Inpaint Processor

Overview

The Inpaint Processor fills in or modifies selected regions of an image based on a mask and text prompts. It supports multiple models via the model parameter and accepts additional model‐specific options through options_json. Additionally, the processor offers pre‐processing and post‐processing resize options to control the resolution of the image fed to the inpainting API and the final output quality.


Example Usage

- processor:
    processor: "inpaint"
    inputs:
      source: "input/input.png" # Filepath to the source image to be inpainted.
      mask: "mem/mask.png" # Filepath to the mask image defining the inpainting region.
    params:
      prompt: "a single vehicle on a solid white background"
      negative_prompt: "walls, cars"
      model: "sdxl" # The model to use (default is "sdxl"; alternative: "stable diffusion").
      options_json: "{}" # JSON string with additional model-specific options.
      pre_width: "1024" # Optional. If >0, resize the source image and mask to this width before inpainting.
      pre_height: "768" # Optional. If >0, resize the source image and mask to this height before inpainting.
      post_width: "0" # Optional. If >0, resize the inpainted result to this width after processing.
      post_height: "0" # Optional. If >0, resize the inpainted result to this height after processing.
    outputs:
      target: "output/output.png" # Filepath where the final inpainted image will be saved.

Inputs

Key Required Description
source Yes Filepath to the source image to be inpainted.
mask Yes Filepath to the mask image defining the inpainting region.

Params

Key Type Default Description
prompt str N/A The text prompt describing the desired content for the inpainted area.
negative_prompt str N/A A text prompt specifying undesired elements to avoid in the inpainted output.
model str "sdxl" The model to use for inpainting. Currently supported models include "sdxl" (default) and "stable diffusion".
options_json str "{}" A JSON string containing additional model-specific options.
pre_width int 0 Optional. If greater than 0, the source image and mask will be resized to this width before inpainting.
pre_height int 0 Optional. If greater than 0, the source image and mask will be resized to this height before inpainting.
post_width int 0 Optional. If greater than 0, the inpainted result will be resized to this width after processing.
post_height int 0 Optional. If greater than 0, the inpainted result will be resized to this height after processing.

Outputs

Key Required Description
target Yes Filepath where the final inpainted image (saved as a JPEG converted to PNG) will be saved.