Image Generation
These packages generate still images from a prompt; fo-fal also edits from a reference image. Each renders the catalogue entries marked for its platform and supplies their options and output receipts; the host imports finished images as ordinary project media. Choosing a different backend does not create a new plugin type.
fo-cui renders Z-Image Turbo on ComfyUI, with model stacks and GPU constraints. fo-fal takes its models and options from the catalogue, including FLUX.2 klein 4B and GPT Image 2.5 editing for character sheets. fo-openai renders GPT Image 2.5 from text only. See each guide for supported settings and cost behavior.
Subtypes
Section titled “Subtypes”- Text to image — implemented
- Image to image / reference-guided editing — implemented by fo-fal and fo-google