Conversation

Notices

1. Vivi Nella Verita :Debian_logo: (verita84@poster.place)'s status on Friday, 03-Feb-2023 05:19:33 JST Vivi Nella Verita :Debian_logo:
   Stable Diffusion not working man
   •  (mint@ryona.agency)'s status on Friday, 03-Feb-2023 05:19:32 JST 

     in reply to
     @verita84 What the fuck are you using instead of industry standard webui?
   • Vivi Nella Verita :Debian_logo: (verita84@poster.place)'s status on Friday, 03-Feb-2023 05:21:43 JST Vivi Nella Verita :Debian_logo:

     in reply to

     • 
     @mint
     
     No idea what I am doing. Following /g/
     
     https://rentry.org/cputard
      likes this.
   •  (mint@ryona.agency)'s status on Friday, 03-Feb-2023 05:23:14 JST 

     in reply to
     @verita84 I see. As far as I remember, webui has CPU-only mode as well, but it's slow as all hell. What GPU do you have?
   • Vivi Nella Verita :Debian_logo: (verita84@poster.place)'s status on Friday, 03-Feb-2023 05:33:08 JST Vivi Nella Verita :Debian_logo:

     in reply to

     • 
     @mint
     
     Intel, it's an 2019 Thinkpad
   •  (mint@ryona.agency)'s status on Friday, 03-Feb-2023 05:33:08 JST 

     in reply to
     @verita84 Yeah, not sure if it can be accelerated. Best support is on Nvidia (both Win/Lin), followed by RDNA+ AMD (Linux-only) and Apple M1/M2. I think there was some Vulkan/ONNX-based version, but if not, you could go with CPU-only, though it would be painful.
     https://rentry.org/nai-speedrun - webui quick start guide focuses on getting leaked NovelAI, but you could substitute it with whatever other model you have on hand.
   •  (mint@ryona.agency)'s status on Friday, 03-Feb-2023 05:35:29 JST 

     in reply to

     • 
     @verita84 Yeah, there is an ONNX version, but no one tested it on Intel integrated GPUs, and it's *really* basic, you even have to supply your own script with prompts.
     https://rentry.org/ayymd-stable-diffustion-v1_4-guide