怎麼填變量：[深棕色頭髮] 可以改髮色和人物描述；把女性肖像換成"一隻貓的頭像""一個蘋果""一隻手"，就能得到不同主題的素描步驟圖（步驟描述裏的"五官、頭髮"要改成對應的結構，比如蘋果寫"外輪廓 → 明暗交界線 → 投影 → 高光"）。想做 6 步，把網格改成 3×2，並把中間步驟合併。

示例圖是白紙上的 3×3 九宮格：左上只有淡淡的橢圓和十字輔助線，往右往下依次出現線稿、淺調子、深調子，到右下角是一張完成度很高的寫實鉛筆肖像——盤發、碎髮、垂墜耳環和深色露肩上衣，九格里人物一致。

常見問題：

• 後幾格看起來一樣：這是正常的（原作第 7～9 格就是完成稿的微調），想要差異更明顯就減少到 6 格。
• 第 1 格太"完整"：強調"只有幾何輔助線，幾乎看不到五官"。
• 用於教學：AI 生成的過程圖是"示意"，不完全等於真實繪畫順序，講解時請結合實際技法。

適合：繪畫教學、美術課講義、畫畫帳號內容、素描練習參考。

英文原版

text
Goal: Create a vertical 3-by-3 tutorial-style progression sheet showing the same young woman’s portrait being developed from a faint construction drawing into a polished realistic graphite pencil sketch on clean white drawing paper.

Canvas: Portrait-oriented white paper background, 4:5 aspect ratio, divided into exactly 9 equal rectangular panels arranged in a 3 columns x 3 rows grid. The grid divisions are very subtle, created by slight tonal differences in the paper rather than visible borders.

Subject: The same front-facing young woman appears centered in every panel from the shoulders up. She has {argument name="hair color" default="dark brown hair rendered in graphite"}, thick softly arched eyebrows, almond-shaped light eyes, a straight delicate nose, full lips with a faint closed-mouth smile, an oval face, and a calm elegant expression. Her hairstyle is a voluminous loose updo with wispy curled strands falling beside both cheeks. She wears dangling earrings and an off-shoulder dark dress or neckline that forms a soft V shape across the bottom of the portrait.

Panel count and progression: Include exactly 9 portrait panels. Panel 1, top left: extremely faint initial construction sketch with an oval head, centerline, horizontal facial guide lines, rough neck, shoulders, and hair mass. Panel 2, top center: light clean line art with facial features, hair outline, earrings, shoulders, and neckline established. Panel 3, top right: early shaded portrait with more developed eyes, lips, hair texture, earrings, and dark neckline beginning. Panel 4, middle left: stronger graphite rendering, fuller hair shading, clearer facial planes, eyes, lips, shoulders, and dress. Panel 5, middle center: refined realistic portrait with balanced midtone shading, detailed hair strands, defined earrings, and darker neckline. Panel 6, middle right: nearly finished portrait with high contrast hair, smooth skin shading, detailed eyes, and darker clothing. Panel 7, bottom left: finished graphite portrait with polished face and hair, strong neckline shadow, and soft shoulder shading. Panel 8, bottom center: finished portrait variant with slightly smoother lighting and crisp facial detail. Panel 9, bottom right: final polished portrait, high realism, clean graphite values, detailed hair, earrings, expressive eyes, soft skin transitions, and dark off-shoulder garment.

Visual style: Extremely realistic traditional graphite pencil drawing, monochrome black-and-white, visible pencil grain, delicate crosshatching and smudged blending, soft eraser highlights, clean white drawing paper texture. The top-left panel should be very pale and sketchy, and each subsequent panel should gradually increase in contrast, detail, and realism until the final bottom-right portrait looks complete.

Constraints: Keep the woman’s identity, proportions, pose, expression, hairstyle, jewelry, and clothing consistent across all 9 panels. No text labels, no watermark, no colored ink, no digital UI elements, no extra objects, and no background scenery.

改編自 Hania Ai 發佈、YouMind-OpenLab/awesome-gpt-image-2 收錄的提示詞，許可證 CC BY 4.0，本站已翻譯並改寫（修改內容見 changes）；示例圖來自同一條目。