分镜预览四格分镜展示提示词中的画面方向,实际视频效果以生成结果为准。
- 模型
- sora-2
- 时长
- 4s
- 画幅
- 9:16
- 分辨率
- 720p
- 参考图片
- 0
- 参考音视频
- — / —
怎样让镜头更好Cookbook 原始 image-input prompt:「She turns around and smiles, then slowly walks out of the frame.」—— 这里加上官方推荐的 Shot-List 包装
生成后检查什么本站尚未实测。检查主体一致性、运镜连续性与动作完整性;参考原始来源核对效果。
社区来源 · 本站未实测OpenAI Cookbook — Use image input for more control ↗ · cclank/lanshu-awesome-ai-video-kit
MIT (repository)
完整提示词 v1
Style: Soft contemporary portrait cinema, naturalistic skin tones, gentle film grain, painterly background blur.
[Use image input as first-frame reference: a young woman standing in soft afternoon light, three-quarter angle, environment locked in by the image.]
Cinematography:
- Camera: locked off medium close-up matching the reference frame; very subtle handheld breathing
- Lens: 50mm; shallow DOF preserving the reference's separation between subject and background
- Lighting: maintain the reference's soft directional key and ambient fill
- Mood: warm, unguarded, momentary
Actions:
- She turns around
- A small, genuine smile reaches her eyes
- She holds the look for one beat
- She slowly walks out of the frame to camera right; light continues to play on the empty background
Background Sound: Quiet room tone; faint cloth rustle as she turns; a single distant door. No music.
API note: pass the reference image via input_reference; size must match the target video resolution (jpeg/png/webp).
来源模型:Sora · sora-2
原始参数:{"aspect": "9:16", "duration": "4s"}
示例分类不代表本站已接入该模型。实际可用模型以视频片场为准;跨模型使用提示词时,请检查参数并调整。
复用会带入提示词、模型、时长、画幅和已发布的参考素材。缺少的参考文件可在片场补充;提交前会显示费用。
阅读视频创作指南