richservo comments

Repositories
Issues
Comments

Results 6 comments of


                                            richservo

Still using invisible watermark package that rarely works?

Or just comment it out of your code and move on

ModuleNotFoundError: No module named 'imwatermark'

Pip install invisible-watermark

DPM Solver doesn't support FP16 mode

Use kdiffusion. Works great and rarely exceeds 8GB vram

.half call to fix?

the tensor needs to be .half(). So wherever it's being decoded just add .half() to the end of the final tensor

[Bug] The "decode_first_stage" function in DDPM does not respect the "force_not_quantize" parameter

is this why it uses SO MUCH VRAM to decode? Above 2k isn't possible....so the new upscaller can't even be used on the 768 model.

Streamlit SD-Upscale x4, CUDA out of memory. Tried to allocate 400.00 GiB

This is actually an issue with the decode first stage. Any high res image uses an excessive amount of vram. I can actually encode a tensor of 960x704 but can’t...