Screenshot is of me tunneling into my headless analog video capture server I put together out of an old Thinkserver and a Magewell 1080p capture card, both of which were ebay gambles.
Here’s the flow:
- VCR and Hi8 decks -->
- Kramer VPC-23 AV Switcher -->
- RetroTINK 4k, upscaled to 1080p HDMI -->
- Magewell Pro Capture card -->
- Thinkserver RS140
AntiX headless install on the server. I ssh in to kick off the capture script I wrote, which also publishes a preview video stream to my LAN. I then just kick back and tune in via mpv.
I went as far as setting up an nfs share on the server, so after a capture I can pull it onto my workstation, to trim, edit, etc.
Not op but in my case I’ve seen huge differences between USB cards and high quality cards. 20 years ago I used to use bt878 capture cards. A few years ago I bought a Viewcast Osprey on eBay.
In particular I use a svhs tape player with built in time base corrector. The reason for svhs is because the hardware supports actual svhs output. Vhs tapes are store with luminous and chroma as separate signals. With regular composite output, that data is mixed into one signal. This causes interference patterns like dot crawl and color fringing.
The other big advantage over USB dongles is all the ones I’ve seen don’t let you capture the raw data- it’s pre compressed with lossy like mjpeg or mpeg which is another huge quality drop because you’ll usually want to post process the video with avisynth before compressing it to mpeg4.
My target is for the captured video to look just as good as it originally did on a VHS player on a crt. So the interlaced to progressive conversion is always tricky. I don’t personally do it but one trick to keep resolution is to capture at 60fps and play each 30i field doubled. That way no detail is lost in the source and it uses your eyes to motion blur the fields together.