Re:Show HN: Shortbread App – AI-powered, romantic comics for...
by webwielder2 on Tuesday June 04, 2024 @06:18PM
(#40579541)
Is there a standard, user-friendly way of generating consistent AI images yet? Or is it just starting to emerge in patches like the tool here? Because DALL-E for example is a complete crapshoot; it has no ability for fine-tuning or persistence across images/iterations.
[ Parent
| on HN ]
Re:Show HN: Shortbread App – AI-powered, romantic comics for...
by pjgalbraith on Tuesday June 04, 2024 @07:51PM
(#40580104)
They used LoRa. Alternatively you can use IP-Adapter and ControlNet, there are models for face, style, composition, lighting, etc... Search for "ComfyUI IP-Adapter workflow".
Re:Show HN: Shortbread App – AI-powered, romantic comics for...
by Fengjiao on Tuesday June 04, 2024 @07:40PM
(#40580030)
We use LoRA models with stable diffusion models. They can be trained quickly with minimal 30-50 images of your subject. (https://civitai.com/articles/2099/lora-models-and-how-to-use...)