Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Upscaler? #41

Open
mr-september opened this issue Sep 20, 2024 · 1 comment
Open

Upscaler? #41

mr-september opened this issue Sep 20, 2024 · 1 comment

Comments

@mr-september
Copy link

In the A1111 Tiled Diffusion extension, there's an option for selecting an upscaler (e.g. ESRGAN). I can't seem to find that in this ComfyUI extension. Does it exist? Or is done with other workarounds?

@Adreitz
Copy link

Adreitz commented Nov 29, 2024

You will need a series of nodes for this. After your initial generation, you will need a normal VAE Decode node. You should then feed the image into an Upscale Image (using Model) node and use a Load Upscale Model node to choose which model you want to use. If the model upscales to a factor other than the one you want to use (for example, I use 4xUltraSharp, but want 3x the initial image generations) feed the upscaled output into an Upscale Image By node using Lanczos method and set the factor based on the ratio between the model's factor and your desired factor -- in my case, 0.75. Then re-encode the upscaled image using the VAE Encode (Tiled) node to avoid going out of memory. You can then send the latent to your tiled diffusion step.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants