A working example, with the files

Read AUTOMATIC1111 parameters from a PNG

An original PNG can carry its prompt and generation settings in a parameters record. Open the file in the saved-prompt extractor to copy the text and see the fields that this copy actually contains.

By isGenAI · Published September 20, 2026

Extract a saved prompt Try the example files

Follow the steps

Download the parameters example below, or choose an original PNG you are allowed to inspect.
Open the saved-prompt extractor and choose the file. Read the positive prompt first, then open the saved details.
Compare the negative prompt, recorded model and generation settings. Download the JSON result if you need to keep the values together.

What the example demonstrates

We added one known parameters record to an image drawn in code. The expected positive prompt starts “Controlled example: a paper boat”. The negative prompt is “text, blur”. Seed 42, 20 steps and CFG scale 7 are values in the record, not settings inferred from the picture.

The example model name is demonstration-only. These pixels were not generated by that model. You can use this fixture to check whether another reader returns the same saved values.

Use the original file

AUTOMATIC1111 documents PNG generation information and a PNG Info workflow. For isGenAI, choose the downloaded file itself. Copying a preview into another app can produce a different file, even when it looks the same.

If the reader finds no supported record, try the original export. Repeating the same check does not reconstruct data that is absent from this copy.

What to keep with the result

Save the file together with its source and the extracted JSON. A recorded model name is a label; it does not provide model weights or verify how the pixels were created. For a signed provenance record, use the separate Content Credentials checker.

Download · Run · Compare

Try the verified example

We drew the picture by code and attached controlled records. The saved prompts and model names are demonstration data. These files test record reading; no image generator, screenshot app or social network was used to create these test results.

parameters-tEXt.png

Observed output

{
  "format": "automatic1111",
  "prompt": "Controlled example: a paper boat on a blue table",
  "negative_prompt": "text, blur",
  "recorded_model": "demonstration-only",
  "settings": {
    "seed": "42",
    "steps": 20,
    "sampler": "Euler",
    "cfg_scale": 7,
    "width": 1024,
    "height": 768
  }
}

Seed values in the JSON are strings so that large integers retain their precision. A null result here means no supported saved prompt was returned.

Hashes and reproduction

parameters-tEXt.png · 13,371 bytese76d2af4e6c273e1ba8ca6cf2b5b7885ceda2321bce893f0bb5f9e64ace17c9f

Fixtures and observations are CC0-1.0. Download all observations and versions or read the procedure. The complete Node.js reproduction script (requires sharp) creates the same test inputs and compares bytes and pixels. The isGenAI source checkout additionally verifies saved fields with npx tsx scripts/build-guide-lab.ts.

For the separate export study, explore the 18 metadata-retention cases.

Sources and method

The linked primary sources describe the formats. The downloadable fixture outputs were checked with automated assertions against the current isGenAI reader. These guides were prepared with AI assistance and reviewed against those examples.

Report an error or unsupported example. For a signed record, inspect Content Credentials.