Seedance 2.0
适合质量优先和更长镜头控制的提示词生成。
Generate the final video with synced audio when the selected model supports it.
你基于提示词生成的视频会显示在这里。
Preview the kinds of prompts and outputs this tool flow is designed for.
Cinematic Mountain Scene
A majestic mountain range at golden hour, camera slowly flying through valleys
Urban City Timelapse
New York City timelapse at night, cars leaving light trails, buildings glowing
Ocean Sunset
Calm ocean waves at sunset, camera slowly zooming out, peaceful atmosphere
Text to Video 是纯提示词驱动的工作流,无需上传素材即可把创意转成视频,适合概念验证、广告钩子探索和叙事预演。
当你有明确文案意图但没有固定视觉参考时,优先用 Text to Video。
当需要严格遵循角色、动作节奏或风格一致性时,切换到 Reference to Video。
建议按“主体-环境-运镜-光线-节奏”顺序书写,并每次只改一个变量。
当视觉身份必须一致,或文生结果出现持续风格漂移时切换。
可以。通常先低成本批量打样,再把入选提示词交给高质量模型重渲染。