Edit / inpaint image

Edits an input image using a prompt and a mask (inpainting). Only registered when the active image runner_type is an edit/inpaint variant.

Request body

promptstring required

Text prompt describing the desired output image

prompt_2string

Optional secondary prompt routed to the second text encoder (SDXL)

negative_promptstring

Negative prompt describing what to avoid

negative_prompt_2string

Optional secondary negative prompt for the second text encoder

num_inference_stepsinteger

Number of denoising steps

guidance_scalenumber

Classifier-free guidance scale

guidance_rescalenumber

Guidance rescale factor (mitigates over-saturation at high guidance)

seedinteger

Random seed for reproducible sampling

number_of_imagesinteger

Number of images to generate per request

crop_coords_top_leftinteger[]

SDXL crop conditioning [top, left]

timestepsnumber[]

Optional explicit list of timesteps for the scheduler

sigmasnumber[]

Optional explicit list of sigmas for the scheduler

lora_pathstring

Path to a LoRA weights file to load before generation

lora_scalenumber

Scale applied to the LoRA

image_return_format'JPEG' | 'PNG'

Encoding format for the returned base64 image bytes

image_qualityinteger

Quality factor for JPEG encoding (ignored for PNG). Bounds match the Python service's Pydantic constraint (ge=50, le=100).

imagestring

Base64-encoded input image. Required for image-to-image and edits.

maskstring

Base64-encoded mask image. Required for edits/inpainting.

strengthnumber

Image-to-image strength (0..1); higher values diverge further from the input image.

Image(s) edited successfully

imagesstring[] required

Generated images, base64-encoded

generation_timenumber

Wall-clock generation time in seconds (omitted when zero)