AI Image Prompt Extractor

Extract prompts from PNG metadata of AI-generated images (ComfyUI, A1111, NovelAI).

📁

Drag and drop a PNG file here or click to select

Select File

Files are processed in your browser and never uploaded to a server

How to Use

  1. Drag and drop the original AI-generated PNG, click to select it, or paste it with Ctrl+V.
  2. Positive/negative prompts and generation parameters such as seed, steps, CFG, sampler, model and LoRA are shown automatically.
  3. Use the Copy button on each prompt, and open Raw Metadata if you need the full original data.
  4. Drop another image into the second slot to switch between the two and compare their prompts.

FAQ

Which image generators are supported?

PNGs saved by ComfyUI (including Qwen Image Edit and split-sampler workflows), Stable Diffusion WebUI (A1111 / Forge) and NovelAI are supported. Other PNGs with text metadata can still be inspected under Raw Metadata.

Why does it say "No metadata found" instead of showing a prompt?

Images re-downloaded from X (Twitter), Instagram or messaging apps, screenshots, and images converted or compressed to another format lose their generation data. Please use the original PNG file saved by the generator.

Does it work with JPG or WebP images?

Only PNG files are supported for now. Generation data is read from PNG text chunks (tEXt, iTXt, zTXt).

The prompt shows "{Runtime auto-generated caption: ...}". What does that mean?

In ComfyUI, a node such as QwenVL or Florence wrote the prompt from the image while the workflow was running. That generated text is not stored in the metadata, so only the node name and its instruction are shown.

Are my images sent to a server?

No. Files are read and analyzed entirely inside your browser and are never uploaded to any server.