Hi all! I'm a bit of a "newb" when it comes to video capturing, so figured I'd post my question here.
I capture VHS tapes using the Elgato Video Capture and the software that came with it. I know it's not the best, but I'm mostly happy with the results! Except: the digital videos created have some problems with what I believe is "frame interpolation".
Here's one rip I uploaded to YouTube which has plenty.
Here's also a screenshot from the source file.
[Attachment 48778 - Click to enlarge]
So the question is, what's the easiest/best way to solve this issue? Should I use different software with different settings? Should I look into some post-processing using Handbrake?
+ Reply to Thread
Results 1 to 11 of 11
Oops, I meant to post this in the "Newbie" sub-forum, and I can't find a way to delete or move my post... Sorry!
Bad capture device.
Even worse software.
That's blended deinterlace, not interpolation.
This cannot be solved. The video must be re-captured with better hardware, software, and methods.
And it can look better, quite easily, with the right tools being used.
Perhaps the OP should post some of his original capture file here to the thread, not the Youtube processed version
The video has been blend deinterlaced. There's no telling where along the line the problem occurred. So you need to do some detective work. Turn of any deinterlacing option in the capture device/software. Then capture a short (1 minute) segment with motion -- including a smooth panning shot. Upload the captured video to this web site (up to 500 MB is accepted) without any conversion.
Thank you for the responses so far!
I did some brief testing, and included below you will find a short copy of the file that was created by the capture software (plus a screenshot). Not a smooth panning shot, unfortunately. If it's important, I can see if I can find something similar in one of my VHS tapes.
Also included are a few screenshots from the software that came with the Elgato Video Capture. As you can see, not many options here. Is there any other software I should try instead? Preferably something that doesn't require too many hours to get up and running.
From the looks of the video file the VHS tape was an analog NTSC to PAL conversion with field blending. On top of that the capture has been blend deinterlaced adding even more blending.
Is that MP4 file exactly what the device captured? Or did you re-encode it? The software was set up to "Preserve Source Format" which, I presume, should give you interlaced video. But the file is encoded progressive and there are no comb artifacts. If that is the file the El Gato captured they must have blend deinterlaced before writing the video data to the file. And you're screwed -- there's no fixing it.
Yes, the video file I uploaded was what the software spit out, no re-encoding on my part (but perhaps the software does it when processing the file at the end?). I also tried capturing with "Preserve Source Format" turned off during the same test session, but the results seemed pretty much the same to me. See the attachment to this post for that clip.
That's a very low quality, low bitrate capture, encoded as progressive. Get rid of that Elgato crap. If you intend to rework, modify, and cleanup VHS you're definitely going down the wrong road.- My sister Ann's brother
The new video is similar the last. It looks like the capture device is deinterlacing with a blend algorithm. Have you tried capturing other tapes to compare? Try some home camcorder video or live sporting events. Those are usually pure interlaced video.
There's just too much blending in those caps. You won't be able to remove it and keep smooth motion. Most of the character animation is at ~8 fps. And during the 3 times each of those frames is repeated there is usually 1 that shows no blending. You can use SRestore(frate=8.??) to retain those clear frames while throwing out the blended ones. But that will make the panning shots very jerky.
Does this happen for non-animated programs? I am assuming this is material is PAL, so it shouldn't be telecined, but who knows.