Curious Question Regarding Alignments in Convert

Getting errors or found a bug when converting faces from a trained model? Post about them here


Forum rules

Read the FAQs and search the forum before posting a new topic.

This forum is for reporting errors with the Convert process. If you want to get tips, or better understand the Convert process, then you should look in the Convert Discussion forum.

Please mark any answers that fixed your problems so others can find the solutions.

Locked
User avatar
ianstephens
Posts: 117
Joined: Sun Feb 14, 2021 7:20 pm
Has thanked: 12 times
Been thanked: 15 times

Curious Question Regarding Alignments in Convert

Post by ianstephens »

Hi there,

We have 8K source videos and have extracted all alignments from the source video along with our training faces.

When converting it's almost impossible to run at 8K-->8K.

So we tried 8K-->4K (50%) output scale. This works better but it gets to a point where it runs out of memory - as if there is not a complete garbage collection going on somewhere.

But the real question I wanted to ask is... we have extracted the alignments from an 8K input video. If we downscaled the source video to 4K before convert (using ffmpeg) - could these alignments be used on the 4K video? Or do we need to re-run the downscaled 4K version through extraction afresh?

Thank you for your tips, help, and suggestions.

User avatar
torzdf
Posts: 2651
Joined: Fri Jul 12, 2019 12:53 am
Answers: 159
Has thanked: 129 times
Been thanked: 622 times

Re: Curious Question Regarding Alignments in Convert

Post by torzdf »

You could probably write a script that updates the alignments to a different frame size (or inject some code which does the calculation). There is no automatic way to do this though.

I would need to look into the running out of memory issue. That shouldn't happen (although admittedly i have not tested at those resolutions)

My word is final

User avatar
deephomage
Posts: 33
Joined: Fri Jul 12, 2019 6:09 pm
Answers: 1
Has thanked: 2 times
Been thanked: 8 times

Re: Curious Question Regarding Alignments in Convert

Post by deephomage »

Hey ianstephens, how much RAM does your PC have? I've extracted faces from 4K footage before, with 64 Gb. of RAM and an RTX 2080ti (11 Gb. VRAM). S3fd is resource-intensive, and 4K extract is going to be slow,

Locked