There are 3 scripts in post #32. The first one shown uses the same QTGMC that your friend declares as his "all-encompassing" VHS cleaner, plus a little something for extra cleaning and some help against the color banding that's in many of the shots, and it's the one-step kind of script you requested. Because it produces progressive video, there should be less aliasing in the output than in the original -- although, let's face it, some of those shots are really unusable, no matter what you do to them.
You have source tapes that were crudely assembled from a variety of other sources. There is no consistency of production values from scene to scene; literally, the video was just patched, never really processed. You have some shots that are already progressive, some that were cropped and resized while still interlaced, some that were telecined and incorrectly deinterlaced or field-blend deinterlaced. The all-in-one script which you requested can't solve all the problems in each shot, because no two shots have the same problems. One can't even find two consecutive shots that have similar levels or correct color balance. Some shots look like direct captures, others look like 2nd and 3rd generation tape dupes, some had crude denoisers applied while still interlaced, some were deinterlaced and alternate frames were discarded (and thus are progressive but play at half the original temporal resolution)....and on and on. So the first scripts you saw preceding post #32 were customized for each shot, then the shots had to be reassembled into the main video. As I expected, you disliked that idea -- no one would blame you for that, as it's a pain in the neck that most newcomers don't see coming until they eventually see why pros and advanced users work with individual segments that way, when needed.
If you still see aliasing in some shots after using QTGMC, then the defect existed in the original video. I just now looked at one shot in your
VG13samp.mov sample posted July 20. The shot is interlaced (and has the wrong IRE, bad gamma levels, and invalid luma and chroma values), and displays some easily visible angular aliasing and strong interlace combing. After using SangNom on the interlaced original, most of the aliasing is gone. After QTGMC, the progressive version shows no aliasing. After some color work and a little denoise, it was reinterlaced and a small part of the aliasing returned -- indicating poor field or chroma alignment in the original, possibly from an inferior encoder or cheap capture card or incorrect colorspace conversion. After applying the santiag() light-duty plugin to the new reinterlace, all aliasing was gone.
Unfortunately one of the shots that precedes the one discussed above can't use those filters: the preceding shot is already progressive, with seriously bad line twitter due to a really no-good deinterlace method. Deinterlacing that shot makes it worse. One could minimize the problems by processing separately with QTGMC in progressive/repair mode and using upsample/downsample routines with specific resizing plugins to help heal the bad lines, all at the cost of some detail. Then that shot would have to be patched back into the others, after the others have had their own separate processing.
I'm pretty certain that you don't want to follow that route. If you still find your other party's one-shot cure-all script to produce satisfactory results, then rest assured that you're using the same basic filter and the same frame discarding method that he used to get progressive video. There is no anti-aliasing mentioned in the third party script you showed us. Either he's putting you on, or he's working with video sources that are quantum leaps cleaner than those you have at hand.
Quote:
Originally Posted by pinheadlarry
also i'm still wondering about the file for streaming vs dvd
|
I guess it should be explained again.
The script with QTGMC posted as the top script in post #32 produces double-frame-rate, deinterlaced, progressive video. Save it in its own file. Call this step Script #1, and call the file Output #1.
The short script in the middle (Script #2) opens that Output #1 file, discards every Odd frame (just as you friend's script did, in exactly the same way), and produced a progressive original-frame-rate video suitable for web streaming. Some of the shots won't have such smooth movement, because you've discarded frames from shots that were already progressive before you used QTGMC, but you friend's videos will have the same problem. Call this file ouput #2.
Script #3 at the bottom of post #32 re-interlaces the Output #1 file from Script #1 and produces an original-frame-rate video for DVD. If you want to cheat a little, you can take the progressive video from the middle step (progressive Output #2), above, and use it for web streaming as described and, by telling a little white lie to your encoder, encode it as if it were interlaced. That's the way many of the shots in your video samples were encoded.
Do not run the #3 re-interlace script on Output #2 file. The #2 file cannot be reintelaced (and even if you did, the #2 file would run at 14.98 fps).
Quote:
Originally Posted by pinheadlarry
The original video is night and day compared to the final output with those jagged lines.
|
if the original looks better, then forget the scripts and run it through whatever NLE you have available.