AI Inspection Utility

AI Metadata Viewer

Audit embedded generative metadata, view synthesis prompts and seed numbers, and inspect software identifiers locally on your device.

Drop your image here

or choose a file from your device

Supported formats:JPGJPEGPNGWebP
Processed locally in browser memory

What Information Can You Inspect?

When generative platforms export files, they embed structured diagnostic records into header containers. The AI Metadata Viewer exposes these parameters in an organized, human-readable format:

Synthesis PromptsInspect the full positive prompt text and negative prompt exclusions used during generation.
Diffusion SettingsView sampler algorithms (Euler a, DPM++ 2M Karras), inference step counts, and CFG scale.
Model & Hash DetailsIdentify base checkpoint models, LoRA weights, and embedding hashes recorded in the file.
Software PlatformIdentify source environments such as Automatic1111 WebUI, ComfyUI, Midjourney, or DALL-E.

Zero-Upload Local Parsing

The entire inspection runs in browser memory using the HTML5 File API and typed binary readers. Image bytes never cross a network interface, guaranteeing that your unreleased visual concepts and prompting formulas remain strictly confidential.

Supported AI Formats

  • PNG Files: Reads embedded parameters and prompt text chunks.
  • JPEG Images: Extracts EXIF UserComment and XMP parameter blocks.
  • WebP Images: Parses extended RIFF metadata chunks.

Limitations of AI Metadata Inspection

Metadata inspection only reveals declarative tags stored in file headers. If an image was processed through social media platforms, messaging applications, or screenshot utilities, header chunks are typically stripped. Consequently, the absence of AI metadata does not prove that an image is human-created.

How to Inspect AI Metadata in 3 Steps

  1. Load your image: Drag and drop your image file into the viewer zone above.
  2. Click "Inspect Metadata": The local parser reads header bytes without server uploads.
  3. Review extracted tags: Copy generation prompts, review seed integers, or verify model checkpoints.