Can I reuse a model??

[roydavisParticipant]

So, If I train a new model for swapping face A to face B, can I reuse that model as a starting point to swap face A with face C? Is there a benefit to doing this?

[richard7311Participant]

Absolutely re-use a model! The more you use it, the faster it gets at learning new faces.
Ideally, your model should have 8-10 Meg iterations.
If, as you say, you want to use it again to fake A to C, you can then copy it to use it
to train other src sets. You don’t have to make more than one highly trained model, but it
does shave off a bit of time when you use a new dst set.

[vaibhavjhaParticipant]

So how do we do this? Just add a new data_dst file and run the entire process only for dst?

[Author]

Posts