stable-diffusion-webui icon indicating copy to clipboard operation
stable-diffusion-webui copied to clipboard

[Feature Request]: show all image progress for batch size>1 and medvram

Open arpitest opened this issue 2 years ago • 2 comments

Is there an existing issue for this?

  • [X] I have searched the existing issues and checked the recent builds/commits

What would your feature do ?

running the UI without --medvram shows all image progress in a grid while iterating in txt2img. with the new "Image creation progress preview mode" option approx cheap/nn progress preview modes it could be done for --medvram too.

Proposed workflow

set batch size >1 at txt2img set "Image creation progress preview mode" to approx cheap or approx nn mode generate! and watch all the images progressing in a grid.

Additional information

No response

arpitest avatar Jan 06 '23 19:01 arpitest

changing this line in shared.py fixes it:

parallel_processing_allowed = not cmd_opts.lowvram # and not cmd_opts.medvram

since preview grid can be switched off in settings, i can't see reason for automatic disabling based on medvram param

arpitest avatar Feb 05 '23 21:02 arpitest

When I appiled the above fixes with —medvram options, I received the following error. Is there anything else I need to take care?

Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument mat1 in method wrapper_addmm)

blueball81 avatar Feb 26 '23 09:02 blueball81

That error message sounds like the reason this feature is disabled on --medvram. :)

(How much even is "medium" VRAM? 4GB? 8GB? 16GB?)

I'd like to second this request, if it's possible to do. It's very convenient to see a preview of a full batch of images while they are being generated.

Technologicat avatar Apr 09 '23 23:04 Technologicat