原文
GPT Image 2 · en
Scene:
A square material-focused image where lighting, scale, and surface behavior are the main event. Build the frame around the dominant material language shown with full confidence. Keep the lighting, camera distance, and spatial organization tuned for a high-signal research image rather than a casual snapshot. The image belongs to the Ancient Book Restoration theme and should feel immediately legible as a prompt atlas theme for ancient book restoration, testing world-knowledge, editing-workflow with gpt image 2..
Subject:
Center the study on the dominant material language shown with full confidence. Translate the theme into a concrete visual statement instead of an abstract mood board: the viewer should understand what kind of image this is, what is being tested, and why the composition is difficult. Use the slug idea of ancient book restoration as a literal guide for subject matter, materials, and staging.
Important details:
Microstructure matters: refraction, fibers, scratches, pores, edge wear, translucency, and dust should all feel physically specific. Preserve a disciplined square composition, clear focal hierarchy, and enough secondary information to make the image feel complete rather than underspecified. The image should look polished, novel, and exhibition-ready, with topic-specific credibility coming from the way surfaces, structure, and visual language reinforce Ancient Book Restoration. Avoid generic filler elements that do not serve the main concept.
Use case:
Ancient Book Restoration single study 1/8, testing world-knowledge, editing-workflow through the surface hero variation.
Constraints:
No readable text, no watermark, no logos, preserve exact 1:1 aspect ratio, no generic stock macro, no broken scale cues, no plastic-looking texture when it should feel natural, no meaningless blur, no watermark, no fake labels.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。