OutOfMemoryError: CUDA out of memory.

[mts1010]

Hello, I am running into a VRAM issue with SD. I am trying to enhance some clouds on a JPEG image, original resolution 3840x2560 px. If I set the resize scale to 1 I get the following message:

OutOfMemoryError: CUDA out of memory. Tried to allocate 4.69 GiB (GPU 0; 12.00 GiB total capacity; 9.19 GiB already allocated; 1.05 GiB free; 9.21 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

I can only get SD to work if I set the resize scale to 0.5. The problem is I need the output in original resolution 3840x2560 px. Is 12GB VRAM really not enough for this kind of resolution? Or is there something I can do?

[w-e-w]

you can try Tiled VAE from this extention https://github.com/pkuliyi2015/multidiffusion-upscaler-for-automatic1111
https://github.com/pkuliyi2015/multidiffusion-upscaler-for-automatic1111/wiki/Tiled-VAE

[mts1010]

Thanks, with tiled VAE I can keep scale at 1. Couldn't get Tiled diffusion to work with scale 1.