The Upscale or Variation (UOV) feature lets you either enlarge an image beyond its original resolution or create alternative versions of it with varying degrees of change. You can use it on any image — not just ones generated by Fooocus — making it a general-purpose post-processing and variation tool.Documentation Index
Fetch the complete documentation index at: https://mintlify.com/lllyasviel/Fooocus/llms.txt
Use this file to discover all available pages before exploring further.
Accessing UOV
Go to Input Image → Upscale or Variation tab. Upload your source image, choose a UOV method, and click Generate.UOV Methods
Fooocus provides six UOV modes:Disabled
UOV is turned off. The input image tab has no effect on generation. This is the default state.
Vary (Subtle)
Generates a variation of the input image with small, subtle differences — different lighting, minor detail changes, or slight compositional shifts — while preserving the overall scene.
Vary (Strong)
Generates a more significant variation. The output will share the general subject and composition of the input but may differ noticeably in style, details, and colors.
Upscale (1.5x)
Upscales the image to 1.5× its original resolution using the Fooocus upscale pipeline, adding fine detail during the process.
Upscale (2x)
Upscales the image to 2× its original resolution. More detailed enhancement pass compared to Fast 2x.
Upscale (Fast 2x)
A faster 2× upscale that skips some of the enhancement steps. Good for quick results when maximum quality isn’t required.
UOV with Enhancement Pipeline
When using the Enhance tab alongside UOV, you can control when the UOV step runs relative to enhancement passes:Processing Order
| Option | Behavior |
|---|---|
Before First Enhancement | UOV runs first, then enhancement passes are applied to the upscaled/varied result |
After Last Enhancement | All enhancement passes complete first, then UOV is applied to the final enhanced image |
Prompt Types for Enhancement UOV
When UOV is used in the enhancement pipeline, you can choose which prompts apply during the UOV step:| Option | Behavior |
|---|---|
Original Prompts | Uses the main generation prompts during the UOV step |
Last Filled Enhancement Prompts | Uses the prompts from the most recently configured enhancement pass |
Describe Feature
The Describe tab (also under Input Image) is a reverse-generation tool: upload an image and Fooocus will generate a text prompt that describes it. This is useful for:- Understanding what prompts could recreate a similar image
- Building a starting prompt for further variation or upscaling
- Learning effective SDXL prompt structures
- Photograph
- Art / Anime
Type:
PhotographOptimized for real-world photography. Generates a prompt describing camera settings, lighting, subject, and scene in photographic terms.Step-by-Step Upscaling Workflow
Select a UOV method
Choose from Vary (Subtle), Vary (Strong), Upscale (1.5x), Upscale (2x), or Upscale (Fast 2x) depending on your goal.
Upscaling uses dedicated upscale models stored in
Fooocus/models/upscale_models/. These are downloaded automatically on first use.