"Configure and troubleshoot the extension's DemoFusion panel for
Scanned 9/8/2026
Install to Claude Code
npx -y skills add VectorSpaceLab/AREX-Skill --skill demofusion --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Demofusion?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/vectorspacelab-demofusion)More formats (shields.io, HTML) on the badges page.
---
name: demofusion
description: "Configure and troubleshoot the extension's DemoFusion panel for
staged multi-scale upscaling, local/global windows, random jitter, mixture
mode, noise inversion, and ControlNet/StableSR interop."
disable-model-invocation: true
metadata:
disco-role: operating
license: NOASSERTION
---
# DemoFusion
Use this sub-skill for the extension's separate **DemoFusion** panel. DemoFusion performs staged multi-scale denoising/upscaling with local tiled windows and global views. It is not the same route as the Tiled Diffusion panel.
## When to read
Read this when the user asks about:
- enabling DemoFusion in AUTOMATIC1111;
- staged x2/x3/x4 upscale behavior;
- latent window size, overlap, local/global batch sizes;
- random jitter, mixture mode, Gaussian filter, cosine scales, or sigma;
- DemoFusion with img2img, keep-input-size, or noise inversion;
- why the panel says not to open DemoFusion with Tiled Diffusion;
- DemoFusion `UniPC` incompatibility, speed, OOM, or image-size infotext behavior.
For normal Tiled Diffusion/MultiDiffusion/Mixture of Diffusers, route to [tiled-diffusion](../tiled-diffusion/SKILL.md). For VAE encode/decode OOM, route to [tiled-vae](../tiled-vae/SKILL.md).
## Core operating facts
- The panel title is **DemoFusion**.
- The UI text warns: do not open DemoFusion with Tiled Diffusion.
- The only method choice is `DemoFusion`.
- DemoFusion asserts that it is not compatible with `UniPC`.
- Scale factor is used as an integer staged upscale count.
- Window size defaults to 128 and is bounded by the latent canvas size.
- Window overlap defaults to 64.
- Local tile batch size and global window batch size both default to 4.
- Random jitter defaults on.
- Cosine scale defaults are `3`, `1`, and `1`; sigma defaults to `0.6`.
- Txt2img exposes denoising strength for the substage with default `0.85`.
## References
- [Workflows](references/workflows.md): basic txt2img/img2img DemoFusion setup, staged upscale controls, and parameter meanings.
- [Troubleshooting](references/troubleshooting.md): conflict with Tiled Diffusion, sampler incompatibility, stochastic jitter, OOM, and slow staged runs.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!