Newbie capturing Japanese VHS using Amarec with GV-USB2 (Deinterlace Issue)

[waxstone]

Hi everyone,

I was hoping someone could let me know if this workflow is good for capturing VHS via Amarec in great quality. Like many others I've had issues with Vdub, specifically audio sync, which makes it a no-go.

My hardware setup is: Panasonic NV-SVB10 -> S-Video -> GV-USB2 -> Windows 10 PC -> Amarec

These are my Amarec settings:


[Attachment 82347 - Click to enlarge]


[Attachment 82348 - Click to enlarge]


[Attachment 82349 - Click to enlarge]

When playing back on VLC is looks fine interlaced standard, but when I play it apply deinterlace via VLC I notice that the quality kinda drops when you look at it frame by frame, as if there are tears within the image, I've uploaded a video capture sample.

Deinterlace OFF:

[Attachment 82350 - Click to enlarge]

Deinterlace ON: (notice the black sharp cut offs in the text)

[Attachment 82351 - Click to enlarge]

I need some advice before moving forward:

• Do these settings look good? Any video codec alternatives?
  
  
• If I wanted to batch encode, let's say, 20 x 3-hour video captures, is there a simple workflow you would recommend?
  
  
• Are there tutorials and workflows/scripts recommended for denoise, crop, resizing to the correct aspect ratio, and finally encoding?
  
  
• I'd deinterlace the video files into something lossy, in a smooth 60p. 80% of the video files do not need to be in lossless quality, but enough quality where you wouldn't notice a drop (if that makes sense).
  

The end output for most files doesn't need to be lossless. I'd ideally like to batch convert a bunch of 3-hour tapes into a high-quality lossy file where you generally wouldn't be able to tell the difference between it and the lossless. I'd also be keeping clips or selections in lossless.

Any help is much appreciated! I'm on Discord. Maybe someone would be able to walk-through this with me. Willing to help for your generosity ($)

[lordsmurf]

- Amarec shows deinterlacing is TFF there, not off.
- In VLC, use Yadif 2x as the mode.
- Best deinterlacer is QTGMC, and just encode a copy (NOT ORIGINAL!) version to view. Use Hybrid, pay attention to deinterlace settings.

[lollo]

The "Deinterlacing" setting in Graph2 (Preview) tab are only for AmarecTV display windows, and has no effect on capture.

Set "Frame rate" in "Recording" tab to 999, to force the frame rate of the capture card rather than defining it in software.

Enable "Make a report file" in the "Advanced" tab, to keep trace of inserted and dropped frames at software level.

Your setings are OK, many others (including myself) use HuffYUV rather than Lagarith, but that's fine.

A real time deinterlacing with VLC is not that effective.

Are there tutorials and workflows/scripts recommended for denoise, crop, resizing to the correct aspect ratio, and finally encoding?

there is no resizing to correct aspect ratio required. In AviSynth crop to 704x480 masking head switching noise, deinterlace with QTGMC or inverse telecine with TIVTC/TFM, denoise with TemporalDegrain2 or similar, perform some sharpening with LSFmod or similar, and encode to h264 with ffmpeg adding 4:3 DAR flag (there are many example in the forum)

Additional defects in the capture require specific plugins in AviSynth.

[Alwyn]

I analyse all my captures with VDub2, not VLC. You can quickly deinterlace using the internal filter (et it on Double Rate, TFF or BFF as the case may be) and then step through, frame by frame (or field by field) to check what's going on.

[lollo]

I analyse all my captures with VDub2, not VLC. You can quickly deinterlace using the internal filter and then step through, frame by frame (or field by field) to check what's going on.

Which, from a technical point of view, is wrong as well. The architecture of the video must be assessed on the original fields, without any deinterlacing operation (as said many times).

[waxstone]

The "Deinterlacing" setting in Graph2 (Preview) tab are only for AmarecTV display windows, and has no effect on capture.

Set "Frame rate" in "Recording" tab to 999, to force the frame rate of the capture card rather than defining it in software.

Enable "Make a report file" in the "Advanced" tab, to keep trace of inserted and dropped frames at software level.

Your setings are OK, many others (including myself) use HuffYUV rather than Lagarith, but that's fine.

A real time deinterlacing with VLC is not that effective.

Are there tutorials and workflows/scripts recommended for denoise, crop, resizing to the correct aspect ratio, and finally encoding?

there is no resizing to correct aspect ratio required. In AviSynth crop to 704x480 masking head switching noise, deinterlace with QTGMC or inverse telecine with TIVTC/TFM, denoise with TemporalDegrain2 or similar, perform some sharpening with LSFmod or similar, and encode to h264 with ffmpeg adding 4:3 DAR flag (there are many example in the forum)

Additional defects in the capture require specific plugins in AviSynth.

