PUBLIC BREAKDOWN · 公开案例

Japan Vlog日本 Vlog

A travel vlog workflow that turns model-based reference stills into seven directed clips with GPT Image 2 and Kling.

旅行 Vlog 工作流：用 GPT Image 2 与 Kling 将模特参考静帧转化为七段定向短片。

GPT Image 2Kling 3.0CapCut

Workflow · 工作流程

1
Generate reference stills for each scene using GPT Image 2 with consistent character identity.
用 GPT Image 2 为每个场景生成参考静帧，保持角色身份一致。
2
Animate each still into a directed clip with Kling 3.0.
用 Kling 3.0 将每张静帧动画化为定向短片。
3
Assemble clips in CapCut with transitions and music.
在 CapCut 中组装片段，添加转场与配乐。

分解步骤Breakdown Steps

Breakdown Step · 分解步骤

Seven key scenes are planned first — plane, street, ramen shop, temple, and more.

先规划七个关键场景：飞机、街道、拉面店、寺庙等。

scene-01-plane.png

scene-02-street.png

scene-03-ramen.png

A beginner-friendly breakdown of how reference images became a sci-fi friendship montage — from character concepts to final edit.

一个适合初学者的案例分解：从角色概念图到最终剪辑，展示参考图如何变成科幻友谊蒙太奇。

A mixed reference-driven workflow using Seedance for speaking anchor shots and Kling for inserts.

混合参考驱动工作流：Seedance 负责口播锚点镜头，Kling 负责插入镜头。

A frozen-time editorial workflow starting from GPT Image 2 stills, turning them into slow-orbit Kling O1 clips.

时间冻结编辑工作流：从 GPT Image 2 静帧出发，转为 Kling O1 慢速环绕片段。