VHS PAL cutting and upscaling

[prodarek]

Hello,
I am trying my hand but to no avail.. someone can show me how should look property cutting and upscaling? I will be very grateful

Tools what I have avidemux and Hybrid.

this is my file :

Video
ID                                       : 256 (0x100)
Menu ID                                  : 1 (0x1)
Format                                   : MPEG Video
Format version                           : Version 2
Format profile                           : Main@Main
Format settings, BVOP                    : No
Format settings, Matrix                  : Default
Format settings, GOP                     : Variable
Format settings, picture structure       : Frame
Codec ID                                 : 2
Duration                                 : 37 s 920 ms
Bit rate mode                            : Variable
Bit rate                                 : 9 090 kb/s
Maximum bit rate                         : 9 300 kb/s
Width                                    : 720 pixels
Height                                   : 576 pixels
Display aspect ratio                     : 4:3
Frame rate                               : 25.000 FPS
Standard                                 : PAL
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Interlaced
Scan order                               : Top Field First
Compression mode                         : Lossy
Bits/(Pixel*Frame)                       : 0.877
Time code of first frame                 : 00:04:24:12
Time code source                         : Group of pictures header
GOP, Open/Closed                         : Open
Stream size                              : 41.1 MiB (92%)
Color primaries                          : BT.601 PAL
Transfer characteristics                 : BT.470 System B/G
Matrix coefficients                      : BT.470 System B/G

https://drive.google.com/file/d/1UHP9ixQwmRCLIm_9XvIN_3EjbAd9LGX6/view

[Sharc]

What exactly is the problem you are struggling with?

Your capture is pretty good quality, but if you absolutely want to process it you can basically use the video filters which are included in Avidemux and Hybrid, like:
- (bob-)deinterlace filter for deinterlacing the capture
- crop filter for cutting off the left/right black borders and the head swithching noise at the bottom
- resize (upscale) filter for scaling your cropped picture to the desired frame size and aspect ratio
Finally re-encode and remux the video + audio using the encoder and container of your choice (x264, mp4 for example)

You can do this with either Avidemux or Hybrid, but perhaps you may find Avidemux easier to use (menu Video -> Filter).
You may also google for YouTube tutorials.

[prodarek]

- (bob-)deinterlace filter for deinterlacing the capture

I tried QTGMC in Hybrid and resultat is awesome this is good choice?

- crop filter for cutting off the left/right black borders and the head swithching noise at the bottom

Yes I want remove borders and head noise but I'm not sure how to set good size and after this upscale to 1080p?

I need export to Huffyuv after every process?

remove deinterlacing > save file to Huffyuv
crop > save file to Huffyuv or maybe crop and upscale can be done in one process?


What about noise and sharpen?
I notices after export to x264 QR-17) image is cleaned from details.

Thank you.

[Sharc]

Yes, QTGMC is fine

Then:
crop(12,2,-18,-10) #crop(12 left, 2 top, 18 right, 10 bottom) to frame size = 690x564
spline36resize(1440,1080) #upscale to 1080p, 4:3, square pixels (you can use a resizer of your choice)

Finally reencode and remux.

No need for intermediate huffyuv etc.
No need to do more denoising because QTGMC already applies some denoising. More denoising would just loose more details.
Also keep in mind that upscaling does not add or "recover" any more details.

Also, it would be better do use the original (lossless) .avi capture (if you have it) as your source for post processing, rather than the mpeg2 encoded version which you posted. It would avoid the mpeg2 compression step of your .ts file, means finally 2 compression steps which deteriorate the quality.

[dellsam34]

How did your file end up in mpeg-2? How did you capture it? Basically for this kind of processing you want to start from a lossless file, AVI 4:2:2 720x576 50 fields/s, The process for a lossless file goes like this:
- De-interlace using QTGMC in fast mode to avoid excessive sharpening.
- Crop to the bare active video area in a ratio close to 704:576, meaning that if Vcrop is 570 lines Hcrop should be around 570x704/576=696 pixels, give or take, few pixels off are not going to be noticeable.
- Resize to 1440x1080, I personally find LanczosResize to be the best, less stair steps, yet less blurring.
- Encode to h.264

Frame Work:

AviSource("C:\file path\file.avi")
AssumeTFF() OR AssumeBFF()
QTGMC(Preset="Fast")
Crop(HL, VT, HR, VB) (H horizontal, V vertical, L left, R right, T top, B bottom)
LanczosResize(1440, 1080)

Using AVSPmod will give you a visualization frame of all steps above.