I’ve corrected my Amarec settings, all looking good (except I’m not sure why HuffYUV doesn’t appear, so using Lagrith)


I tested was able to get started with AVISynth using this tutorial: http://macilatthefront.blogspot.com/2021/01/deinterlacing-with-avisynth-and-qtgmc.html
I applied QTGMC, and it looked good until closer inspection where there were artefacts from other frames present in the preview (see photos)
Original code:
SetFilterMTMode ("QTGMC", 2)
FFMPEGSource2("Test.avi", atrack=1)
ConvertToYV12()
AssumeTFF()
QTGMC(preset="Slower", EdiThreads=2)
BilinearResize(640,480)
Prefetch(6)


[Attachment 82367 - Click to enlarge]

I reached out to another member who said that this can be due to it being animation which is 24fps. They recommended I use StaxRip which has a whole bunch of scripts built in, specifically IVTC.
IVTC looks like it solved (though there were lots of options) and frame by frame it looks good. Now the aspect ratio doesn’t look correct and I’m confusing myself with the front-end. I can’t even get a good video output. I'd prefer to stick with the script.

What would be the next steps to get IVTC this working on a .avs script, I can't find clear tutorials for the following:

• inverse telecine with TIVTC/TFM
• denoise with TemporalDegrain2
• perform sharpening with LSFmod

Now I’ve realised that I have 100+ hours of Japanese broadcast footage, which is mainly animation but also contains commercials which I really want to keep intact with the video file.

Is it true that QTGMC will only work on the standard broadcast footage which is 29.97fps (like commercials), and IVTC will only work on the anime footage which is 24fps? Is there no way to bulk apply this to a large 2 hour video?

There will be certain clips I’m extracting such as commercials etc. Trying to get my head around this workflow.

[Alwyn]

Fixed it for you:

The architecture of the video must is best be assessed on the original fields

That is your preference. It's not mine. I find it's impractical to create an AVISynth script just to set up a quick deinterlace filter process for analysing fields. I wouldn't even know what to look for on a Separate Fields that I can't already see in DR Deinterlacing. Please give us an example of something that would be revealed by the Separate Fields script verses a quick VDub Deinterlace.

[Sharc]

@Alwyn: at least suggest bob- or double rate deinterlacing rather than just saying deinterlacing....
Apart from this, field separation is the way to go, as any deinterlacer synthesizes something which is not in the original footage.

[Alwyn]

at least suggest bob- or double rate deinterlacing rather than just saying deinterlacing....

Fair call, I've edited my post above.

[lollo]

Fixed it for you:

The architecture of the video must is best be assessed on the original fields

That is your preference. It's not mine. I find it's impractical to create an AVISynth script just to set up a quick deinterlace filter process for analysing fields. I wouldn't even know what to look for on a Separate Fields that I can't already see in DR Deinterlacing. Please give us an example of something that would be revealed by the Separate Fields script verses a quick VDub Deinterlace.

https://forum.videohelp.com/threads/413554-Deinterlacing-in-2024/page5#post2735907

[Alwyn]

In my opinion, and I sit to be corrected, which I'm sure will happen!

I just had a look at your video double-rate deinterlaced and it's not interlaced, it's Progressive Segmented Frame. Every second frame (was a field) is basically the same as the last, so you have pairs of frames. So you should not use QTGMC on it (at least, the clip you gave us in the first post.

To "fix" that, just do a AVISynth/TFM() on it, or in VDub, apply the Field Delay filter.

So, in your script, replace QTGMC with:

TFM()
TemporalDegrain2(degrainTR=3) #change the "3" to lower the noise reduction; BTW this hash symbol signifies non-executable notes in a script.

I know nothing about LSFMod so somebody will hopefully chime in on that.

You may need extra DLLs for those; AVISynth will spit a warning at you if needed.

[lollo]

I applied QTGMC

QTGMC is a deinterlacer and should be applied on interlaced material (it may also fixes defect on progressive source, but that's another story)

First of all you have to understand the nature of your video, some very basic stuff here:


more complicated here:


search the forum for other examples

What would be the next steps to get IVTC this working on a .avs script, I can't find clear tutorials for the following:

• inverse telecine with TIVTC/TFM
• denoise with TemporalDegrain2
• perform sharpening with LSFmod

There are no tutorials, but several example of processing in the forum.

Is it true that QTGMC will only work on the standard broadcast footage which is 29.97fps (like commercials), and IVTC will only work on the anime footage which is 24fps? Is there no way to bulk apply this to a large 2 hour video?

There will be certain clips I’m extracting such as commercials etc. Trying to get my head around this workflow.

You can try some single rate deinterlacing which will act only on interlaced segments like this:

TDeint(mode=0, full=false, cthresh=9, chroma=false, MI=16, edeint=QTGMC(FPSDivisor=2), emask=TMM(mode=0), slow=2)

but the best would be to work differently (deinterlace or inverse telecine, or whatever) for thr specific portions of the video.

edit: here a sample for you https://forum.videohelp.com/threads/412444-Help-with-guidance-on-Avisynth-scripts for an interlaced source

[Alwyn]

Are you using AVISynth Plus, 64 bit? If so, I have something for you.

[Alwyn]

@Lollo "but possible for determining the fields architecture (interlaced, pulldown, telecine, phase-shift)", yes, I know (all the other dot points are irrelevant and personal preference only).

All of those aspects are easily discovered by DR deinterlacing except for phase shifting, which I have never seen and wouldn't know what to look for. Do you have an example of it, and what's the issue with it?

[lollo]

All of those aspects are easily discovered by DR deinterlacing except for phase shifting, which I have never seen and wouldn't know what to look for. Do you have an example of it, and what's the issue with it?

Do you really need an example to understand? Sharc partially already answered. And read the documents I attached earlier.

In any case, stay with your flow, I do not care. Just do not propose it to newbies