AI Prompt Extractor (PNG Info)
Read the prompt and settings saved inside AI images. Free, no sign-up, and your images stay on your device.
or drop PNG, JPEG or WebP files here. Add several at once.
Your images stay on your device.
Most AI image tools write the prompt and the generation settings into the image file when they save it. This page reads them back so you can see exactly how a picture was made, copy the prompt into your own tool, or load a ComfyUI workflow again. It is a PNG Info viewer that also works for JPEG and WebP images, and for many tools besides Stable Diffusion.
How to extract a prompt from an AI image in 3 steps
- 1
Add your image
Drop a PNG, JPEG or WebP made with AI, or tap Choose image. You can add many at once.
- 2
Read the prompt
See the prompt, negative prompt and a clean table of settings: seed, steps, sampler, CFG scale, size, model and more.
- 3
Copy or download
Copy the prompt with one tap. For ComfyUI images, download the workflow as a .json file.
What it reads
- ✓Stable Diffusion web UI: AUTOMATIC1111, Forge and SD.Next images, including LoRAs, embeddings, hires fix and model hash.
- ✓ComfyUI: the prompt wired into the sampler, the settings, and the full workflow to download.
- ✓NovelAI: the prompt, undesired content, character prompts and settings, even when only the picture kept them.
- ✓Midjourney: the prompt from website downloads, with --ar, --v, --stylize and other parameters listed.
- ✓InvokeAI, Fooocus and SwarmUI: plus RuinedFooocus, Easy Diffusion, Draw Things and Civitai images.
- ✓ChatGPT, Firefly and Gemini: these do not save the prompt, but the page tells you when the file names the app that made it.
What this cannot do
- It does not guess prompts. It only shows what the file itself saved. It cannot describe a picture or invent the prompt for an image that has none.
- Removed details are gone. Sites like Discord, X, Instagram and WhatsApp often remove the prompt on upload, and screenshots never have one. Nothing can bring it back.
- Some apps never save it. ChatGPT, Gemini and Adobe Firefly keep the prompt to themselves. You may still see which app made the file.
- Unusual ComfyUI setups. If a workflow builds its prompt with rare custom nodes, the prompt may not show. The workflow download still has everything.
- It does not change your file. To remove the prompt before sharing, use the AI metadata remover.
PNG Info and prompt extractor FAQ
What is PNG Info in Stable Diffusion?
PNG Info is the tab in the AUTOMATIC1111 Stable Diffusion web UI that shows the prompt and settings an image was made with. Most Stable Diffusion tools save these details inside the image file. This page does the same job in your browser, without installing anything, and it also reads images from ComfyUI, Forge, NovelAI, InvokeAI, Fooocus, SwarmUI and Midjourney.
How do I extract the prompt from an image?
Drop the image on this page or tap Choose image. If the file still has its prompt, you see it right away with the negative prompt, seed, steps, sampler, CFG scale, size, model and LoRAs. Press Copy prompt to paste it into your own tool. It works best on the original file from the app that made it, not a copy saved from social media.
How do I load a ComfyUI workflow from an image?
Add a ComfyUI image here and press Download workflow (.json). Then drag the .json file onto the ComfyUI window to load the whole workflow with every node. You can also press Copy workflow to keep it on your clipboard. Dragging the original PNG straight into ComfyUI works too, as long as the file was not re-saved by another app.
Can I get the prompt from a Midjourney image?
Often, yes. Images downloaded from the Midjourney website usually keep the prompt, the parameters such as --ar and --v, and the Job ID. This page shows the full prompt and lists each parameter. Images saved from Discord or from social media usually have none of this, so there is nothing to read.
Why does it say no prompt is saved in my image?
Many sites and apps, such as Discord, X, Instagram and WhatsApp, remove these details when you upload or send an image. Screenshots never have them. ChatGPT, Gemini and Adobe Firefly do not save the prompt either; when the file says which app made it, this page tells you. The tool cannot guess a prompt from the picture itself.
Is my image uploaded? Is it free?
Your images stay on your device. They are read in your browser and never sent to a server. There is no account and no sign-up, and the tool is free. You can add up to 50 images at a time.
Related tools
Questions or a file that reads wrong? Email team.kyiron@gmail.com.