How to use custom pictures?

[rgzn-aiyun]

Hello, can you use custom pictures for reasoning?

[mchong6]

You can use custom images by doing GAN inversion. You can try using projector.py in this repo for that (details in README) or use any other standard GAN inversion techniques. Do note that the current stylegan2 I use is 256x256 while many GAN inversion (such as e4e) uses 1024x1024.

You should also be able to run my method on 1024x1024 images if you swap out the pretrained models

[rgzn-aiyun]

You can use custom images by doing GAN inversion. You can try using projector.py in this repo for that (details in README) or use any other standard GAN inversion techniques. Do note that the current stylegan2 I use is 256x256 while many GAN inversion (such as e4e) uses 1024x1024.

You should also be able to run my method on 1024x1024 images if you swap out the pretrained models

Why does it take half an hour to generate after running?

[mchong6]

It shouldnt take that long, are you using a GPU?

[rgzn-aiyun]

It shouldnt take that long, are you using a GPU?

Yes, this question is weird?