分镜叙事

二维角色的真实世界四格

让同一位原创成年动画角色出现在四种真实摄影场景中,保留脸型与服装,并协调透视和接地关系

更新日期:

完整提示词

英文提示词包含建议的示例需求。生成前请修改主体和细节。原始示例输入和生成参数尚未完全核实。

推荐模型: GPT Image 2.5 Flare

Create a single two-by-two contact sheet showing one original adult character visiting four everyday locations. Use an uploaded original character reference image as the identity anchor; if none is supplied, construct the adult from the detailed written brief and keep that design throughout. Preserve the same facial proportions, hairstyle, outfit, palette, and signature accessory in every panel. Render the character with flat cel-animation fills, crisp ink contours, and a small number of drawn shadow shapes. Render each environment photographically, including natural depth, surface detail, and believable camera exposure. Do not turn the character into a photoreal person or a three-dimensional figurine. Instead, connect the two visual styles through shared perspective, correct feet placement, contact shadows, and plausible interaction with furniture or handheld objects. Give each panel a different modest action, with the hands readable and the face unobscured. Keep the four cells equal in size, separated by narrow neutral gutters, and vary the setting without adding duplicate characters inside a cell. Remove visible brand names and incidental readable signs. Aim for a consistent illustrated traveler whose presence feels spatially grounded across all four photographs. User request: An original 32-year-old courier with a short black bob, amber eyes, a mustard jacket, charcoal trousers, and a red messenger bag. Four scenes: choosing a pear at a market stall, waiting beside a crosswalk, reading a map in a cafe, and sitting on a tram. Keep the character flat and hand-drawn; make all backgrounds photographic.

提示词技巧

  • 上传原创角色参考图,或详细描述成年角色;四格都重复指定发型、服装与标志性配饰

来源与署名

由 OmniArt 改编和重写,以下列出来源署名、修改说明及使用条件

  • 2D Anime Character Real-World Grid · @seiiiiiiiiiiru

    EvoLinkAI/awesome-gpt-image-2-API-and-Prompts

    The upstream repository is licensed under CC0 1.0, which does not establish rights to third-party community prompts or media. This template is independently authored; no source text or media is copied.

    Independently authored from a general workflow concept. Uses four photographic environments with a consistent original cel-drawn adult character and explicit perspective/contact-shadow controls. No source prompt text or media copied. GPT Image 2.5 Flare is a recommendation, not a verified example model.

来源与署名

相关提示词配方