denoising-diffusion-gan
denoising-diffusion-gan copied to clipboard
NVlabs
Image to Image
I've got a fork with a script that can do img2img here: https://github.com/JCBrouwer/DenoisingDiffusionGAN
It's been a while since I've used it though, and I'm not sure if my version still supports loading in the original checkpoints from this repo, but maybe you'll be lucky. Otherwise it might be a good starting point...
Thanks Hans,
Can you please specify the train/ test scripts for image to image.
With regards Jignesh
On Wed, Aug 2, 2023 at 1:26 AM Hans Brouwer @.***> wrote:
I've got a fork with a script that can do img2img here: https://github.com/JCBrouwer/DenoisingDiffusionGAN
It's been a while since I've used it though, and I'm not sure if my version still supports loading in the original checkpoints from this repo, but maybe you'll be lucky. Otherwise it might be a good starting point...
— Reply to this email directly, view it on GitHub https://github.com/NVlabs/denoising-diffusion-gan/issues/36#issuecomment-1660988814, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMHML6A3RNG6372FLVF2G63XTFNNDANCNFSM6AAAAAA276375Y . You are receiving this because you authored the thread.Message ID: @.***>
The relevant test script is interpolate.py
The readme has a short description of the parameters, the relevant one for img2img is --init_video which allows the script to perform vid2vid transfer.
You can rewrite the script fairly easily to use a single image I think, but otherwise it should also "just work" if you make a 1 frame video haha
To load the official checkpoints you might need to set a couple of the "model related options" (towards the bottom of the list in the readme)
Hi can we use this code for image to image translation?
Hi @JigneshChowdary , got any chance to make it work for image2image? Would be greatly appreciated if you could share ur trial.