Encoding:

ffmpeg -i Input.avi -vf setsar=sar=1/1 -c:v libx264 -crf 18 -x264opts colorprim=bt709:transfer=bt709:colormatrix=bt709:force-cfr -c:a aac -b:a 192k -movflags +faststart Output.mp4

[prodarek]

"Also, it would be better do use the original (lossless) .avi capture "
Yes I know but I use pioneer dvr-560h recorder so far... soon I will test crop and upscaling. Thx.

[dellsam34]

Yes as I mentioned above it is better to use the original lossless capture, I guess we posted at the same time. Upload a short lossless sample here (<500MB) and I will give it a go and then you can use the same parameters I used.

[prodarek]

Hmm I'm not sure what you mean this sample file "mpeg-2" this is my only file what I have, I use pioneer dvr-560h to capture image from vcr and burn on disc. And this file is from dvd disc .

[prodarek]

So far I tried Hybrid I don't know yet how to use script

QTGMC_Fast_Lanczos_x264_QR18_696x570_to_1080
https://drive.google.com/file/d/1c9g7-8lTByf1cSJsUHoYwhNa4eufhN5a/view?usp=sharing

QTGMC_Fast_Lanczos_x264_QR18_690x564_to_1080
https://drive.google.com/file/d/1GighongG8p4zdl4DDyYPBAGxpdd6b-qa/view?usp=sharing

[dellsam34]

Hmm I'm not sure what you mean this sample file "mpeg-2" this is my only file what I have, I use pioneer dvr-560h to capture image from vcr and burn on disc. And this file is from dvd disc .

Capturing straight to mpeg-2 leaves no room for editing, It is acceptable for direct playback. Anything you do to it requires re-encoding, after you do all those processes mentioned above there will be no quality left in the final file. Maybe it's time to start learning how to capture lossless, plenty of threads here and the learning curve is quite steep.
But it all depends on someone expectations, if you are happy with what you are doing that's all what matters.

[prodarek]

Hmm I'm not sure what you mean this sample file "mpeg-2" this is my only file what I have, I use pioneer dvr-560h to capture image from vcr and burn on disc. And this file is from dvd disc .

Capturing straight to mpeg-2 leaves no room for editing, It is acceptable for direct playback. Anything you do to it requires re-encoding, after you do all those processes mentioned above there will be no quality left in the final file. Maybe it's time to start learning how to capture lossless, plenty of threads here and the learning curve is quite steep.
But it all depends on someone expectations, if you are happy with what you are doing that's all what matters.

Yes I planning to this, first I wanted to test some things. I need to read about capture cards.

[Sharc]

QTGMC_Fast_Lanczos_x264_QR18_690x564_to_1080
https://drive.google.com/file/d/1GighongG8p4zdl4DDyYPBAGxpdd6b-qa/view?usp=sharing

It is about as good as it can get from your mpeg-2 source.

[Selur]

It is about as good as it can get from your mpeg-2 source.

Since the target isn't denoising etc., but more upscaling while keeping the target as similar to the source as possible, NNEDI3 probably is a better choice than lanczos.
Personally I would add Deblock_QED and TemporalDegrain:


Cu Selur

[prodarek]

Awesome! can you show me all settings to obtain such a result?

[Selur]

I simply enabled:

• Filtering->Vapoursynth->DeBlock->Deblock_QED, and
• Filtering->Vapoursynth->DeGrain->TemporalDeGrain

I didn't try to tweak any settings or try different filters, I just wanted to deponstrate that if keeping information isn't the goal there are different filters that might improvde the visual appearance,...


In general you might want to:

• enable: "Filtering->Filter view" (by default this will cause the AviSynth/Vapoursynth preview to show a dual view preview which shows the original on the left and the filtered version on the right.
• enable "Filtering->Synth auto referesh" which will automatically refresh the preview on any filtering change.

and then at least try to learn a bit of the effect of different filters by trial&error.

