How to use the pre-training model to generate other images?

jgkwak95 / SURF-GAN

[ECCV 2022] Official Pytorch implementation of "Injecting 3D Perception of Controllable NeRF-GAN into StyleGAN for Editable Portrait Image Synthesis"

MIT License

127 stars 14 forks source link

How to use the pre-training model to generate other images? #7

Closed hxl835097324 closed 2 years ago

hxl835097324 commented 2 years ago

Hi @jgkwak95, thanks a lot for sharing the code for your work. When I was learning your project code, I didn't find a place to input my own pictures.

Thank you very much !

hxl835097324 commented 2 years ago

zs = sample_latent((num_id, 9, 6), device=device, truncation=opt.psi) Is it this line of code?

jgkwak95 commented 2 years ago

Hi @hxl835097324, Thanks for your attention.

SURF-GAN generator is a random image generator which takes a latent vector as input. Of course you can find a target latent vector through GAN inversion, but the results will probably be bad because of low resolution. That is why we distilled to StyleGAN.