logo Toolmaxy
New

AI Image Prompt Metadata Viewer

View and extract prompts, seeds, models, and generation parameters embedded in PNG images generated by Midjourney, Stable Diffusion, ComfyUI, and more.

Drop your AI images here

PNG, WebP, JPEG from Stable Diffusion, ComfyUI, Midjourney

Select Images

What is the AI Image Prompt Metadata Viewer?

The AI Image Prompt Metadata Viewer is a lightweight client-side helper built to read the hidden parameter values written into images by popular artificial intelligence generators. When tools like Stable Diffusion, ComfyUI, Fooocus, or NovelAI produce an image, they automatically save your settings — including your positive and negative prompts, seeds, sampler names, CFG values, and even the complete visual nodes workflow.

This utility parses these files locally. You can use it as a stable diffusion metadata viewer to retrieve the prompt settings of images you generated weeks ago, or to see how another artist structured their generation setup. This makes it easy to see prompt from image files or to retrieve a lost node graph configuration without any complex installation.

Since all operations happen inside your browser, the tool operates with zero server uploads, keeping your prompts and images completely private. You can drag and drop multiple images at once to quickly inspect parameters in a clean side-by-side dashboard. If you want to delete these parameters and protect your privacy before sharing your work, you can use our Image Metadata Viewer & Remover to strip all EXIF and textual chunks.

How to Extract Prompts from PNG Files

  1. Upload your image: Drag and drop your AI-generated PNG, WebP, or JPEG files directly onto the upload zone, or click the selection button to choose files from your local storage.
  2. Choose the active file: If you selected multiple files, a sidebar list will appear. Click on any thumbnail in the sidebar list to load its generation parameters in the main panel.
  3. Inspect prompts and parameters: Read the positive prompt and negative prompt in the formatted text fields, and copy them with one click. View key settings such as CFG scale, seed, model, size, and sampler in the parameters panel.
  4. Export ComfyUI Workflows: If a ComfyUI image is loaded, a dedicated tab will appear allowing you to view and download the raw workspace JSON file. You can drag this file directly back into ComfyUI to reload the nodes.
  5. View raw chunks: Select the "Raw Metadata Chunks" tab to inspect the unmodified binary properties found inside the image headers.

Supported AI Generators & Metadata Chunks

Generator File Type Metadata Location Extracted Fields
Stable Diffusion WebUI (A1111) PNG / JPEG / WebP tEXt "parameters" / EXIF UserComment Prompt, Negative, Seed, CFG, Sampler, Model, Steps, Size
ComfyUI PNG iTXt "prompt" & "workflow" Complete node graph parameters & loaded models
NovelAI PNG tEXt "Description" & "Comment" Prompt, Undesired Content (Negative), Seed, Sampler, Steps
Fooocus PNG tEXt "Fooocus" Full prompt JSON properties, samplers, and seeds
Midjourney (Downloaded) PNG / JPEG iTXt "Description" / EXIF Title Core positive prompt text descriptions

Technical Details: How We Extract Prompt Metadata

The tool works by analyzing the raw binary stream of the selected image files. PNG files are organized into sequential blocks called chunks. Each chunk contains a length header, a 4-letter type identifier (such as `tEXt` or `iTXt`), the payload, and a cyclic redundancy check (CRC) footer. The browser reads these blocks using a DataView interface.

For compressed chunks (like `zTXt` or compressed `iTXt` keys), the application pipes the compressed bytes through the browser's native DecompressionStream API using a deflate filter, unpacking the textual JSON strings directly in memory. This eliminates the need for large external unzip scripts.

JPEG and WebP files place metadata inside EXIF APP1 or RIFF EXIF chunks. The parser navigates the TIFF directory trees within these blocks to read tag headers. For example, Stable Diffusion stores its parameters inside the EXIF `UserComment` tag (tag index `0x9286`) as an undefined byte array. Our parser isolates this tag, strips the character prefix (such as ASCII or UNICODE), and decodes the remaining bytes into clear, editable strings.

Frequently Asked Questions

How does the AI Image Prompt Metadata Viewer work?

When you upload an image generated by Stable Diffusion (A1111/Forge), ComfyUI, Midjourney, NovelAI, or Fooocus, the tool parses the binary chunks of the file (such as tEXt, iTXt, or description tags in PNG/JPEG/WebP) to find embedded metadata containing the original prompt, negative prompt, sampler, seed, model name, and other configuration parameters.

Which AI generators and formats are supported?

We support PNG, WebP, and JPEG files generated by: Stable Diffusion WebUI (Automatic1111, SD.Next, Forge, Fooocus), ComfyUI (raw workflows or parameter lists), Midjourney (when saved with metadata), and NovelAI. Note that some social platforms or chat apps (like Discord or Twitter) compress images and strip metadata, which makes extraction impossible for those shared files.

Are my images uploaded to any server?

No. Just like all tools on Toolmaxy, the AI Image Prompt Metadata Viewer processes your images 100% locally in your web browser. No data, image files, or prompts are sent to our servers, ensuring your generation parameters and images remain completely private.

How do I extract a ComfyUI workflow from a PNG?

When you drop a ComfyUI generated PNG, this tool will detect the embedded JSON workflow. It will display the prompt text and offer a dedicated button to copy or download the entire JSON workflow file, which you can drag directly back into ComfyUI to load the workspace.

Why is my image showing "No AI Metadata Found"?

This happens if the image was not generated by a supported AI engine, or if the metadata has been stripped. Web services, image hosts, and messaging apps (like Discord, WhatsApp, or Twitter/X) automatically compress and strip EXIF/PNG chunks for optimization and privacy, removing the embedded AI metadata.

Does this tool support batch processing?

Yes! You can drag and drop multiple AI-generated images at once. The tool will parse each image and display the prompts, negative prompts, models, and seeds in a clean list, allowing you to copy individual prompts or export them in batch.

More Image Tools

View all