Implementation:AUTOMATIC1111 Stable diffusion webui Run postprocessing

Knowledge Sources	stable-diffusion-webui
Domains	Image Processing, Pipeline Design, Batch Processing
Last Updated	2026-02-08 00:00 GMT

Overview

Concrete tool for dispatching image postprocessing across single-image, batch-upload, and batch-directory modes provided by stable-diffusion-webui.

Description

run_postprocessing is the main entry point for all image postprocessing in the AUTOMATIC1111 Stable Diffusion WebUI. It accepts images from three distinct input modes (single image, batch upload, batch directory), normalizes them into a common processing loop, runs the full postprocessing script pipeline (upscaling, face restoration, custom scripts) on each image, and saves the results with appropriate metadata.

The function uses an inner generator get_images() to yield (image_or_path, filename) pairs from whichever input mode is selected. Directory mode supports lazy loading where file paths are yielded first and only resolved to PIL Image objects in the main loop. After processing, each image and its extra outputs are saved with embedded PNG info including postprocessing parameters. The function also supports caption file management with configurable actions (prepend, append, keep, or replace) for existing captions.

Usage

Use run_postprocessing when you need to apply the full extras/postprocessing pipeline to images. It is called directly by the Extras tab in the WebUI and by the API endpoint for postprocessing operations.

Code Reference

Source Location

• Repository: stable-diffusion-webui
• File: modules/postprocessing.py
• Lines: 9-129

Signature

def run_postprocessing(
    extras_mode,
    image,
    image_folder,
    input_dir,
    output_dir,
    show_extras_results,
    *args,
    save_output: bool = True
) -> tuple[list, str, str]:

Import

from modules.postprocessing import run_postprocessing

I/O Contract

Inputs

Name	Type	Required	Description
extras_mode	int	Yes	Input mode selector: 0 = single image, 1 = batch upload, 2 = batch directory
image	PIL.Image	No	The single input image (required when extras_mode == 0)
image_folder	list	No	List of uploaded image files or PIL.Image objects (required when extras_mode == 1)
input_dir	str	No	Filesystem path to input directory (required when extras_mode == 2)
output_dir	str	No	Filesystem path to output directory; used when extras_mode == 2 and non-empty, otherwise falls back to configured output path
show_extras_results	bool	Yes	Whether to include processed images in the returned output list when in directory mode (extras_mode == 2)
*args	tuple	Yes	Arguments passed through to the postprocessing script pipeline (ScriptPostprocessingRunner.run)
save_output	bool	No	Whether to save processed images to disk (default: True)

Outputs

Name	Type	Description
outputs	list[PIL.Image]	List of processed images (may be empty in directory mode if show_extras_results is False)
infotext	str	HTML-formatted string containing postprocessing parameters for the last processed image
empty_string	str	Always returns an empty string (reserved for UI compatibility)

Usage Examples

Basic Usage

from modules.postprocessing import run_postprocessing
from PIL import Image

# Single image mode
img = Image.open("input.png")
results, info_html, _ = run_postprocessing(
    extras_mode=0,
    image=img,
    image_folder=None,
    input_dir="",
    output_dir="",
    show_extras_results=True,
    *script_args,
    save_output=True,
)

Batch Directory Mode

# Process all images in a directory
results, info_html, _ = run_postprocessing(
    extras_mode=2,
    image=None,
    image_folder=None,
    input_dir="/path/to/input/images",
    output_dir="/path/to/output",
    show_extras_results=False,
    *script_args,
    save_output=True,
)

Related Pages

Implements Principle

• Principle:AUTOMATIC1111_Stable_diffusion_webui_Postprocessing_input_selection