Cu Selur

[prodarek]

Thx, I have also diffrent question about Hybrid and AviSynth/Vapoursynth, linux manjaro this is my native system but so far I cant enable AviSynth/Vapoursynth I dont know why but Hybrid dont see anything.

I have installed :
https://aur.archlinux.org/packages/hybrid-encoder
vapoursynth R58-1

So everything what I doing this is a virtual machine with win10, can you help me? thank you.

[Selur]

a. That is something that you should lead your post with.
b. There are some users over at forum.selur.net like Wisper who use Arch Linux and Hybrid with Vapoursynth, they probably can help you, if you ask them on how to setup Hybrid on Arch Linux based system

Cu Selur

Ps.: There is no AviSynth support on Linux in Hybrid.

[prodarek]

Ok thank you I will check.

[Sharc]

..... NNEDI3 probably is a better choice than lanczos.

Any plans to incude it in the resizing filters collection. (Crop/Resize Tab -> Resize method dropdown selection)?

[Selur]

Nope, Hybrid does offer to overwrite that selection using "Filtering->Avisynth/Vapoursynth->Frame->Resizer".

[Sharc]

Thanks. Apparently I have missed this option
(actually I hardly ever upscale).

[prodarek]

I know about this function from here https://forum.videohelp.com/threads/404634-NTSC-VHS-summary-6-Hybrid-for-Conversions-D...scaling-to-MP4

[Brad]

Frame Work:

AviSource("C:\file path\file.avi")
AssumeTFF() OR AssumeBFF()
QTGMC(Preset="Fast")
Crop(HL, VT, HR, VB) (H horizontal, V vertical, L left, R right, T top, B bottom)
LanczosResize(1440, 1080)

Using AVSPmod will give you a visualization frame of all steps above.

Encoding:

ffmpeg -i Input.avi -vf setsar=sar=1/1 -c:v libx264 -crf 18 -x264opts colorprim=bt709:transfer=bt709:colormatrix=bt709:force-cfr -c:a aac -b:a 192k -movflags +faststart Output.mp4

Technically, your script should include ColorMatrix(mode="Rec.601->Rec.709") or your ffmpeg command-line should do this conversion. Currently you're starting with 601 and then telling x264 that it's 709.

[prodarek]

@Brad
Thx for tips.

[dellsam34]

Technically, your script should include ColorMatrix(mode="Rec.601->Rec.709") or your ffmpeg command-line should do this conversion. Currently you're starting with 601 and then telling x264 that it's 709.

The resized file is still lossless AVI, the conversion to rec.709 is only needed when encoding to h.264 which I believe the ffmpeg commend does convert to 709, Doesn't it?

[Selur]

Strickly speaking the conversion to 709 is not needed, it's only used by users which do either not properly set the VUI information in the video stream or do not trust that the player they use properly follow the VUI tags.

Cu Selur

[lollo]

Strickly speaking the conversion to 709 is not needed, it's only used by users which do either not properly set the VUI information in the video stream or do not trust that the player they use properly follow the VUI tags.

AFAIK, when converted back to RGB for display, HD is generally considered as Rec709. The proper way to upscale is to add the ColorMatrix conversion (YUV 601 to YUV 709) before the resizing.

In addition, I suspect that an upscaled 1280x1440 HuffYUV file as mentioned have no flags or tags.

[Selur]

AFAIK, when converted back to RGB for display, HD is generally considered as Rec709. The proper way to upscale is to add the ColorMatrix conversion (YUV 601 to YUV 709) before the resizing.

Only if there is no color matrix information etc. present. Same way with 4k and Rec.2020. If no color matrix info is present Rec.20202 should be assumed, but it could be everything and if done properly VUI information should be considered during conversions.

HuffYUV file as mentioned have no flags or tags.

I agree when you use a format that has no VUI signaling and you are not doing the conversion 'manually' (set matrix&co), then you should stick with the assumed default.

[lollo]

Indeed, we are talking about VHS PAL

[Selur]

Source doesn't really matter, whether the used format supports vui tagging and whether the used software properly respects the tagging is the question.