Posts: 11
Threads: 4
Joined: Nov 2023
Hi,
I recently came into possession of some footage, I am not sure of the format it was filmed in. By the looks of it, it has been deinterlaced, albeit poorly, as there is a lot of ghosting present in the footage. I've tried re-deinterlacing it using QTGMC and Bob, as well as applying an sRestore filter but have been unable to improve the ghosting. Is there anything I can do to make the footage look any better? I have attached a sample to the post.
Posts: 10.616
Threads: 57
Joined: May 2017
28.11.2023, 18:54
(This post was last modified: 28.11.2023, 18:56 by Selur.)
Sorry to say this, but since each frame and each color plane has the ghosting and you get different strong ghosting in the frames depending on the motion, I don't think you can get rid of these. (looks like someone applied blend deinterlacing to the source)
Cu Selur
----
Dev versions are in the 'experimental'-folder of my GoogleDrive, which is linked on the download page.
Posts: 11
Threads: 4
Joined: Nov 2023
That's fine! Appreciate the help.
Posts: 258
Threads: 34
Joined: Aug 2022
this profile 422 is special for interlaced video
try - in avisynth mode - bvdf with TFF/BFF or other options and definitely Top Field First or Bottom Field First
Posts: 10.616
Threads: 57
Joined: May 2017
@humanoid86: his source is not interlaced
----
Dev versions are in the 'experimental'-folder of my GoogleDrive, which is linked on the download page.
Posts: 258
Threads: 34
Joined: Aug 2022
(29.11.2023, 05:36)Selur Wrote: @humanoid86: his source is not interlaced
(29.11.2023, 03:07)humanoid86 Wrote: this profile 422 is special for interlaced video
Posts: 10.616
Threads: 57
Joined: May 2017
So you suggest applying something that is special for interlaced video on content that is not interlaced.
That just seems wrong,...
----
Dev versions are in the 'experimental'-folder of my GoogleDrive, which is linked on the download page.
Posts: 258
Threads: 34
Joined: Aug 2022
csp422 and MOV container...
probably the original codec was prores 422 or ffv1 8/10 422
Posts: 2
Threads: 0
Joined: Nov 2023
I tried to solve it with Topaz AI but apparently it is not yet intelligent enough.
Posts: 10.616
Threads: 57
Joined: May 2017
29.11.2023, 16:10
(This post was last modified: 29.11.2023, 16:12 by Selur.)
Yes, haven't seen any papers where folks try to undo blending.
(Topaz currently it only improves the blending even more)
Cu Selur
----
Dev versions are in the 'experimental'-folder of my GoogleDrive, which is linked on the download page.
