Using and Creating Fooocus Presets: Built-in and Custom

Presets are JSON files stored in the presets/ folder that define a complete generation configuration in one place — including which checkpoint and LoRAs to load, default styles, sampler, scheduler, CFG scale, aspect ratio, and even which model files to auto-download if they are missing. Switching a preset instantly reconfigures the entire pipeline, making it easy to jump between photorealistic, anime, and fast-generation workflows without manually adjusting every setting.

Built-in Presets

Fooocus ships with the following presets out of the box:

Preset	Launch Flag	Batch File	Main Checkpoint	Default Performance
`default`	(none)	`run.bat`	`juggernautXL_v8Rundiffusion.safetensors`	Speed
`anime`	`--preset anime`	`run_anime.bat`	`animaPencilXL_v500.safetensors`	Speed
`realistic`	`--preset realistic`	`run_realistic.bat`	`realisticStockPhoto_v20.safetensors`	Speed
`lcm`	`--preset lcm`	—	`juggernautXL_v8Rundiffusion.safetensors`	Extreme Speed
`lightning`	`--preset lightning`	—	`juggernautXL_v8Rundiffusion.safetensors`	Lightning
`playground_v2.5`	`--preset playground_v2.5`	—	`playground-v2.5-1024px-aesthetic.fp16.safetensors`	Speed
`pony_v6`	`--preset pony_v6`	—	`ponyDiffusionV6XL.safetensors`	Speed
`sai`	`--preset sai`	—	`sd_xl_base_1.0_0.9vae.safetensors`	Speed

Missing model files listed in a preset’s checkpoint_downloads, lora_downloads, embeddings_downloads, or vae_downloads sections are automatically fetched from the configured URLs the first time that preset is loaded.

Switching Presets

Browser UI
Command Line

From Fooocus 2.3.0 onward, you can switch presets directly in the web interface without restarting. Open the Preset dropdown (visible in the top section of the UI) and select any preset from the list. Fooocus will load the new configuration immediately.To prevent users from changing presets through the UI, launch with --disable-preset-selection.

Pass --preset PRESET_NAME when launching Fooocus to start with a specific preset:

# Start with the anime preset
python entry_with_update.py --preset anime

# Start with the realistic preset
python entry_with_update.py --preset realistic

# Start with the LCM fast-generation preset
python entry_with_update.py --preset lcm

On Windows you can also use the dedicated batch files:

run_anime.bat
run_realistic.bat

Useful CLI Flags for Presets

Flag	Description
`--preset PRESET`	Load the named preset at startup
`--disable-preset-selection`	Hide the preset dropdown in the browser UI, locking the current preset
`--always-download-new-model`	Automatically download any model files declared in `checkpoint_downloads` / `lora_downloads` that are not present locally

The `previous_default_models` Field

When Fooocus starts and the file named in default_model is not found on disk, it falls back through the list in previous_default_models in order, using the first one it finds. This ensures that users who already have an older version of a model can still load the preset without an immediate download.

"previous_default_models": [
    "juggernautXL_version8Rundiffusion.safetensors",
    "juggernautXL_version7Rundiffusion.safetensors",
    "juggernautXL_v7Rundiffusion.safetensors",
    "juggernautXL_version6Rundiffusion.safetensors",
    "juggernautXL_v6Rundiffusion.safetensors"
]

Preset JSON Format

A preset file is a plain JSON object. All keys are optional — any key omitted from the preset file falls back to the value in config.txt or Fooocus’s built-in defaults. The default.json preset is shown below with annotations explaining each field:

{
    // ── Model selection ──────────────────────────────────────────────────────
    "default_model": "juggernautXL_v8Rundiffusion.safetensors",
    "default_refiner": "None",
    "default_refiner_switch": 0.5,

    // ── LoRA slots: [enabled, filename, weight] ──────────────────────────────
    "default_loras": [
        [true,  "sd_xl_offset_example-lora_1.0.safetensors", 0.1],
        [true,  "None", 1.0],
        [true,  "None", 1.0],
        [true,  "None", 1.0],
        [true,  "None", 1.0]
    ],

    // ── Sampler settings ─────────────────────────────────────────────────────
    "default_cfg_scale": 4.0,
    "default_sample_sharpness": 2.0,
    "default_sampler": "dpmpp_2m_sde_gpu",
    "default_scheduler": "karras",
    "default_performance": "Speed",
    "default_overwrite_step": -1,

    // ── Prompt defaults ───────────────────────────────────────────────────────
    "default_prompt": "",
    "default_prompt_negative": "",

    // ── Style and resolution ──────────────────────────────────────────────────
    "default_styles": [
        "Fooocus V2",
        "Fooocus Enhance",
        "Fooocus Sharp"
    ],
    "default_aspect_ratio": "1152*896",

    // ── Auto-download declarations ────────────────────────────────────────────
    "checkpoint_downloads": {
        "juggernautXL_v8Rundiffusion.safetensors": "https://huggingface.co/lllyasviel/fav_models/resolve/main/fav/juggernautXL_v8Rundiffusion.safetensors"
    },
    "lora_downloads": {
        "sd_xl_offset_example-lora_1.0.safetensors": "https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/resolve/main/sd_xl_offset_example-lora_1.0.safetensors"
    },
    "embeddings_downloads": {},

    // ── Fallback models ────────────────────────────────────────────────────────
    "previous_default_models": [
        "juggernautXL_version8Rundiffusion.safetensors",
        "juggernautXL_version7Rundiffusion.safetensors",
        "juggernautXL_v7Rundiffusion.safetensors",
        "juggernautXL_version6Rundiffusion.safetensors",
        "juggernautXL_v6Rundiffusion.safetensors"
    ]
}

Standard JSON does not allow comments (//). The annotations above are for illustration only — remove them if you copy this block into a real preset file.

