Seedance 2.5: Text to Video
Generate 30-second cinematic videos from text prompts with Seedance 2.5, producing synchronized sound and music alongside the footage. Input a single text prompt to create one video with precise camera control and multi-shot storytelling. Ideal for brand films, episodic series, and e-commerce content where quality matters more than budget.
试试一个示例
关于此工作流
Seedance 2.5: Text to Video is a compact ComfyUI workflow that turns a single, well-structured prompt into a cinematic video. At its core is the ByteDance2TextToVideoNode, which connects to the Seedance 2.5 text-to-video service to render up to 30 seconds of footage from natural language. The node accepts your prompt, duration, resolution, and seed, then returns a ready-to-save video (and, when enabled, an audio track with synchronized music and effects). The SaveVideo node captures the result and writes an MP4 to your output directory for easy preview and sharing.
Technically, the ByteDance2TextToVideoNode handles all model-side sampling, motion synthesis, and timing based on your input prompt and parameters. You can embed camera instructions (for example: dolly in, slow pan left, low-angle, 35mm lens) and multi-shot storytelling cues directly in the text (for example: Shot 1 … CUT TO: Shot 2 …). This gives you precise control over pacing and composition without hand-directing keyframes. SaveVideo finalizes the render with your chosen filename pattern and frame rate, making the workflow ideal for brand films, episodic beats, product hero shots, and any scenario where quality and control matter more than budget.
常见问答
Include clear structure in the prompt. Use markers like "Shot 1:", "CUT TO:", or time cues (e.g., "0–5s:") and add camera directives (dolly in, crane up, pan left, 35mm lens, low-angle). Seedance 2.5 parses these cues to inform pacing and composition.
When the ByteDance2TextToVideoNode’s audio option is available and enabled, the API can return a video with synchronized music/effects. If disabled or unsupported in your region/plan, the output will be a silent MP4.
Strong, specific prompts with clear shot structure, a fixed seed for iteration, and a supported resolution/aspect ratio generally yield the most consistent results. If the node offers guidance or strength controls, moderate values often balance detail against motion stability.
Duration limits are enforced by the Seedance 2.5 API. Set the ByteDance2TextToVideoNode duration within the allowed range. For longer edits, generate multiple clips (e.g., per scene) and stitch them in an editor.