(on Wikipedia)
https://github.com/CompVis/stable-diffusion
https://ommer-lab.com/research/latent-diffusion-models/
https://stability.ai/
- These guys use Stable Diffusion: https://creator.nightcafe.studio/
-
NSFW and waifu was unstablediffusion.net
-
-
2023-04-12 - I used Easy Diffusion
Stuff ∞
Recommended budget video cards to use for Stable Diffusion ∞
This knowledge is decent as of 2023-04-12, but development and improvement are ongoing!
-
- As of 2023-04-12 efforts for NVIDIA RDNA exist but those perform worse. It is unknown if that type of card will ever outperform Nvidia in terms of ROI.
-
The third digit to be high
- e.g. 2060 is better than 3050
-
8G or more of VRAM
- The higher the better, this is more important than that third digit.
Examples:
- 2050 8G
- 2060 8G is widely available and cost effective.
- 3050 12G
- 2060 12G is a great choice but a significant cost increase.
-
3060 12G
-
- "Ti" carts are nice, but it is too expensive to consider.
-
Used / refurbished cards save a lot of money, but be aware that those cards may have been used for crypto mining and at worse their fans might be wearing down. Also, warranty matters.
Random knowledge I don't understand ∞
Stuff here: https://pastebin.com/9RmqDhSm
Prompts:
- https://rentry.org/hdgpromptassist
-
More TODO
Inpainting/Outpainting:
Upscaling:
Wildcards: https://rentry.org/NAIwildcards
Create Embeddings, Hypernetworks, LoRAs:
- https://rentry.org/simplified-embed-training
- https://rentry.org/HNSpeedrun
- https://rentry.org/lora_train
Animation:
ControlNet:
Where to get models ∞
Alternatives / other implementations ∞
See AI Art
- Easy Diffusion
- https://invoke-ai.github.io/InvokeAI/
- NVIDIA: https://rentry.org/voldy
- CPU: https://rentry.org/cputard | https://rentry.org/webui-cpu
- AMD (and more): https://rentry.org/sdg-link
-
Alternative UI (Node-based): https://github.com/comfyanonymous/ComfyUI
Last updated 2023-09-21 at 14:56:41