Field Reference

Model fields

Field	Type	Description
`default_model`	string	Checkpoint filename (must be in `path_checkpoints`)
`default_refiner`	string	Refiner checkpoint filename, or `"None"`
`default_refiner_switch`	float 0–1	Fraction of total steps at which to swap to refiner
`default_vae`	string	VAE filename, or `"Default (model)"` to use the checkpoint’s built-in VAE
`previous_default_models`	list of strings	Ordered fallback list used when `default_model` is not found

LoRA fields

Field	Type	Description
`default_loras`	list	Each element is `[enabled (bool), filename (str), weight (float)]`
`default_loras_min_weight`	float	Minimum LoRA weight the UI slider allows
`default_loras_max_weight`	float	Maximum LoRA weight the UI slider allows
`default_max_lora_number`	int	Number of LoRA rows shown in the UI

Sampler and generation fields

Field	Type	Description
`default_cfg_scale`	float	Guidance scale (CFG)
`default_sample_sharpness`	float	SAG sharpness enhancement weight
`default_sampler`	string	Sampler key (e.g. `"dpmpp_2m_sde_gpu"`)
`default_scheduler`	string	Noise scheduler (e.g. `"karras"`)
`default_performance`	string	`"Quality"`, `"Speed"`, `"Extreme Speed"`, `"Lightning"`, or `"Hyper-SD"`
`default_overwrite_step`	int	Override step count; `-1` follows the performance mode
`default_overwrite_switch`	int	Override refiner switch step; `-1` follows preset
`default_aspect_ratio`	string	Resolution as `"width*height"`
`default_image_number`	int	Number of images per generation run

Prompt and style fields

Field	Type	Description
`default_prompt`	string	Pre-filled positive prompt
`default_prompt_negative`	string	Pre-filled negative prompt
`default_styles`	list of strings	Active style names loaded at startup

Download fields

Field	Type	Description
`checkpoint_downloads`	object	`{ "filename": "url", … }` — downloaded if file is missing
`lora_downloads`	object	`{ "filename": "url", … }`
`embeddings_downloads`	object	`{ "filename": "url", … }`
`vae_downloads`	object	`{ "filename": "url", … }`

Creating a Custom Preset

Create the JSON file

Create a new file in the presets/ directory. The file name (without .json) becomes the preset name.

# Example
touch presets/mypreset.json

Define your configuration

Add the keys you want to override. You only need to include the fields you are changing — anything omitted falls back to config.txt defaults.

{
    "default_model": "dreamshaper_xl_v2.safetensors",
    "default_performance": "Speed",
    "default_cfg_scale": 5.0,
    "default_sampler": "dpmpp_2m_sde_gpu",
    "default_scheduler": "karras",
    "default_styles": ["Fooocus V2", "Fooocus Cinematic"],
    "default_aspect_ratio": "1024*1024",
    "checkpoint_downloads": {
        "dreamshaper_xl_v2.safetensors": "https://huggingface.co/your-repo/resolve/main/dreamshaper_xl_v2.safetensors"
    },
    "lora_downloads": {},
    "embeddings_downloads": {}
}

Launch with your preset

Pass --preset mypreset on the command line:

python entry_with_update.py --preset mypreset

Or select it from the Preset dropdown in the browser UI without restarting.

To quickly build a preset from an existing configuration, set REWRITE_PRESET = True in modules/config.py, launch Fooocus with --preset yourpreset, and the current live config will be serialized to presets/yourpreset.json automatically. Remember to set REWRITE_PRESET back to False afterwards.

Get Started

Installation Guides

Core Features

Advanced Usage

Deployment & Operations

Using and Creating Fooocus Presets: Built-in and Custom

Built-in Presets

Switching Presets

Useful CLI Flags for Presets

The `previous_default_models` Field

Preset JSON Format

Field Reference

Creating a Custom Preset

Build docs developers (and LLMs) love

Get Started

Installation Guides

Core Features

Advanced Usage

Deployment & Operations

Documentation Index

​Built-in Presets

​Switching Presets

​Useful CLI Flags for Presets

​The previous_default_models Field

​Preset JSON Format

​Field Reference

​Creating a Custom Preset

Build docs developers (and LLMs) love

Built-in Presets

Switching Presets

Useful CLI Flags for Presets

The `previous_default_models` Field

Preset JSON Format

Field Reference

Creating a Custom Preset