2️⃣ 每个镜头一张固定图,人设锁定
分享链接 · https://openimages.ajiang.me/p/2097640118366486580/

camera_lens · shallow-depth-of-fieldcomposition · medium-shotcomposition · multi-panelhuman_attributes · facial-hairhuman_attributes · short-hairquality_flags · embedded-text
提示词(原文,逐字保留)
2️⃣ Une image fixe par plan, persona verrouillé
Le plan 1 est généré en texte → image : il devient l'ancre du persona.
Tous les autres plans sont générés en image → image avec l'ancre en référence et un verrou écrit sur le visage, les cheveux, la barbe, le vêtement.
Sur les plans produit, le packshot réel est passé en seconde référence.
👉️Modèle: gpt-image-2-5-flare-image-to-image
👉️ Format: aspect_ratio 9:16 · resolution 1K
👉️Coût: 6 crédits / image
🤖 Mon prompt
Keep the EXACT same claymation character as the FIRST reference image (same face, same short dark-brown clay-coil hair, same short beard, same lean build, same grey knitted hoodie). The Mushilo pouch must be the REAL packshot from the LAST reference image: keep the packaging, colours and label text EXACTLY as shown. <la scène du plan>. Stop-motion claymation still photographed on a miniature set. Real plasticine modeling clay: matte surfaces with visible fingerprints and tool marks, slightly lumpy hand-sculpted shapes, chunky rounded proportions, round bead eyes, hair sculpted as clay coils, clothes sculpted in clay with knitted texture. Handmade diorama built from clay, felt and cardboard, shallow depth of field, warm practical studio lighting. NOT a smooth 3D render, NOT Pixar. 9:16 vertical. No text, no watermark.
中文译文
2️⃣ 每个镜头一张固定图,人设锁定
镜头1采用文→图方式生成:它成为人设的锚点。
所有其他镜头采用图→图方式生成,以锚点作为参考,并对脸部、头发、胡须、服装进行书面锁定。
在产品镜头中,真实的实物图作为第二参考传入。
👉️模型:gpt-image-2-5-flare-image-to-image
👉️格式:aspect_ratio 9:16 · resolution 1K
👉️费用:6 积分 / 张
🤖 我的提示词
Keep the EXACT same claymation character as the FIRST reference image (same face, same short dark-brown clay-coil hair, same short beard, same lean build, same grey knitted hoodie). The Mushilo pouch must be the REAL packshot from the LAST reference image: keep the packaging, colours and label text EXACTLY as shown. <该镜头的场景>. Stop-motion claymation still photographed on a miniature set. Real plasticine modeling clay: matte surfaces with visible fingerprints and tool marks, slightly lumpy hand-sculpted shapes, chunky rounded proportions, round bead eyes, hair sculpted as clay coils, clothes sculpted in clay with knitted texture. Handmade diorama built from clay, felt and cardboard, shallow depth of field, warm practical studio lighting. NOT a smooth 3D render, NOT Pixar. 9:16 vertical. No text, no watermark.
来源与署名
原文由 @Lucas_IA_ 发布在 X:查看原推文。 本页逐字保留原文并提供机器翻译的中文解读;版权归原作者所有。
同工具的更多提示词
- #AIArt #AIgirl Prompt ➡️ https://t.co/pfafZV9Bzt. GPT Image 2.5 is insane! ----…@listudio
- 4:5 竖版社交媒体海报,超高分辨率,8K 场景 / 背景: 平面赛车黄,带细腻渐变(浅黄 → 深琥珀) 颗粒质感 2–3% 柔和矩形叠加层,透明度 3% 产…@Diplomeme
- #AIArt #AIgirl 提示词 ➡️ https://t.co/pfafZV9Bzt. GPT Image 2.5 太疯狂了! ---- 提示语· pr…@listudio
- 创建一个迷人的手工微缩旅行场景,以 [ICONIC STRUCTURE](标志性建筑)作为主要焦点。 将这座地标展示为一个精美雕刻的小型 3D 模型,具有柔和…@Naiknelofar788
- 参考图片人物,生成3*3共9种不同表情包,可以出现中文和英文,浅色渐变背景 https://t.co/eVZRaxIOa8@DeepBlueX0
- 这就是GPTimage2.5的破坏力啊…… #GPTimage25 #AIphoto -------------Prompt---------- Photor…@AI_money_club