midihex

I've noticed that too cvar66, think that is a direct gfpgan 1.4 issue- it's a pain when doing thin plate spline upscales For the main issue above I'm thinking its...

Fixed in PR #3874 as far as the extras tab goes, still needs fixing when using highres.fix

you can add --vae-path to arguments to do most of this So --vae-path path/to/vae in set COMMANDLINE_ARGS=

Reading the discussion It looks to me like the photopea Dev doesn't understand the logic behind this feature, or doesn't realise that SD is another local app with an API...

this isn't a bug, this is just a moan. Train your own face restoration model - gfpgan offers that option, or take the moaning to discussion. not here.

With absolute respect, its a great sounding TTS but it's a horror show spectrally with many tells and wouldn't fool anyone with any sort of credibility. Your choice though.

> @midihex What do you think the right move here is? Would love to hear your thoughts well there's now real-time voice cloners within the past few months that can...

It should be a full path EG C:\Comfy\ComfyUI\main.py, give that a shot

On the comfyui GitHub it mentions some extra steps to make cui work with amd, pip install torch-directml And add --directml in the backend extra Args box But do check...

Might be worth asking for help over on the comfyui hub Also I didn't realise that you'd installed comfyui via the swarm install process so your original link that started...