Read the FAQs and search the forum before posting a new topic.
This forum is for discussing tips and understanding the process involved with Training a Faceswap model.
If you have found a bug are having issues with the Training process not working, then you should post in the Training Support forum.
Please mark any answers that fixed your problems so others can find the solutions.
I am a newbie and have questions about A-face training.
Assume A-face video is 5 minutes
the frame rate is 30fps/s (a total of 9000 images)
If I want to completely replace the video, then 9000 images should be kept during training?
or like B-face pick different angles. expressions. lighting.
Another question
I have a model that trains well
I want to replace other person's videos with the same B-face.
Do I retrain a new model?
or change person A-face and Inherit the same model to continue training?
You can keep them all, but it isn't necessary. We have the "Extract every N" option to limit the number of faces that you can extract. Getting different angles. expressions + lighting from multiple sources is preferable though.
Ideally you require a new model, however you can re-use the weights from the previous model using the load/freeze weights feature.
My word is final
