Beakfx's MBQ Viewer Deciphers ComfyUI Metadata Without The JSON Soup

MBQ Viewer is a standalone desktop application that reads the embedded metadata inside ComfyUI PNG files and displays it in a clean, color-coded panel next to your images. It also includes a companion node called MBQ Wedge that lets you sweep any numeric parameter across a range, queuing one image per value from a single click. Together, they eliminate the need to dig through raw JSON just to see what settings produced a particular result.
Developer Beakfx built MBQ Viewer after finding that existing image viewers either ignored ComfyUI's prompt data or dumped it as unreadable text blocks. The tool soft-fails gracefully. If a PNG lacks a prompt chunk, it still shows the image without crashing or displaying garbage. The project is fully open source and currently seeking beta testers to catch edge cases outside the creator's own workflows.
A viewer built for comparing image sweeps
- Color-coded metadata panel with three readability tiers.
- OpenGL-accelerated canvas for smooth pan and zoom.
- Zoom lock pins crop position across multiple images.
- Wedge node sweeps any float or int parameter.
- Delete and restore images directly from the viewer.
- Search box highlights matching text across workflows.
- Scroll freeze keeps metadata panel position steady.
- Works on any ComfyUI PNG, no runtime needed.
This tool suits anyone who generates lots of images in ComfyUI and needs to compare results efficiently. Photographers testing denoise values or artists fine-tuning CFG scales can lock zoom on a specific detail and flip through an entire sweep to see exactly where quality changes. Privacy-conscious users will appreciate that everything runs locally, with no cloud uploads or external dependencies required.
What to know before installing MBQ Viewer
The Windows installer is not yet code-signed, so SmartScreen will show a warning that requires clicking "Run anyway" to proceed. Mac support remains untested, though the Python source should theoretically work across platforms. The viewer only supports ComfyUI's SaveImage output format. Images from A1111, InvokeAI, or NovelAI will display but won't populate the workflow panel.
"Every viewer I tried either didn't know about ComfyUI's embedded metadata, or showed it as raw JSON soup. I wanted something that reads the prompt chunk out of each PNG and displays it properly." — Source: Reddit