Stability Sketch Control

This is the "I can't draw, but I can sketch" node. StabilityControlSketch takes a rough outline - a scribble, a line drawing, a stick-figure composition - and generates a finished image that follows your lines, the same way the scribble/lineart flavors of ControlNet do locally. The difference is that the control model runs on Stability's servers, so you skip the local ControlNet checkpoint, the preprocessor, and the VRAM entirely.

If you've used ControlNet locally, the mental model transfers directly: your sketch is the condition (it decides where things go), and your prompt is the content (it decides what things are). Draw a blobby layout of "big circle on the left, vertical line on the right," prompt "a red balloon next to a lighthouse," and the model does the rest. It's the classic composition-first workflow, and it's genuinely fun - this is one of the easiest ways to iterate on layout without wrestling with a local control stack.

How it works

The node POSTs your image plus prompt to Stability's /v2beta/stable-image/control/sketch endpoint. The one dial that matters is control_strength (0–1, default 0.7): how strictly the output must follow your sketch. At 0.7 the sketch guides composition but the model has room to reinterpret; push it up and the output stays glued to your lines; drop it and the sketch becomes a suggestion.

A practical note on what to feed it: the model reads your image as lines. Flat, high-contrast line art or a clean grayscale sketch works best. Feed it a full-color render and expect it to track the edges of that, not the "intent" of the scene.

Inputs and outputs

• image - required. Your sketch or outline.
• prompt - required. Describe the finished image you want.
• negative_prompt - optional, empty to start.
• control_strength - 0–1, default 0.7. The main dial.
• seed - set for reproducible layouts.
• style_preset - 18 named looks (anime, line-art, cinematic…); none by default.
• output_format - png / jpeg / webp.
• api_key - per-node override.

Output: a single IMAGE, straight into Save Image.

Installing it

Part of the aicu-comfyui-stability-ai-api pack:

cd ComfyUI/custom_nodes
git clone https://github.com/aicuai/aicu-comfyui-stability-ai-api.git
cd aicu-comfyui-stability-ai-api
pip install -r requirements.txt

Or ComfyUI Manager → search "aicu-comfyui-stability-ai-api" → install → restart. You'll need a paid Stability AI key from platform.stability.ai: node api_key field, STABILITY_API_KEY env var, or the auto-generated config.ini in the pack folder.

Gotchas

The API validates inputs before spending your credits: sides must be at least 64px, total pixels between 4,096 and ~9.4MP, and the aspect ratio has to be between 1:2.5 and 2.5:1 - extreme panoramas get rejected. And remember the honest trade-off: local ControlNet is free and its scribble mode is excellent, so this node is really a convenience play for people without the VRAM or patience. The moderation layer is also Stability's, so if your sketch trends toward content the platform blocks, you'll get a 403 - locally that's nobody's business but yours.