Image prompt archive
GPT-Image-2 Prompt Patterns
A full-site archive of adapted GPT-Image-2 prompt patterns based on public community cases, rewritten into reusable templates with controls, failure modes, and attribution.
47 adapted prompt patterns across portraits, posters, character sheets, UI mockups, and community benchmark cases.
Every page keeps source attribution but replaces verbatim prompt reuse with a stronger derivative template and editorial guidance.
Use the archive to compare renderer behavior, rewrite visual briefs, and pressure-test GPT-Image-2 workflows before relying on an output.
How this archive is curated
Each entry starts from a public image-generation case, then turns the idea into a reusable control model. The page should help a reader understand what to preserve, what to change, what can fail, and where attribution belongs. That makes the archive useful for prompt design, model evaluation, client review, and internal creative QA instead of serving as a gallery of copied examples.
Archive quality gates
| Gate | What the page should help you do | Why it matters |
|---|---|---|
| Source boundary | Understand the public case and attribution before adapting the prompt. | Prevents the archive from becoming a copied-prompt dump. |
| Control model | Pick which subject, camera, layout, identity, or typography controls should change first. | Makes the pattern reusable for a real brief instead of a one-time recreation. |
| Failure review | Reject outputs that break anatomy, text, layout, continuity, attribution, or source context. | Turns prompt browsing into a lightweight evaluation workflow. |
| Commercial readiness | Record the final prompt, model, edits, and rights review before publishing. | Creates a safer path from experiment to client, product, or campaign work. |