Critique and Script Advice for Captured Lagarith Footage
 

Hi all! I hope everyone is doing well and staying safe. I am reaching out for some advice on overall critique on my capture and then on advice on what scripts in Virtualdub/Avisynth might work best for these samples.

The samples come from footage from some old home VHS tapes that were recorded with a JVC Camcorder.

My current setup looks like this: JVC S9911U (S-video) → AVT-8710 (S-Video) ---> Pinnacle 700-USB

I've included 3 samples, the first two are from the same tape and the 3rd is from a different tape.

I have been experimenting with some scripts and I have been able to get QTGMC to work but I would like your advice and critique based on these samples as I feel I am in over my head at this point. :smack:

Thanks so much! I truly appreciate it.

-X

Hi all,

I attempted to write a script to deinterlace and clean up clip3. If you have any advice or critiques, it would be greatly appreciated. I've included the script below as well as the "restored" clip3. At first glance, I think I can remove Coloy YUV if the gain is simply zero and from what I read sharpen might not do much.

Thank you for any help :)

I downloaded your videos yesterday, but have not had a chance to dive into it much as I am a bit busy. But I have a few quick notes:

--If I recall, your clips have an interleaved color format. ColorYUV was giving me negative values. This problem resolves when you convert to a planar color space. Keep this mind as you work with different filters, as some will only work in certain color spaces.

--ColorYUV with a gain_y = 0 will do nothing, obviously, because it is equal to 0. If you increase or decrease this you should see a change. Whether you need ColorYUV will depend on how the luma is distributed in your video, for which you should use Histogram("Levels") to check. Likewise for your Tweak function, the default value for cont is 1.00, so writing it out is not necessary unless you plan to change it.

--Your QTGMC has an FPSDivisor = 2. This will cut your frame rates in half. If you are trying to save space, you may want this, but otherwise people usually keep all the frames because the video will look smoother. Otherwise you are throwing out half of the data.

--LimitedSharpenFaster usually works better for sharpening, if I recall. You can read more about it here:
http://avisynth.nl/index.php/LimitedSharpen

I do not think I will be able to help for a few weeks as I am using a different computer, but if you are lucky perhaps our resident expert Sanlyn will be able to offer his wisdom :wink2:

Thanks so much for the tips!

In regards to ColorYUV, that is what I figured, I will run the histogram in VirtualDub and apply what I have learned from other posts that Sanlyn has made in regards to this.

In regards to FPSDivisor, I was under the impression that I would want to keep the video at its native frame rate, so this is why I have included it, to keep it at around 29 frames. Should I not do this and let it double? I'll play around with it and see how it looks.

LimitedSharpenFaster, I will research this tonight and apply it.

I look forward to if and when you're able to dig into the clips in the future, if not, you've already put me on a better path, thanks!

As a point of reference, this assertion has always slightly irritated me.

It's not half.
Example: 4:2:2 to 4:1:1 is half. This is not.

Deinterlacers reconstruct fields into frame. Some degree of data is interpolated. It's not 200%, double the data (fields are 50% size of frames). The inverse, the drop-frame method to retain 25/29.97fps, is about the same loss as the 50/59.94fps overall interlace gain. It's not 0, but it's equally not 50%/half. There are many times where the interpolated motion is minimal to identical between frames, so a 50% frame reduction doesn't equal a 50% motion reduction.

Make sense?

25% is a much better guestimate, but it's heavily dependent on content.

All this noted, these days I do err more towards 59.94 NTSC / 50 PAL more than not. We have 16tb drives now, after all. About 5 years ago, I was more in favor of drop-frame. Much of it has to do with streaming specs and overhead at the time, and in years prior. I no longer do studio streaming work, mostly just for myself now.

This assumes deinterlace is even. needed. And in many cases, it is not, and never will be.

Again, point of reference. :)

Thanks for the clarification. Dropping the frames may make less of a difference for a video with little motion, like an interview.

As a side point, I had started to keep the double frame rate for deinterlaced files for YouTube. However, every 60 fps SD video I have uploaded to YouTube has been down converted to 30 fps. From what I have read, YouTube will reconvert each and every video, and they don't support 60 fps for SD, so keeping the double frame rate is not worth it if your only intention for the deinterlaced version is to upload online.

Thanks for the info in regards to the framerate. My mindset was to always use 29 frames in order to avoid the "soap opera" effect but it actually doesn't look too bad. I may play around with it a bit more. My end goal is to have an archive (mpeg2?), an h264/MP4 for online viewing, and then another for Bluray/DVD (unless the archive can be used for this as well).

Looking over the script I pasted above, I have it set to output RGB so I can then take it into Virual Dub for some filters. I know I set the Video Options to "processing mode" once I do this, however once I create the new AVI, do I change the color depth or lagarith compression to anything other than YUY2?

Thanks!

If you plan to output to a final video file with no further steps, then set it to YUY2, if that is what the option says (I do not have it front of me). If you plan to process it in a NLE that does not accept YUV color space, then you would set it to RGB.