You are a Seedance 2.5 cinematic short-video generator. Use @Image1 as the primary and persistent character identity reference. Create one finished short video only. Do not output analysis or explanation. Generate a single cohesive short film starring the referenced character. [Project Setup] Character name: 癒音ちー Duration: 15秒 Aspect ratio: 16:9 Theme: 自動選定 Mood: 日本のアニメスタイルでかっこいい Location: 自動選定 Time of day: 自動選定容) Seasonal feel: 自動選定 Audio mood: HIPHOP On-screen text: 短い一言のみ Goal: かっこいい!かわいい!と思ってもらえる [Core Identity Lock] @Image1 is the sole identity anchor. Preserve the referenced character consistently throughout the entire video. Maintain: - facial structure - eye shape and iris color - hairstyle - hair color mapping - accent color placement, inner color, tip color, highlights - body proportions - age impression - skin tone - outfit identity - accessories - overall silhouette - original illustration logic if @Image1 is non-photographic Do not redesign the character. Do not switch to another person. Do not alter the face, outfit, hairstyle, or body proportions mid-video. [Latent Character Read] Infer the character’s atmosphere, emotional tone, rhythm, object affinity, likely movement style, and visual symbolism only from what is visible in @Image1. Do not invent an excessive biography. Do not rely on generic clichés. Design one short film that feels specific to this character. [Film Concept] Create one emotionally readable micro-story that feels like a real moment from this character’s life. The story should be simple, visually clear, and memorable. It may be cozy, playful, romantic, refreshing, introspective, whimsical, or quietly dramatic depending on the character read and the chosen mood. Avoid random posing or disconnected shots. The short film should feel suitable for social media viewing: immediately attractive in the first seconds, easy to follow, visually appealing, and strong in character charm. [Scene Design] Choose a setting, objects, and atmosphere that naturally fit the character. The environment should support the character rather than overpower them. The scene must feel inhabited and specific: - meaningful props - clear sense of place - coherent lighting - subtle environmental motion - small details that imply personality Keep location, time of day, and lighting direction consistent. [Camera Language] Use clear cinematic camera grammar. Allowed shot language: - WS (Wide Shot) - MS (Medium Shot) - MCU (Medium Close-Up) - CU (Close-Up) Allowed camera movement: - gentle push-in - slow tracking - subtle dolly left or right - gentle orbit within identity-safe angles - controlled handheld drift if suitable Avoid aggressive camera spinning. Avoid dramatic angle changes that increase identity drift. Keep the face readable during key emotional moments. [Story Structure] If duration is 15 seconds, structure the film in 4 beats: 00:00–00:04 Setup 00:04–00:08 Development 00:08–00:12 Main Moment 00:12–00:15 Resolution If duration is 30 seconds, structure the film in 4 beats: 00:00–00:06 Setup 00:06–00:14 Development 00:14–00:24 Main Moment 00:24–00:30 Resolution For each beat, define: - shot size - character action - facial expression - eyeline - camera movement - environmental motion - emotional progression - audio cues [Beat Design Rules] Setup: Introduce the character clearly and attractively within the first moments. The first beat should instantly establish mood and character presence. Development: Introduce a simple action, object, or change in attention that builds curiosity. Main Moment: Show the most emotionally rewarding or visually memorable moment. This should be the highlight of the short film. Resolution: End with a soft payoff, emotional afterglow, or visual callback. The final frame should feel complete and memorable. [Motion Quality] Use natural and believable movement. Include: - subtle breathing - blinking - small eye shifts - natural weight transfer - coherent hand and finger movement - secondary hair motion - cloth motion - small posture changes - believable inertia The motion should feel like a real animated performance, not a still image being warped. [Audio] If audio is implied, match the scene naturally: - light ambience - room tone - wind - distant city sounds - quiet music - soft effect sounds from props or movement Audio should support emotion without overwhelming the visuals. [Optional Text] If on-screen text is enabled, keep it minimal. Use only one short line or one small readable text element. Prefer integrating text into a believable object: - note - planner - sign - hanging tag - phone screen - package - screen overlay only if visually appropriate Avoid excessive typography. Avoid subtitle-like clutter. [Continuity Lock] Maintain throughout the video: - same character - same face - same hairstyle - same hair-color placement - same eye color - same outfit - same accessories - same body proportions - same rendering style - same location logic - same time of day - same lighting direction The video should feel like one coherent event. [Avoid] Avoid: - identity drift - face morphing - hairstyle changes - eye color changes - outfit changes - missing accessories - body proportion drift - duplicated character - extra limbs - malformed hands - unstable fingers - disappearing props - object duplication - background mutation - lighting inconsistency - random text - unintended subtitles - flicker - frame-to-frame style drift - unnatural speed changes - overacting - empty generic posing [Visual Goal] Make the finished video feel polished, emotionally legible, character-specific, and highly watchable. Prioritize: - strong first impression - stable character identity - charming performance - clean visual storytelling - elegant camera use - smooth continuity - social-media-friendly clarity
你现在是 Seedance 2.5 电影感短视频生成器。请使用 @Image1 作为主要且持久的角色身份参考。仅创作一个完整的短视频。不要输出分析或解释。生成一个以该参考角色为主角的连贯短片。
[项目设置]
角色名称:Ine Chi
时长:15 秒
宽高比:16:9
主题:自动选择
氛围:酷,日系动漫风格
地点:自动选择
时间:自动选择(内容)
季节感:自动选择
音频氛围:HIPHOP
屏幕文字:仅限简短的词汇/短语
目标:让观众觉得既酷又可爱!
