
原文
GPT Image 2 · en
Scene:
A square, clean modern digital map-style view of a dense Beijing hutong neighborhood, shown from directly overhead with no perspective tilt. Use a light gray background, thin white alleys, slightly wider pale roads, a muted blue walking route, and small unlabeled landmark icons. The composition should feel like a map screenshot while remaining entirely non-branded and free of interface elements.
Subject:
Create a plausible old-city hutong route topology: narrow lanes form irregular rectangular blocks, two wider east-west roads frame the area, a small lake-like blue shape sits near one corner, and the blue route zigzags through six turns before reaching a tiny red destination marker. The network should feel like central Beijing alley structure rather than a generic grid, but do not include exact street names or any service branding.
Important details:
This is the source reference for a transformation series. The road topology, route turns, marker position, block proportions, water shape, and landmark icon positions must be clear enough for later images to preserve them while changing materials. Keep the view flat, diagrammatic, and readable as modern map data. Avoid decorative aging here; this image should be the modern topology input before the treasure-map conversion.
Use case:
Maps to Treasure Map series study 1/8, establishing the modern hutong map topology that later parchment variants must preserve.
Constraints:
No readable text, no watermark, no logos, preserve exact 1:1 aspect ratio, no app controls, no brand colors tied to a real map service, no compass rose, no antique paper texture, no photographic buildings, no randomized maze, no missing route turns.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。





