BAMoore UX Portfolio

Pixel Peeper turns batches of Stable Diffusion images into an interactive analysis workflow. Users upload PNGs, surface the prompt and negative-prompt metadata hidden inside, and explore which terms recur across the set. Selecting any ranked term highlights the images it influenced, transforming long, unreadable prompt strings into a visual decision-making tool for AI artists, prompt designers, and creative teams iterating across large image sets.

The problem

AI image generation is fast, but learning from the results is slow. A creator might produce a batch of images and know that some feel stronger than others without being able to articulate why: a style phrase, a subject descriptor, a repeated setting. The signal is already there in the image metadata, just buried.

Pixel Peeper upload state with feature cards — The entry point keeps the workflow clear: upload a batch first, then move into metadata analysis once images are available.

How it works

The experience is intentionally simple, designed around the rapid iteration loop AI artists already use.

Drop in a batch of PNG images and review the grid
Choose to analyze positive prompts, negative prompts, or both
Filter out low-frequency terms to focus on real patterns
Review a ranked list of recurring prompt terms
Select a term to highlight the images it influenced
Export the findings or sort visually with quality scoring

Pixel Peeper prompt analysis controls for uploaded images — Once images are loaded, users choose which prompt metadata to inspect and set frequency thresholds before running analysis.

Pixel Peeper ranked recurring prompt terms — The analysis output ranks recurring terms so patterns become easier to scan, compare, and act on.

Design value

The key design move is turning invisible metadata into a visual decision-making tool. Instead of forcing users to read raw prompt strings one image at a time, Pixel Peeper organizes information around the real question: what patterns are present across this batch?

The result supports pattern recognition across large image sets, reduces friction in creative review, and helps users learn from their own experimentation while balancing analytical power with a lightweight, focused interface for a niche but real workflow.