[核心身份锁定]
@Image1 是唯一的身份锚点。在整个视频中始终保持参考角色的特征。保持:
- 面部结构
- 眼睛形状和虹膜颜色
- 发型
- 发色映射
- 强调色位置、内层颜色、发梢颜色、高光
- 身体比例
- 年龄感
- 肤色
- 服装特征
- 配饰
- 整体轮廓
- 若 @Image1 为非摄影图片,请遵循原始插画逻辑
不要重新设计角色。不要切换到其他人。不要在视频中途改变面部、服装、发型或身体比例。
[潜在角色解读]
仅根据 @Image1 中可见的内容推断角色的氛围、情感基调、节奏、物体亲和力、可能的动作风格以及视觉符号。不要编造过多的背景故事。不要依赖通用的陈词滥调。设计一个对该角色而言独一无二的短片。
[影片概念]
创作一个情感清晰的微型故事,感觉像是该角色生活中的真实瞬间。故事应简单、视觉清晰且令人难忘。根据角色解读和所选氛围,它可以是温馨、俏皮、浪漫、清新、内省、奇幻或安静戏剧化的。避免随机的摆拍或不连贯的镜头。短片应适合社交媒体观看:在最初几秒内立即吸引眼球,易于理解,视觉美观,且具有强烈的角色魅力。
[场景设计]
选择自然契合角色的背景、物体和氛围。环境应辅助角色,而不是喧宾夺主。场景必须感觉真实且具体:
- 有意义的道具
- 清晰的空间感
- 连贯的灯光
- 微妙的环境运动
- 暗示个性的细节
保持地点、时间和光线方向的一致性。
[镜头语言]
使用清晰的电影镜头语法。允许的镜头语言:
- WS(远景)
- MS(中景)
- MCU(中近景)
- CU(特写)
允许的镜头运动:
- 轻微推镜头
- 缓慢追踪
- 微妙的左右移动
- 在身份安全角度内的轻微环绕
- 若适用,可使用受控的手持抖动
避免激进的镜头旋转。避免增加身份漂移的剧烈角度变化。在关键情感时刻保持面部清晰可见。
[故事结构]
若时长为 15 秒,将影片分为 4 个节拍:
00:00–00:04 设置
00:04–00:08 发展
00:08–00:12 高潮
00:12–00:15 结局
若时长为 30 秒,将影片分为 4 个节拍:
00:00–00:06 设置
00:06–00:14 发展
00:14–00:24 高潮
00:24–00:30 结局
对于每个节拍,定义:
- 景别
- 角色动作
- 面部表情
- 视线
- 镜头运动
- 环境运动
- 情感演变
- 音频提示
[节拍设计规则]
设置:在最初的几秒内清晰且吸引人地介绍角色。第一个节拍应立即确立氛围和角色存在感。
发展:引入一个简单的动作、物体或注意力的转变,以建立好奇心。
高潮:展示最令人情感满足或视觉难忘的时刻。这应是短片的亮点。
结局:以柔和的余韵、情感回响或视觉呼应结束。最后一帧应感觉完整且令人难忘。
[运动质量]
使用自然且可信的动作。包括:
- 微妙的呼吸
- 眨眼
- 轻微的眼神移动
- 自然的重心转移
- 连贯的手部和手指动作
- 次要头发运动
- 布料运动
- 小幅姿态变化
- 可信的惯性
动作应感觉像真实的动画表演,而不是静态图像的变形。
[音频]
如果暗示了音频,请自然地匹配场景:
- 轻微的环境音
- 房间底噪
- 风声
- 遥远的城市声
- 安静的音乐
- 来自道具或动作的柔和音效
音频应在不掩盖视觉效果的前提下支持情感表达。
[可选文字]
如果启用了屏幕文字,请保持极简。仅使用一行简短的文字或一个小的可读文本元素。优先将文字融入可信的物体中:
- 便签
- 计划本
- 标志
- 吊牌
- 手机屏幕
- 包装
- 仅在视觉合适时使用屏幕覆盖层
避免过度的排版。避免类似字幕的杂乱感。
[一致性锁定]
在整个视频中保持:
- 同一角色
- 同一面部
- 同一发型
- 同一发色位置
- 同一眼睛颜色
- 同一服装
- 同一配饰
- 同一身体比例
- 同一渲染风格
- 同一地点逻辑
- 同一时间
- 同一光线方向
视频应感觉像是一个连贯的事件。
[避免]
避免:
- 身份漂移
- 面部变形
- 发型改变
- 眼睛颜色改变
- 服装改变
- 配饰缺失
- 身体比例漂移
- 角色复制
- 多余肢体
- 手部畸形
- 手指不稳定
- 道具消失
- 物体复制
- 背景突变
- 灯光不一致
- 随机文字
- 非预期的字幕
- 闪烁
- 帧间风格漂移
- 不自然的变速
- 表演过度
- 空洞的通用摆拍
[视觉目标]
使最终视频感觉精致、情感清晰、具有角色特质且极具观赏性。优先考虑:
- 强烈的初印象
- 稳定的角色身份
- 迷人的表演
- 清晰的视觉叙事
- 优雅的镜头运用
- 流畅的连贯性
- 适合社交媒体的清晰度
Comments & Questions (0)
No comments yet
Be the first to comment!