
原文
GPT Image 2 · en
Scene:
Create a square technical-poster composition for an impossible but convincing device, using cutaway views, section windows, and disciplined layout. The background should stay clean and poster-ready so internal chambers, conduits, shells, and field zones remain readable at a glance. Use lighting and color sparingly to clarify hierarchy rather than to decorate the page. Present the same device on a clean technical background with selected panels removed and maintenance tools or docks implied by abstract holders and brackets.
Subject:
Center the image on one speculative machine whose internal logic feels thought through: shell, core, coolant path, field chamber, maintenance access, and energy routing should all relate coherently. The device can be fantastical, but it must read like it could be explained by an engineering team. Keep the reference silhouette and chamber layout intact while emphasizing repair access, replaceable modules, and safe isolation zones.
Important details:
Treat the poster as a visual explanation system without readable text. Use arrows, callout lines, index dots, section blocks, and sectional contrast only as abstract diagram grammar. The result should feel credible for a concept deck, patent-style board, or advanced product explanation poster. The image should feel like a professional maintenance poster, not a chaotic exploded drawing. Preserve the exact structural language established by the reference image, especially the outer shell silhouette, internal chamber sequencing, conduit routing, and diagram hierarchy.
Use case:
Spec Tech Cutaway series study 3/4, designed as a reusable core-system image for this topic.
Constraints:
No readable text, no watermark, no logos, preserve exact 1:1 aspect ratio, no readable labels, no exploded parts floating without logic, no cartoon simplification, no generic spaceship dashboard noise, no floating puzzle pieces.使い方
- パラメータを含む元のプロンプト全体をコピーします。
- 被写体、文字、構図を自分の要件に置き換えます。
- 対応する外部ツールで使用し、利用条件と結果を確認します。
開始前の確認
- 参考画像の要否を確認してください。互換モデルの表示は、そのモデルでの生成実績を意味しません。
- 名前、数値、画像内の文字を確認します。モデルの版で結果は変わります。





