search

TemryL / ComfyUI-IDM-VTON

ComfyUI adaptation of IDM-VTON for virtual try-on.
GNU General Public License v3.0
370 stars 40 forks source link

convert return images to tensor, to ensure broader compatibility #30

Closed cardosofelipe closed 4 months ago

cardosofelipe commented 4 months ago

The current node version returns images as a list. This causes errors if we try to use the output image as input for another node. For example if we pipeline two IDM-VTON, using the output IMAGE of the first as human_img input of the second, we got the following error:

Error occurred when executing IDM-VTON:

'list' object has no attribute 'squeeze'

This was an example, but this happens also with other nodes that uses the IMAGE as input. Just by converting the list in tensor fixes the problem.

TemryL commented 4 months ago

Resolves https://github.com/TemryL/ComfyUI-IDM-VTON/issues/27

That's right, thanks for your contribution!