Loading...
Loading...
See the prompt, seed, model and workflow hidden in a ComfyUI image. This free AI metadata viewer and checker also reads Midjourney, Stable Diffusion and NovelAI files, plus EXIF data. No login required.
Drop an image here or click to browse
PNG, JPEG, WebP, or MP4 — up to 100MB. PNG files are analysed in your browser.
No file handy?
Found the recipe? Numonic keeps it attached to every image from now on, so you never have to dig it out again. Try Numonic free →
Every AI image tool embeds its generation data differently. The inspector reads the raw fields and recognises the formats below automatically.
ComfyUI embeds the full workflow graph and the prompt JSON in PNG tEXt/zTXt chunks. The inspector parses both, detects the tool, shows the prompt, counts the workflow nodes, and lets you copy the workflow JSON so you can reload it in ComfyUI.
Where ComfyUI stores metadata →AUTOMATIC1111 and most Stable Diffusion UIs write a "parameters" text chunk containing the prompt, negative prompt, model, sampler, steps, CFG scale, and seed. The viewer parses that block into a readable summary you can copy from.
PNG metadata vs workflow JSON →Midjourney images downloaded from the website carry the prompt, job ID, and version in embedded fields (Discord downloads usually lose them). For a privacy-focused view of exactly what your Midjourney exports reveal before you share them, use the dedicated Midjourney inspector.
Midjourney Metadata Inspector →NovelAI writes the prompt and generation settings into PNG text chunks, with the settings as JSON in a "Comment" chunk. They appear in the PNG Chunks tab; the Summary tab does not recognise NovelAI yet. The same applies to other tools that write their own tEXt, zTXt, or iTXt chunks.
How to read NovelAI metadata →NovelAI saves the prompt and generation settings in two places: in PNG text chunks, where the settings sit as JSON in a “Comment” chunk, and in a second copy hidden in the image's alpha channel (“stealth pnginfo”), which NovelAI also signs so it can verify its own images.
When the file has text chunks it is parsed in your browser and never leaves your machine. The Summary tab does not recognise NovelAI yet, so read the values from the chunks.
This viewer does not read the alpha-channel copy. If an image's text chunks have been stripped, for example by a re-save or a social-media upload, NovelAI's own NovelAI Inspect may still recover it. A screenshot loses both copies.
NovelAIで生成した画像には、プロンプトと生成設定が2か所に保存されています。PNGのテキストチャンク（設定は「Comment」チャンクにJSON形式で格納）と、画像のアルファチャンネルに埋め込まれた「stealth pnginfo」です。
このツールはテキストチャンクを読み取ります。NovelAIからダウンロードした元のPNGをアップロードし、「PNG Chunks」タブで「Comment」の内容を確認してください。テキストチャンクがある場合、解析はブラウザ内で行われ、ファイルは端末の外に送信されません。
アルファチャンネル内の情報は、現在このツールでは読み取れません。再保存やSNSへのアップロードでテキストチャンクが失われた画像は、NovelAI公式の「NovelAI Inspect」（novelai.net/inspect）で確認できる場合があります。スクリーンショットでは、どちらの情報も失われます。
Quick answers to the most common questions.
The chunk-level detail behind what this viewer parses, with extraction code.
Which metadata survives saving, converting, and re-exporting — and which does not.
What Midjourney embeds, and where it gets lost between Discord and disk.
Why platforms strip the metadata you need, and how to keep provenance anyway.
Privacy-focused companion tool: see what your Midjourney exports reveal before sharing.
Numonic automatically extracts, indexes, and preserves the metadata behind every AI-generated asset — so the workflow, prompt, and parameters you inspected here stay searchable and reproducible across your whole library.