You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Commenting out that line gives the expected behavior.
To Reproduce
Steps to reproduce the behavior:
Go to the img2img tab.
Keep the example image and enter any prompt.
The script encounters an error:
File "C:/Code/stable-diffusion/scripts/webui.py", line 1343, in img2img
output_images, seed, info, stats = process_images(
File "C:/Code/stable-diffusion/scripts/webui.py", line 859, in process_images
samples_ddim = func_sample(init_data=init_data, x=x, conditioning=c, unconditional_conditioning=uc, sampler_name=sampler_name)
File "C:/Code/stable-diffusion/scripts/webui.py", line 1249, in sample
xi = x0 + noise
RuntimeError: The size of tensor a (128) must match the size of tensor b (64) at non-singleton dimension 3
Expected behavior
The image should be cropped.
The text was updated successfully, but these errors were encountered:
You can change this locally for now, just change init_img to image in the line you mentioned.
It was mentioned here, but not implemented for some reason: #398 (comment)
Describe the bug
In init() the line
image = init_img.convert("RGB")
is repeated after resizing, leading to the original size being used: https://github.com/hlky/stable-diffusion-webui/blob/45fafec68a14187bc65f8e2b7015e3f4ace3c358/webui.py#L1193This then gives an error when the input image is not 512x512.
Commenting out that line gives the expected behavior.
To Reproduce
Steps to reproduce the behavior:
Expected behavior
The image should be cropped.
The text was updated successfully, but these errors were encountered: