OpenZhenZhen AI 提示词画廊 · gpt-image-2.5 · image

六种表情的3D成年角色肖像拼贴

3D Portrait Collage of an Adult Character with Six Expressions

以3×2网格呈现同一位参考绑定的成年角色六张风格化3D肖像，分别表现喜悦、惊讶、严肃、可爱、俏皮和自信，搭配六种鲜艳纯色背景。

六种表情的3D成年角色肖像拼贴 — AI image creation
六种表情的3D成年角色肖像拼贴

提示词 / Prompt

中文提示词

以用户上传的参考图作为主要可见人物的面部身份依据。在不与身份参考冲突的情况下，保留来源的场景、造型、服装、姿势、构图、光线和视觉风格。将人物渲染为风格化3D插画，保留参考图的面部身份特征；不要转成写实相机人像。创作一幅整洁的3×2拼贴画，准确包含六格精致3D漫画肖像，每格都是同一位明确成年角色米拉，一位原创成年角色，并忠实绑定ref-10a-v1。米拉，一位原创成年角色默认值：“米拉，一位原创成年角色”。六张近景肖像依次呈现清晰不同的表情：喜悦、惊讶、严肃、可爱、俏皮和自信。六格保持相同面部结构与身份，只自然变化表情、轻微头部角度和手势。采用流畅精致的角色设计、柔和环境光，并让每格使用一种不同的鲜艳纯色背景。六格等距对齐，肖像尺度一致，边界清楚，完整保持3×2网格。整体活泼讨喜但不夸张扭曲。不要额外角色、重复格子、文字、标签、标志、水印或无依据的分辨率承诺。

English prompt

Use the uploaded reference image as the facial identity reference for the main visible person. Keep the source scene, styling, clothing, pose, composition, lighting and visual style unless they conflict with the identity reference. Render the person as a stylized 3D illustration that preserves the reference facial identity cues; do not turn the result into a photorealistic camera portrait. Create a clean 3×2 collage of exactly six polished 3D caricature portrait panels featuring the same clearly adult character, Mira, an original adult character, faithfully bound to ref-10a-v1 in every panel. Default Mira, an original adult character: “Mira, an original adult character”. Give each close portrait a distinct, readable expression in this order: joyful, surprised, serious, cute, sassy and confident. Keep facial structure and identity consistent across all six panels while varying expression, slight head angle and hand gesture naturally. Use a smooth refined character-design finish, soft ambient light and a different bold vibrant solid-color background for each cell. Align the six panels evenly with clear boundaries, consistent portrait scale and no gaps that break the 3×2 grid. Keep the mood lively and appealing without exaggerated distortion. No extra characters, duplicate panels, text, labels, logos, watermark or unsupported resolution claims.

查看 API 模型

探索更多图片和视频提示词 · 生成图片 · 生成视频 · 在 AI 画布中创作

  • 特斯拉电动车核心部件示意全景

    横向概念示意海报：以未指定车型的特斯拉电动车为中心，周围安排若干模块放大细节。电池包、电驱电机、逆变器/电力电子、车载充电器、热管理回路、悬架与制动、充电接口均为整理者选定的示意模块，不是来源列明的清单，也不代表特定车型布局或工程规格。 / A landscape conceptual overview centered on a Tesla EV of unspecified model, surrounded by selected module detail callouts. Battery pack, drive motor, inverter/power electronics,

  • 日系负片里的记忆碎片

    A 9:16 candid Japanese negative-film style portrait. The adult person, unposed walk and quiet neighborhood lane are replaceable curator-selected defaults, not subject or scene details supplied by the source. The currently configured ref-10a-v1 is used as the generic person identity reference and is

  • 1girl

    gpt image2.5生成

  • 法式风情地标美食海报

    A premium, clean and elegant poster concept built from the source’s broad request. Eiffel Tower, Arc de Triomphe, ratatouille and Provençal bouillabaisse are curator-selected illustrative examples, not details stated by the source; they can be replaced in a future variant. The source example image i

  • 梦回江户浮世绘古典人物像

    A reference-led Edo-period ukiyo-e portrait transformation that preserves the supplied person’s facial identity while restyling hair, clothing and accessories in traditional Japanese fashion. The current ref-10a-v1 is the generic person preview for this workflow, not a claim about any external named

  • 盛世长安万国来朝图

    A text-driven historical reconstruction of the source scene: an aerial view of Kaiyuan-era Chang’an, Emperor Xuanzong and Yang Guifei looking out from Hanyuan Hall, and a distant tribute procession. The configured generic portrait is used only as a face-identity anchor for Yang Guifei in this genera