I'm digitalizing some old VHS-C tape recorded with a JVC camcorder from the late 80's (i dont have the reference anymore).
I'm playing them in a Toshiba RDXV50 VCR/DVD combin.
It's playing fine for most of them except for a group of tapes that seem to have a tracking issue. These tapes belong all to the same time period, so i feel like it was a setting in the camcorder that got changed because they all got the exact same line artefact that show up at the exact same place.
Here is a sample : https://youtu.be/yCLd-oXuzzg
So my question are :
Is it a tracking issue or something else?
can i fix it with a VCR who has a manual tracking option?
If so, what are the most accessible VCR that can do that ?
thank you very much!
+ Reply to Thread
Results 1 to 12 of 12
Thank you for your help! Good surprise to find out i have a tracking option on my VCR!
Playing with the tracking result in moving this line upper or lower but i can't totally get rid of it. When i almost manage to put it out of the way there are more lines appearing everywhere or things like that.
Any suggestion what might be the issue ? Again the weird part is that its not only this tape, its like 7 tapes over 50. And these 7 tapes have been recoded one after another. Like if I changed a setting in the camera at the time, or maybe damaged it ?
If you can't get rid of the problem you can try using two captures. For example, one cap with the noise in the top half of the frame, one with the noise in the bottom half. Then paste the good half of one cap with the good half of the other in an editor.
Median would work too. The OP will have to move the noise to three different locations to use a median of 3 caps (or 5 for median of 5 caps). So it takes more caps. On the other hand the filter can automatically adjust to dropped frames.
Why is anybody losing frames?!! If you are losing frames, your setup is not optimal (no tbc, insufficient subsystem horsepower, etc).
Yes, I assumed one would adjust tracking to get 1 w high noisebar, 1 w mid, 1 w low. The median function does not find the average it finds ONLY the most common (middle) value. So if pixel 192:235 has 2 clean & 1 noise value, the middle one is clean. Throws out outliers, whereas mean/average keeps them in the mix in there, even if their weighting is less. Plus it is amazing for reduction of any kind of playback-originated noise, since it favors the non-random.
I've tried capturing twice with the line in a different position but i don't get something clean when i compose both together. the picture flickers at different moment on both cap, there a dropped frames, etc... I will try AVISynth with more captures
FIY I'm working on Linux and capturing with a cheap EasyCap device since its the only thing i found is working on Linux.
My command line to capture is :
ffmpeg -f v4l2 -standard secam -thread_queue_size 1024 -i /dev/video2 -f alsa -thread_queue_size 1024 -i hw:1,0 -ac 2 -ab 320k -vf yadif=1,scale=768:576,format=yuv420p -c:v libx264 -crf 19 -preset veryfast -bf 0 -g 50 -t 00:10:00 -y output.mp4
Last edited by mistercoop; 17th Feb 2020 at 06:01.
You have discovered one of the major flaws with the multiple capture approach: offset fields.