// image · lllyasviel/ControlNet

ControlNet

ControlNet est un projet de image open source, créé par lllyasviel (lllyasviel/ControlNet sur GitHub). Il compte 34 123 étoiles sur GitHub et 3 019 forks, il est publié sous la licence Apache-2.0 et a été mis à jour pour la dernière fois sur GitHub le 25 février 2024.

ControlNet — the original conditional control for diffusion models: steer Stable Diffusion with edges, depth, pose and scribbles instead of prompts alone, Apache-2.0

★ 34,1 k étoiles⑂ 3,0 k forksApache-2.0mis à jour févr. 2024

Voir sur GitHub ↗

▦ visuel — le README de ce projet ne contient aucune capture intégrable,
voici donc un emplacement réservé. Le guide ci-dessous est complet malgré tout.

// détails du projet

Détails du projet

Dépôt: lllyasviel/ControlNet
Étoiles GitHub: 34 123
Forks: 3 019
Licence: Apache-2.0
Dernière mise à jour sur GitHub: 25 février 2024
Catégorie: Image
Type de fiche: Guide d'installation complet

// fonctionnement

Comment ça marche

Create the environment
Per the README: conda env create -f environment.yaml, then conda activate control.
Download models and detectors
Get the pretrained models and annotator detectors from the project Hugging Face page — SD models go in ControlNet/models, detectors in ControlNet/annotator/ckpts.
Launch a Gradio app
Run the app for your condition, e.g. python gradio_canny2image.py for edge control or gradio_pose2image.py for pose control, and generate in the browser UI.

// fonctionnalités

Pourquoi on l'adore

Control beyond the prompt

Edges, depth, pose, segmentation, scribbles and line art become hard constraints the diffusion model must follow.

Locked + trainable design

The base model stays frozen and safe; only the trainable copy learns the condition, so small datasets suffice.

Nine ready Gradio apps

One app per condition type, each with the relevant detector (Canny thresholds, OpenPose, MiDaS depth and others) built in.

Transferable to community models

Control weights can be transferred to customised Stable Diffusion checkpoints, as documented in the project discussions.

// lancez-le

Lancez-le vous-même

Create the conda env

The README install path.

conda env create -f environment.yaml
conda activate control

Add the weights
Download models and detectors from the Hugging Face page into the folders the README names.
```
https://huggingface.co/lllyasviel/ControlNet
```
Run an app
Example: Canny edge control in the browser.
```
python gradio_canny2image.py
```

// cas d'usage

Ce que vous pouvez créer

Keeping a pose, layout or product shape fixed while restyling an image by prompt
Turning sketches, scribbles and line art into finished illustrations
Recolouring and restyling photos while preserving their structure (HED/depth control)
Research and training of new conditioning types on small datasets

// faq

Questions fréquentes

Who makes ControlNet?

Lvmin Zhang (lllyasviel) and co-authors; this is the official implementation of the ControlNet paper (arXiv 2302.05543).

Is ControlNet free?

Yes — Apache-2.0 licensed. The base Stable Diffusion models it controls carry their own licences, which you should check separately.

Should I use this repo or an extension?

For the original models and research code, this repo. For day-to-day generation, most people use the same ControlNet models through the A1111 extension or ComfyUI nodes, which are also listed on LibreAI.

How does it compare to a well-known alternative?

Against plain Stable Diffusion prompting, ControlNet adds structural guarantees a prompt cannot give. Against IP-Adapter-style methods, ControlNet controls layout and pose from explicit maps, while adapters transfer style/subject from a reference image — many pipelines use both.

How do I get started?

Les autres projets de la catégorie Image avec le plus d'étoiles du catalogue LibreAI — la sélection réaliste d'alternatives si ControlNet ne vous convient pas.

Stable Diffusion Webui★ 165,2 k
Stable Diffusion web UI
ComfyUI★ 136,0 k
The node-graph powerhouse for precise, professional AI image and video generation.
Immich★ 115,5 k
High performance self-hosted photo and video management solution
PhotoCraft★ 73,2 k
PhotoCraft — a clean-room Photoshop reimplementation in pure Rust: layers, masks, adjustment layers, type, vectors, brushes and real PSD files, offline, MIT / Apache-2.0
Fooocus★ 53,3 k
Focus on prompting and generating

// plus en image

Plus en Image

Projets de la catégorie Image proches dans le catalogue, les plus proches en étoiles de ControlNet — continuez à explorer de part et d'autre.

diffusers★ 34,6 k
🤗 Diffusers: State-of-the-art diffusion models for image, video, and audio generation in PyTorch.
czkawka★ 33,9 k
Multi functional app to find duplicates, empty folders, similar images etc.
Real-ESRGAN★ 37,1 k
Real-ESRGAN — practical blind super-resolution for real-world photos and video: restores and upscales degraded images with models trained on pure synthetic data, BSD-3-Clause
Insightface★ 29,9 k
State-of-the-art 2D and 3D Face Analysis Project
aseprite★ 39,9 k
Animated sprite editor & pixel art tool (Windows, macOS, Linux)
InvokeAI★ 28,3 k
A polished creative studio for professionals who want power without the node spaghetti.

Obtenir ControlNet

Gratuit et ouvert : récupérez le code source, mettez une étoile au dépôt et commencez à créer.

Voir sur GitHub ↗← Plus d'outils image