GPU:
- Recommend: Any Nvidia Card from GTX 1070 or newer. Also Valid for same era Quadro/Tesla card or newer
- Minimum: 1060 3gb with additional start parameters for Automatic1111 to work with low VRAM
- AMD: currently not really compatible. AMD fork of Automatic1111 will be tested and noted here on a later date
CPU:
- Any 4 Core CPU should work. CPU isn't that utilized by the AI Generator
RAM:
- 16GB or more. Is used as cache for models
Storage:
- 64GB or more. Only holds models, lora, embeddings and such. Generated images are not saved.
- SSD storage highly recommended for way faster model loading
How to become a worker:
-
Install stable-diffusion-webui and run with
--api
GitHub: Automatic1111
Try to generate a picture to see if it works.
If it works add--no-download-sd-modelto the run command and delete v1-5-pruned-emaonly.safetensors from your models/Stable-diffusion folder -
Download a furry model, for example bb95:
CivitAI: BB95 Furrymix
Put it in your SD models/Stable-diffusion, select it in webui, try to generate something furry. -
Install ngrok, and make it use a static domain, then give us the ngrok link.
Ngrok: Free static domains -
Add basic authentication to ngrok, then give us your login/password:
Ngrok: Authentication with Ngrok -
Install syncthing to sync models, loras and embedding folders so every worker uses the same models (ask other's device IDs in admin chat):
Syncthing
