Weird face shapes in XSeg trainer

[czyzpkk]

Expected behavior

See normal faces as i labeled them.

Actual behavior

XSeg trainer looks like this: (This is from the default Elon Musk video by the way)

Steps to reproduce

I deleted the labels, then labeled again.

Other relevant information

OS : Windows 10 Home - OS Build: 19044.1949 Python version: 3.10

[CodeSmith2000]

That just looks like "Random Warp". Its a method of randomly warping the image as it trains so it is better at generalization. This is fairly expected behavior to make training more robust, unless it is incorrectly masking your faces after it has been trained and applied to merged faces. This is not a problem unless this translates to the final image that gets rendered out. If it is doing this to the merged faces then just turn off random warping in the model parameters and train for a bit longer.

[czyzpkk]

Thank you for your answer. How can i turn off random warping? I looked it up on the internet but cant find anthing.

[CodeSmith2000]

I'm not sure if you can turn off random warping for XSeg training and frankly I don't thing you should, it helps to make the mask training be able to generalize on new data sets.

[joolstorrentecalo]

Did you ever find the answer? If so, would you mind sharing it and closing this issue?