File Format

How to read PNG text chunks (tEXt, zTXt, iTXt)

AI tools store prompts in PNG text chunks — tEXt (plain), zTXt (compressed), and iTXt (UTF-8). Reading them is how prompt recovery from PNGs actually works.

Try it now: AI Metadata Inspector

Free, no login — upload an image and read its embedded metadata in your browser.

What to know

PNG files can carry named text chunks alongside the image data. There are three kinds: tEXt for plain Latin-1 text, zTXt for compressed text, and iTXt for compressed or uncompressed UTF-8. AI tools use these to store the prompt and settings — ComfyUI writes "workflow" and "prompt" chunks; many Stable Diffusion interfaces write a "parameters" chunk.

Reading a chunk means opening the PNG with a tool that lists its text chunks by key and value. Standard EXIF viewers often miss them because they only read camera fields; an AI-aware metadata reader surfaces the chunks and decompresses zTXt/iTXt for you.

How to recover it

  1. 1

    Open the PNG in an AI-aware metadata reader

    Upload it to the AI Metadata Inspector, which lists tEXt/zTXt/iTXt chunks and decompresses them — where generic EXIF viewers show nothing.

  2. 2

    Identify the prompt-bearing chunk

    Look for keys like "prompt", "workflow", or "parameters"; these hold the generation data for the common AI tools.

  3. 3

    Copy out the decoded text

    Read the decoded chunk value to recover the prompt and settings, then save them somewhere durable.

Reading chunks one file at a time is fine for one image; a catalog that parses them on import makes every prompt searchable across your whole library.

Frequently asked questions

Frequently Asked Questions

Quick answers to the most common questions.

Related recovery guides

Never lose a prompt again

Numonic extracts the prompt, model, seed, and settings from every AI image on import and keeps them searchable — so recovery stops being a rescue mission and the prompt is always attached to the work.