Research protocol

The AI Image Failure Atlas: 50 Failure Modes and What to Try Next

A transparent atlas of prompt omission, anatomy, counting, spatial, typography, product, cultural, and production failures with repair paths.

A complex diagnostic map representing categories of image-generation failure

The central idea

Failure examples are more educational than a perfect gallery when they are classified honestly. Each entry should show the request, output, visible failure, likely category, repair attempt, result, and the point where a different workflow is safer.

A repeatable workflow

  1. Collect without cherry-picking

    Preserve failed outputs from controlled tasks and remove private, unsafe, or rights-sensitive material.

  2. Classify the failure

    Use defined categories for omission, contradiction, anatomy, count, relation, text, identity, material, culture, and production.

  3. Test a repair

    Change one prompt or workflow variable, show the result, and avoid claiming causality from a single retry.

  4. Define the exit

    Recommend compositing, photography, vector design, charting, or human illustration when generation remains unsuitable.

Worked example

A packaging example may combine plausible nonsense text, impossible folds, invented certification marks, and inconsistent volume. The repair can remove label content and simplify structure, but final packaging still moves to dielines and verified copy.

Review checklist

  • Failures come from controlled tasks
  • Categories have clear definitions
  • Repairs change one thing
  • A safer workflow is named

Limitations

  • One repair attempt cannot prove a general law
  • Unsafe or private failures should not be published for completeness

Open the debugging workflow

Browse all 50 guides and research protocols