ControlNet

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

34091 stars · 3020 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

ControlNet is a neural network architecture that adds conditional control (edges, poses, depth, etc.) to pretrained text-to-image diffusion models like Stable Diffusion. It uses locked copies of original weights plus trainable copies with zero-initialized convolutions, enabling fine-tuning on small datasets without destroying the base model.

Use cases

• control stable diffusion output with edge maps or sketches
• generate images conditioned on human pose skeletons
• guide image generation with depth maps
• fine-tune diffusion models on small image-pair datasets
• add spatial conditioning to text-to-image generation
• use canny edge or segmentation maps to steer image synthesis

When to choose

• you need precise spatial or structural control over diffusion model outputs
• you want to condition Stable Diffusion on images like poses, edges, or depth
• you want to train a control adapter on a small dataset without breaking the base model

When to avoid

• you only need plain text-to-image generation without extra conditioning
• you need a maintained application with a UI rather than a research codebase
• you work with non-diffusion generative models

Facets

library · maturity maintenance

deep-learning machine-learning image-processing stable-diffusion deep-learning image-processing artificial-intelligence python windows diffusion-models stable-diffusion conditional-generation text-to-image zero-convolution image-to-image neural-network-architecture gpu linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="lllyasviel/ControlNet")