rosinality / style-based-gan-pytorch

Implementation A Style-Based Generator Architecture for Generative Adversarial Networks in PyTorch
Other
1.1k stars 232 forks source link

Some confusing #117

Open XYFFFFFF opened 3 years ago

XYFFFFFF commented 3 years ago

Thanks for your code, this is my first in this field, so i am a little confused in some places. in train(func), what is step/alpha means? thanks a lot !

rosinality commented 3 years ago

Step means training resolution of images. Alpha corresponds to interpolation between previous resolutions and new (larger) resolutions. You can refer to progressive gan / stylegan paper for more details.

XYFFFFFF commented 3 years ago

Thaxs a lot!!

XYFFFFFF commented 3 years ago

Hi, i finish the training now, and i run the generate.py to test my model, but i can't change the style of the image. could u give me some reminders?

rosinality commented 3 years ago

Could you clarify the meaning of changing the style? generate.py will generate style mixing examples.