zhujiapeng / lowrankgan Goto Github PK

After reading your paper and code, I think you use the W space. Is that true? Can I use this method to find an editing direction in W+ space?

What kind of GPU and environment did you used for computing Jacobian?
I seems like it won't end till I die...
I just want to see the feasibility..

Thanks,
Joseph

I would like to know what version of gcc and cuda should be used when running on Linux?

Hi,

Thanks for the interesting work!

I was wondering if a PyTorch version is still planned?

Hello thanks for providing source code for your awesome work!!
Since there's no requirements.txt file, and I am not familiar with tensorflow I am having hard time making the environment for this repository
Can you specify version and requirements for this work?

Thanks in advance !

Hey folks, great work!

I'm following the steps you outline in the Find More Directions section of the readme, for FFHQ with StyleGAN2. I'm finding that all the (4) directions it finds for both the "eyes" and "mouth" region correspond roughly to either open/close of the eyes/mouth. Is this expected?

Did you find any way of locating additional directions at a given region (perhaps more fine-grained ones such as eye colour or lipstick in your experiments)?

Thanks :-)!
James.