast as allows.

(also deselected Show Input Video, Show Output Video, and Show Decompressed Video)

Speed depends on your system, so it varies by machine.

Selecting the differences in the output pane doesn't do anything in the output pane with direct stream copy.