This is just a GUI issue, not showing the preview about it.
Specifically there is a permission issue on your machine accessing the preview file.
Don't worry about it. Whilst you can't see the preview, it won't impact the ability to convert.
This is just a GUI issue, not showing the preview about it.
Specifically there is a permission issue on your machine accessing the preview file.
Don't worry about it. Whilst you can't see the preview, it won't impact the ability to convert.
@ianstephens Thanks for testing. Unfortunately I have been very busy of late, so not much time for development.
I have made a note of your feedback and will investigate as soon as I get an opportunity.
I'm don't know anything, offhand, that will do this, but Faceswap will not be suitable as the style of the swapped face will be very real.
This is not an issue I have seen, but it sounds like the model has collapsed.
Roll back 50k iterations and lower the learning rate
This is likely a bug in the code somewhere.
Sadly, it will be hard to find without the original data.
If you can provide: The video that you are converting + the alignments file, I can look into this (You are welcome to send me a link via private message)
A screen grab of the preview window would be useful...
However, this sounds almost definitely that you selected the extended/components mask when starting training.
The selected mask is 'locked' to the model, so changing it after the model has started training will have no effect.
Unfortunately not directly, as this is 'locked' to the model.
You could edit the model's statefile, although this is unsupported, and I'm not sure there is any value in switching the loss in this way.
I will look into it when I have time (sorry for slow response, I'm very busy at the moment!)
I don't think I've heard of anyone training it to completion, for the reasons you state. It takes a very long time.
Would be interested to know how you get on.
I will also need the source video that this alignments file relates to, to be able to diagnose this, I'm afraid.
Thank you for confirming
I'm afraid not. It is not an issue that has impacted me for a long time. As I said, it is likely to be data related, so make sure you are well covered off on both sides for the scenes you are struggling with.