skills.rest
ShinyGua
ShinyGuaCOMMUNITY
Explore all skills by this author →
GitHub

transferring-styles

Compiles a Visual Intent Object into a ComfyUI img2img style-transfer prompt JSON.

Updated Apr 8, 2026
Search Tags:#image generation#sdxl#comfyui#style transfer#denoise#img2img#prompt json
One-click install
npx skills add https://github.com/ShinyGua/ComfyUI-Agent --skill transferring-styles-shinygua
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: transferring-styles
Source: https://github.com/ShinyGua/ComfyUI-Agent/tree/main/.claude/skills/transferring-styles
Command: npx skills add https://github.com/ShinyGua/ComfyUI-Agent --skill transferring-styles-shinygua

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Applying a new artistic style to an existing image in ComfyUI requires hand-editing a workflow graph with the right denoise, prompts, and image inputs. This Skill automates that by patching a proven low-denoise img2img template with the caller's style prompt and parameters, producing a ready-to-execute ComfyUI prompt JSON. ## Core Features & Use Cases - Template+patch compilation: Reads a Visual Intent Object (VIO) from stdin and patches a hand-authored SDXL img2img template with the style prompt, negative prompt, source image, and sampler settings. - Face-preservation awareness: Automatically lowers denoise to 0.45 when a preserve:["face"] constraint is present, while still honoring explicit denoise overrides. - Deterministic CLI contract: Emits the compiled ComfyUI prompt JSON on stdout and structured error codes on failure, with an eval suite covering basic, face-preserve, and override cases. - Use Case: A planner emits a step with goal=transfer_style and prompt "van gogh" for image a.png; the Skill outputs a complete ComfyUI graph with the style prompt injected and denoise tuned to preserve composition. ## Quick Start Ask the agent to transfer the style of an existing image, for example: apply a film noir style to my photo while keeping the composition unchanged.

Frequently Asked Questions about transferring-styles

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I apply a style to an existing image in ComfyUI?▼

Use an img2img workflow with a low denoise value so the composition is preserved while the style shifts. This Skill compiles your style prompt and source image into a ready-to-run ComfyUI prompt JSON based on an SDXL img2img template.

What denoise value should I use for img2img style transfer?▼

The template defaults to 0.55, which preserves structure while shifting style. When a face-preservation constraint is present, the Skill lowers denoise to 0.45, and any explicit denoise in params overrides both defaults.

How do I preserve faces during AI style transfer?▼

Pass a preserve constraint containing "face" in the Visual Intent Object. The Skill responds by lowering denoise to 0.45; actual face enforcement is handled downstream by a separate preserving-faces Skill.

What input format does the style transfer compiler expect?▼

It reads a JSON Visual Intent Object from stdin with fields like goal, prompt, negative_prompt, inputs.image, params, and model.checkpoint. It exits with a structured error if the JSON is invalid or the goal is not transfer_style.

Can I override sampler settings like steps, cfg, or seed?▼

Yes. The params object accepts steps, cfg, sampler_name, scheduler, denoise, and seed, which are patched directly into the KSampler node. A negative seed is replaced with a time-based random seed.

Related Skills

Explore More Design & Creative Skills→
autonomy-cloudautonomy-cloud
2

image-studio

Generate and edit images from natural-language prompts using the media_generate_image tool.

Official
Intermediate
dawsonblockdawsonblock

cli-anything-comfyui

Manage ComfyUI workflow queues and download generated images via REST API.

Community
Intermediate
fuyucnfuyucn

ref-to-image

Generate new scenes while preserving facial identity using reference-based anchoring.

Community
Advanced
devMoezdevMoez
1

stable-diffusion-image-generation

Generate and transform images via Stable Diffusion pipelines with ControlNet and LoRA.

Community
Intermediate