OpenAI's GPT Image 2.6 prompting guide: 7 tips for precise image edits
mark_k · x · 2026-11-07
OpenAI's official GPT Image 2.6 prompting guide, summarized by markk, targets the common pain of regenerating a whole image when you only wanted one tweak:
- State the format and composition: banners, portraits, and product shots need different layouts—specify where the subject goes.
- Describe what's actually visible instead of style words ("rocket on the right, pink exhaust, dark sky on the left for a headline").
- Put exact text in quotes, then zoom in to verify spelling and legibility.
- Edit one thing at a time: build on the last good image, name what must stay fixed, and paste the original back if a logo needs pixel-perfect fidelity.
- Give every reference image a job: one for layout, one for the logo, one for the palette.
- For API edits, circle the target via a transparent markup layer, asking for the circle to vanish in the output.
- Tune complexity and quality separately: raise quality when fine print is blurry; pick Flare for speed, Sunburst for demanding layouts and precise edits.
More from Multimodal
- 2-step pDMD LoRA for p-h3-image released, fix blurry fast-generation outputs — Illustrious_Row_9971 · 2026-11-07
- Dreamy pastel Midjourney explorations: candy factories and coastal towns in pink clouds — hewarsaber · 2026-11-07
- Midjourney 8.2 test: stunning Munch-inspired expressionist artwork — LudovicCreator · 2026-11-07
- Midjourney style code --sref 4173596805: cinematic grainy night scenes of the Colosseum — sergeantsref · 2026-11-07
- How I built a rhythm racing game with Claude Opus 5.5, six skills, and a voice feedback loop — Groowl · 2026-11-07
- Stylized 2D-to-3D character conversion preserves every design detail — azed_ai · 2026-11-07