A problem I've been having when converting is that during training, the previews show normal looking mouths with good blending, however when I use the preview tool or convert it, the bottom lip of the original person is still there with the new mouth being overlayed on top of it and the lip is still very visible. I'm not sure how to fix this as I've tinkered with all of the mask settings with no change. Hopefully this is descriptive enough for some help. If I could also get the settings the training preview uses that might help as well. Thanks!
Getting errors or found a bug when converting faces from a trained model? Post about them here
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.
Update to the post, the more I look at it, its almost like the new face doesn't quite stretch all the way down onto the other persons face and just ends at the persons bottom lip which is why it can still be seen. Tinkering around with the preview tool makes me think that it's because the box blend doesn't go down far enough, but I still don't know why it looks perfectly fine on the training preview and not anywhere else. I'm still not sure how to fix this but hopefully this can give a little more info.
Without knowing your coverage settings at training time, or seeing an example I can only speculate.
However, I would suspect that it is the "Box Blend" settings that you need to adjust, as they are set too high (preview image for training does not use any blending settings).
From the convert guide (may help):
As a tip, if you change the Color plugin type to "manual-balance" and set the brightness to -100% you can see very clearly where the box/mask edges are, and the impact that changing the settings have:
My word is final