ControlNet
ControlNet is an add-on that conditions a diffusion model on a reference structure - such as a pose skeleton, edge map or depth map - so you control composition precisely, not just with words.
Prompts are great at describing content but weak at pinning down exact layout. ControlNet solves that. It is a neural network that plugs into a diffusion model and feeds it an extra structural hint - derived from a reference image - so the output follows that structure while your prompt fills in the style and content.
Common control types
- Pose (OpenPose): copy a person's exact body pose into a new character or scene.
- Canny / line art: follow the edges of a reference so the composition matches precisely.
- Depth: preserve the 3D layout and spatial arrangement of a scene.
- Scribble: turn a rough doodle into a finished image that respects your lines.
Why it matters
ControlNet gives you reproducible composition control that prompting alone cannot. It is essential when you need a specific pose, a consistent layout across a series, or a faithful sketch-to-render workflow. It pairs naturally with image-to-image and inpainting for fine-grained editing.
Try it in the generator
Put controlnet to work right now - free daily generations, commercial license included.
Related terms
- Image-to-imageImage-to-image (img2img) is the AI workflow that transforms an existing picture according to your prompt, keeping some of the original structure instead of generating from scratch.
- InpaintingInpainting regenerates only a selected (masked) region of an image while leaving the rest untouched - useful for removing objects, fixing details or replacing part of a scene.
- Diffusion modelA diffusion model is the type of AI that powers most modern image generators. It learns to turn random noise into a coherent image by reversing a step-by-step noising process.
- LoRAA LoRA (Low-Rank Adaptation) is a small add-on file that teaches a base model a specific style, character, object or concept - without retraining the entire model.