Fooocus is a free image generator built around SDXL that borrows its idea from Midjourney: you type what you want, and the program handles the sampler, the steps, the prompt expansion and the quality tricks. It still has an advanced panel with styles, image prompts, inpainting and outpainting, but you never have to open it. It was created by lllyasviel and ships with a well tuned default model.

Fooocus is the easiest full-featured local app for Windows users with an NVIDIA card. Choose it if you want good pictures without learning samplers and CFG values. It is limited to SDXL-type models and is no longer in active development, so if you want Flux or video, look at Forge, Invoke or ComfyUI.

Aucune installation requise – exécutez les modèles IA dans le cloud

BitVector Prism est la façon la plus simple de commencer : choisissez un modèle, tapez un prompt et générez dans une application web épurée, sans rien configurer. BitVector est aussi disponible sur Discord et sur le web (SpyGlass).

Essayer BitVector

Ce qu’il vous faut

Windows 10 or 11 with an NVIDIA card with 4 GB VRAM or more and 8 GB of system RAM (16 GB recommended; Fooocus swaps to RAM on small cards). AMD on Windows works through DirectML and is slower. Linux and Mac (Apple Silicon) installs are possible from source. The default models are about 7 GB.

Comment obtenir et installer Fooocus

Download the Fooocus Windows package (a .7z file) from the link on the GitHub page and extract it with 7-Zip.
Run run.bat. The first start downloads the default SDXL model (Juggernaut) and the refiner, several gigabytes, then opens the browser.
Linux and Mac: clone the repository, create a Python 3.10 environment, install the requirements and run python entry_with_update.py (see the README for Apple Silicon notes).
To use your own SDXL checkpoints and LoRAs, drop them in models/checkpoints and models/loras, or edit config.txt to point to the folders of another app such as A1111.
Use run_anime.bat or run_realistic.bat for presets with different default models and styles.

Téléchargez uniquement depuis les liens officiels de cette page. Les installateurs trouvés ailleurs peuvent être obsolètes ou modifiés. ModelTrends.ai n’héberge aucun fichier.

Aucune installation requise – exécutez les modèles IA dans le cloud

PirateDiffusion est réservé à Telegram et conçu pour les pros : des milliers de modèles, LoRA et workflows pilotés par commandes de chat, avec génération illimitée à prix fixe.

Essayer PirateDiffusion

Comment utiliser Fooocus

Type your prompt in the box and press Generate. Fooocus makes two images by default at a quality setting that balances speed and detail.
Tick Advanced to open the panel: choose Speed, Quality or Extreme Speed, the aspect ratio and the number of images.
In the Style tab, pick visual styles such as cinematic, anime or photographic; Fooocus adds the matching prompt words for you. Fooocus V2 expands short prompts automatically.
In the Model tab, select a different base checkpoint and up to five LoRAs with their weights. Add the trigger words from the model page to your prompt.
Tick Input Image to use Upscale or Variation, Image Prompt (reference pictures, face swap, canny or depth) or Inpaint or Outpaint to change part of a picture or extend it.

Conseils et problèmes courants

Negative prompts are rarely needed; the defaults already handle common problems. Use them only to remove something specific.
If the first start fails to download the models, download them manually from the links in the README and put them in models/checkpoints.
The History Log link under the images shows every prompt and seed you used, so you can repeat a result.
Check the box Developer Debug Mode for advanced sampler, scheduler and refiner settings when you do want control.

Modèles utilisables dans Fooocus

Fooocus runs SDXL checkpoints and LoRAs, including Pony and Illustrious variants, from this catalog. It does not run Stable Diffusion 1.5, Flux or video models.

Modèles Stable Diffusion XL 1.0

Aucune installation requise – exécutez les modèles IA dans le cloud

Pas de GPU, pas de téléchargement, pas de Python. Ces services exécutent les mêmes modèles open source sur leurs propres serveurs : vous générez depuis un navigateur, Discord ou Telegram en une minute.

Essayer BitVector

Essayer PirateDiffusion

Liens

Ces liens mènent aux sites des éditeurs. ModelTrends.ai n’est pas affilié à eux et n’héberge aucun logiciel.

Autres applis locales

Comment installer ComfyUI

The node-based powerhouse for Stable Diffusion, Flux, Wan video and almost every new open source model. Free, open source, for Windows, Mac and Linux.