Question / Help Best Settings for Recording VHS

[Dionis Bajrami]

I just bought a 15$ capture card, it has a RCA and a S-Video input (gonna use S-Video for better quality). And connected it with OBS works great.
Changed the canvas to the max Capture card quality 720x567. My Goal is to upscale this to 1280x1024 (5:4) currently using Lancoz.
But i need help with other settings (video and audio) i want the best quality (max 5GB per 1 hour).
Also what filters do i use for audio and video.
Basically wanna make them look better, suggest any option possible>

 

[koala]

The effective resolution of a VHS recording is about 320x240, interlaced, 50 Hz (PAL) or 60 Hz (NTSC), so set your OBS canvas size to this. It makes no sense to record any higher resolution or even upscale. This will only result in even more quality loss. You would only upscale the drops and artifacts. There is simply not more picture information in a VHS recording than that. Record and save as close to the original material as possible and let the player upscale on the fly to the device you are viewing. And don't scale and don't add any letterboxes to the recording to fit a different aspect ratio. PAL/NTSC is a 4:3 ratio and should be recorded as that. If you want to show the video on a non-4:3 device, let the player add letterboxes. But don't add these to the video. If you have VHS video recordings that contain 16:9 material with letterboxes, remove these letterboxes with the crop filter and save the video with this. This results in 320x180 video. This may look strange, but that is what's contained in such a VHS video. It's absolutely mediocre quality compared to state of the art bluray HD recordings.

If you don't feed a VHS recording but a real analog TV transmission, you can record with the max capture card quality of 720x576 and use this as canvas size, because this is what the analog signal contains. But a VHS recording contains only about half of that.

 

[Dionis Bajrami]

The effective resolution of a VHS recording is about 320x240, interlaced, 50 Hz (PAL) or 60 Hz (NTSC), so set your OBS canvas size to this. It makes no sense to record any higher resolution or even upscale. This will only result in even more quality loss. You would only upscale the drops and artifacts. There is simply not more picture information in a VHS recording than that. Record and save as close to the original material as possible and let the player upscale on the fly to the device you are viewing. And don't scale and don't add any letterboxes to the recording to fit a different aspect ratio. PAL/NTSC is a 4:3 ratio and should be recorded as that. If you want to show the video on a non-4:3 device, let the player add letterboxes. But don't add these to the video. If you have VHS video recordings that contain 16:9 material with letterboxes, remove these letterboxes with the crop filter and save the video with this. This results in 320x180 video. This may look strange, but that is what's contained in such a VHS video. It's absolutely mediocre quality compared to state of the art bluray HD recordings.

If you don't feed a VHS recording but a real analog TV transmission, you can record with the max capture card quality of 720x576 and use this as canvas size, because this is what the analog signal contains. But a VHS recording contains only about half of that.

This left me disappointed. So the best i can save it as is 720x567. ill set the output to that too.
I have two questions:
1. How do i remove the interlacing lines thingy?
2. Should i use Downscaling Filters for sharpeining such as Lanczos?

also what about audio

 

[koala]

Deinterlacing is available for several source types in the context menu of that source. Media Sources and Video Capture Sources have these.
Downscaling might be an idea. capture at 720x576 and downscale to 320x240.
For a start, I propose you capture at 720x576 and save exactly this resolution, then downscale in a postprocessing step.
Use simple output mode and not ffmpeg custom output. Choose "Indistinguishable Quality, Large File Size" as output quality and mkv as Recording format. This is going to be your raw recording and you can feed this into the postprocessing software of your choice. Record one without deinterlacing and deinterlace during postprocessing, and one with some deinterlacing.
If you mangle your source during recording too much, you are unable to correct any handling errors later. You will make many errors and you will learn how to produce better video with the right postprocessing, if you do much trial and error. It is much easier to postprocess a raw video without any mangling than to rewind and re-record the tape constantly for every new iteration.

Unfortunately, this is a very old task and much knowledge is lost. Digitizing analog material was done primarily 10-15 years ago. You might find more stuff on the net if you google for this topic.
Visit the virtualdub website and get virtualdub for postprocessing. It's somewhat old software, but it contains many filters for postprocessing digitized analog video. It may be difficult to feed OBS-recorded video to virtualdub, because virtualdub neither supports any of the native container formats of OBS, only *.avi, and it doesn't support x.264.

 

[Dionis Bajrami]

Deinterlacing is available for several source types in the context menu of that source. Media Sources and Video Capture Sources have these.
Downscaling might be an idea. capture at 720x576 and downscale to 320x240.
For a start, I propose you capture at 720x576 and save exactly this resolution, then downscale in a postprocessing step.
Use simple output mode and not ffmpeg custom output. Choose "Indistinguishable Quality, Large File Size" as output quality and mkv as Recording format. This is going to be your raw recording and you can feed this into the postprocessing software of your choice. Record one without deinterlacing and deinterlace during postprocessing, and one with some deinterlacing.
If you mangle your source during recording too much, you are unable to correct any handling errors later. You will make many errors and you will learn how to produce better video with the right postprocessing, if you do much trial and error. It is much easier to postprocess a raw video without any mangling than to rewind and re-record the tape constantly for every new iteration.

Unfortunately, this is a very old task and much knowledge is lost. Digitizing analog material was done primarily 10-15 years ago. You might find more stuff on the net if you google for this topic.
Visit the virtualdub website and get virtualdub for postprocessing. It's somewhat old software, but it contains many filters for postprocessing digitized analog video. It may be difficult to feed OBS-recorded video to virtualdub, because virtualdub neither supports any of the native container formats of OBS, only *.avi, and it doesn't support x.264.

This is very informational, thank you.
I have somethings thats bugged me for years;

Is it better to record it first WITHOUT and "touchups"(filters, downscaling etc) basically as RAW as possible then to edit it in post proc.
OR
is it better to make it look as good as possible in OBS than to do just littlle touchups in postproc(premiere pro)

 

[koala]

It depends on the filters you have in the recording software and in the postprocessing software. Usually, you should do a recording as authentic to the source as possible, and do all postprocessing on the raw video in a second step. Only if your postprocessing software doesn't have appropriate filters but the recording software has, record and beautify the footage at the same time. It's the same as with audio recordings for a movie, for example: the voice actor is recorded as clean and authentic as possible, and any audio effects are put over the existing recording in postprocessing steps.
OBS has so many filters and effects only because it mainly is a live streaming client where you have to do recording, mixing and broadcasting your material in one step in realtime. For postprocessing recordings that were never intended to broadcast life, use the effects of a postprocessing software. These effects can go deeper and don't need to render in realtime.