MrDeepFakes Forums

Some content may not be available to Guests. Consider registering an account to enjoy unrestricted access to guides, support and tools

  • We are looking for community members who are intested in helping out. See our HELP WANTED post.

Need regular training model

selinadfl

DF Vagrant
I have been using Fappenberg DF 192  since Feburary :cool: :cool: ....the conversion is faster and accurate with that model.....but I want to try new DF-u or DF-ud model.

I have downloaded a few of them from the trained model section which are all generic pretrained models.... :rolleyes: :rolleyes:
Due to the few hundred personal SRC faceset the result looks awful.... :shy: :shy:

Needs some regular training model U or UD.....  :) :)
 

Groggy4

NotSure
Verified Video Creator
So, you still want to stick with 192 res? I only have a high dims df-ud 256. It is equal heavy to run as df-192, but around 300ms slower pr iteration.
 

selinadfl

DF Vagrant
sure Groggy4 any model with regular training with loss value less than 0.3............... :) :) :D ............will be really appreciated.
 

TMBDF

Moderator | Deepfake Creator | Guide maintainer
Staff member
Moderator
Verified Video Creator
selinadfl said:
I have been using Fappenberg DF 192  since Feburary :cool: :cool: ....the conversion is faster and accurate with that model.....but I want to try new DF-u or DF-ud model.

I have downloaded a few of them from the trained model section which are all generic pretrained models.... :rolleyes: :rolleyes:
Due to the few hundred personal SRC faceset the result looks awful.... :shy: :shy:

Needs some regular training model U or UD.....  :) :)

If you want a model that was already trained on with random src/dst faces then collect some that you've already used, place parts of few src datasets into src folder and parts of few dst datasets into dst folder and run the model so that it trains on those random faces (again, if that's what you mean by "regular training model").
 
Top