Difference between fs_model.py and projected_model.py under models folder #492

Open
opened 2025-09-24 21:12:15 +02:00 by justin4ai · 0 comments
justin4ai commented 2025-09-24 21:12:15 +02:00 (Migrated from github.com)

Hello,

Training code of the repository uses models/projected_mode.py while the provided checkpoints.zip include discriminator (D1 and D2) weights which correspond to fs_model.py;

fs_model.py has two discriminators while projected_model.py has one.

Anyone knows the difference between those two and could let me know where to get the discriminator weight for projected_model.py version?

Best,
Junyeong Ahn

Hello, Training code of the repository uses ```models/projected_mode.py``` while the provided ```checkpoints.zip``` include discriminator (D1 and D2) weights which correspond to ```fs_model.py```; [fs_model.py](https://github.com/neuralchen/SimSwap/blob/bd7b7686a17f41dd11cfcd5d82f7e4c5eb94b780/models/fs_model.py#L94) has two discriminators while [projected_model.py](https://github.com/neuralchen/SimSwap/blob/bd7b7686a17f41dd11cfcd5d82f7e4c5eb94b780/models/projected_model.py#L57) has one. Anyone knows the difference between those two and could let me know where to get the discriminator weight for ```projected_model.py``` version? Best, Junyeong Ahn
Sign in to join this conversation.