aiunivers/utopianpony-inpainting

Don't use yet! In development.

Public

43 runs

Run aiunivers/utopianpony-inpainting with an API

Use one of our client libraries to get started quickly. Clicking on a library will take you to the Playground tab where you can tweak different inputs, see the results, and copy the corresponding code to use in your own project.

Input schema

The fields you can use to run this model with an API. If you don't give a value for a field its default value will be used.

Field	Type	Default value	Description
image	string		RGB image to inpaint
mask	string		Mask: white=replace (edit), black=keep (preserve original)
prompt	string		Positive prompt
negative_prompt	string	low quality, blurry, distorted, watermark, logo, text, artifacts, chromatic aberration	Negative prompt (defaults include artifact suppression)
steps	integer	40 Min: 1 Max: 150	Inference steps
guidance_scale	number	7.5 Max: 30	CFG guidance scale
scheduler	string	dpm	Scheduler (euler\|euler_a\|dpm)
strength	number	0.55 Min: 0.05 Max: 1	Inpaint strength (higher = more change; lower for subtle edits)
seed	integer		Random seed (blank = random)
enable_safety_checker	boolean	False	Enable safety checker (off by default)
guidance_rescale	number	0 Max: 1	Guidance rescale to mitigate over-saturation / burn from high CFG (0 = off)
mask_threshold	integer	0 Max: 255	Threshold (0-255) to binarize mask after optional inversion (0 = skip)
mask_feather	integer	0 Max: 128	Feather (Gaussian blur radius) applied AFTER binarization for softer edges (0 = none)
invert_mask	boolean	False	Invert mask BEFORE threshold (use if your white/black semantics are reversed)
karras	boolean	True	Use Karras sigmas for smoother scheduler noise (where supported)
preserve_composition	boolean	True	Pass SDXL original/target size hints to better preserve aspect & framing
lora_paths	string		Comma-separated paths (local or under weights/) to LoRA .safetensors for SDXL (applied sequentially)
lora_scale	number	0.8 Max: 2	Base scale for all LoRAs (individual scaling not yet exposed)
upscale_factor	number	1 Min: 1 Max: 2.5	Optional latent second-pass upscale factor ( >1 enables high-res re-inpaint pass)
upscale_steps	integer	15 Min: 5 Max: 60	Steps for high-res second pass (only used if upscale_factor>1)

{
  "type": "object",
  "title": "Input",
  "required": [
    "image",
    "mask",
    "prompt"
  ],
  "properties": {
    "mask": {
      "type": "string",
      "title": "Mask",
      "format": "uri",
      "x-order": 1,
      "description": "Mask: white=replace (edit), black=keep (preserve original)"
    },
    "seed": {
      "type": "integer",
      "title": "Seed",
      "x-order": 8,
      "nullable": true,
      "description": "Random seed (blank = random)"
    },
    "image": {
      "type": "string",
      "title": "Image",
      "format": "uri",
      "x-order": 0,
      "description": "RGB image to inpaint"
    },
    "steps": {
      "type": "integer",
      "title": "Steps",
      "default": 40,
      "maximum": 150,
      "minimum": 1,
      "x-order": 4,
      "description": "Inference steps"
    },
    "karras": {
      "type": "boolean",
      "title": "Karras",
      "default": true,
      "x-order": 14,
      "description": "Use Karras sigmas for smoother scheduler noise (where supported)"
    },
    "prompt": {
      "type": "string",
      "title": "Prompt",
      "x-order": 2,
      "description": "Positive prompt"
    },
    "strength": {
      "type": "number",
      "title": "Strength",
      "default": 0.55,
      "maximum": 1,
      "minimum": 0.05,
      "x-order": 7,
      "description": "Inpaint strength (higher = more change; lower for subtle edits)"
    },
    "scheduler": {
      "type": "string",
      "title": "Scheduler",
      "default": "dpm",
      "x-order": 6,
      "description": "Scheduler (euler|euler_a|dpm)"
    },
    "lora_paths": {
      "type": "string",
      "title": "Lora Paths",
      "default": "",
      "x-order": 16,
      "description": "Comma-separated paths (local or under weights/) to LoRA .safetensors for SDXL (applied sequentially)"
    },
    "lora_scale": {
      "type": "number",
      "title": "Lora Scale",
      "default": 0.8,
      "maximum": 2,
      "minimum": 0,
      "x-order": 17,
      "description": "Base scale for all LoRAs (individual scaling not yet exposed)"
    },
    "invert_mask": {
      "type": "boolean",
      "title": "Invert Mask",
      "default": false,
      "x-order": 13,
      "description": "Invert mask BEFORE threshold (use if your white/black semantics are reversed)"
    },
    "mask_feather": {
      "type": "integer",
      "title": "Mask Feather",
      "default": 0,
      "maximum": 128,
      "minimum": 0,
      "x-order": 12,
      "description": "Feather (Gaussian blur radius) applied AFTER binarization for softer edges (0 = none)"
    },
    "upscale_steps": {
      "type": "integer",
      "title": "Upscale Steps",
      "default": 15,
      "maximum": 60,
      "minimum": 5,
      "x-order": 19,
      "description": "Steps for high-res second pass (only used if upscale_factor>1)"
    },
    "guidance_scale": {
      "type": "number",
      "title": "Guidance Scale",
      "default": 7.5,
      "maximum": 30,
      "minimum": 0,
      "x-order": 5,
      "description": "CFG guidance scale"
    },
    "mask_threshold": {
      "type": "integer",
      "title": "Mask Threshold",
      "default": 0,
      "maximum": 255,
      "minimum": 0,
      "x-order": 11,
      "description": "Threshold (0-255) to binarize mask after optional inversion (0 = skip)"
    },
    "upscale_factor": {
      "type": "number",
      "title": "Upscale Factor",
      "default": 1,
      "maximum": 2.5,
      "minimum": 1,
      "x-order": 18,
      "description": "Optional latent second-pass upscale factor ( >1 enables high-res re-inpaint pass)"
    },
    "negative_prompt": {
      "type": "string",
      "title": "Negative Prompt",
      "default": "low quality, blurry, distorted, watermark, logo, text, artifacts, chromatic aberration",
      "x-order": 3,
      "description": "Negative prompt (defaults include artifact suppression)"
    },
    "guidance_rescale": {
      "type": "number",
      "title": "Guidance Rescale",
      "default": 0,
      "maximum": 1,
      "minimum": 0,
      "x-order": 10,
      "description": "Guidance rescale to mitigate over-saturation / burn from high CFG (0 = off)"
    },
    "preserve_composition": {
      "type": "boolean",
      "title": "Preserve Composition",
      "default": true,
      "x-order": 15,
      "description": "Pass SDXL original/target size hints to better preserve aspect & framing"
    },
    "enable_safety_checker": {
      "type": "boolean",
      "title": "Enable Safety Checker",
      "default": false,
      "x-order": 9,
      "description": "Enable safety checker (off by default)"
    }
  }
}

Output schema

The shape of the response you’ll get when you run this model with an API.

Schema

{
  "type": "string",
  "title": "Output",
  "format": "uri"
}