上传两张含有不同人脸的照片
为了获得最佳效果,请使用需要合成的人脸近景照片!
Create text-to-video or first-frame image-to-video with automatic upload, polling, and preview.
At least 5 characters. Describe camera movement, action, lighting, and style.
Image-to-video requires one first-frame image. Text-to-video can run without it.
Seedance multimodal supports up to 9 images, 3 videos, and 3 audios.
RH-G supports 480p/720p, 6-30 seconds, and five aspect ratios.
Waiting to create task
Keep the all-in-one page for flexible creation, and use specialized cards for faster, cleaner, scenario-based production.
上传产品/人物参考图,输入创意描述,规划视频时长,自动生成适合 Seedance 2.0 的提案版式图。
先由 LLM 分析参考图与需求,生成镜头文案,再调用生图模型生成完整故事版。
包含创意概念、造型设定、场景设计、镜头调度、分镜脚本与声音节奏。
等待生成文案...
生成结果会以截图示例风格呈现:左侧设定,中间场景与调度,右侧分镜,下方氛围与节奏。
Original on the left, generated image on the right
--
裁剪 / 扩图 / 标注 / 画笔
Language, theme, logs and local save folder
Choose a local folder for generated images
Sensitive fields are hidden.
Enter an amount, then scan the QR code to pay
Scan with WeChat to pay