Do I need API keys or extra models to run this?

Yes. OpenAIGPTImageNodeV2 and GeminiNode require valid API keys configured per their node instructions. The Seedance 2.0 step runs through ByteDance2ReferenceNode, which must be installed and properly set up. Verify all nodes load without errors before running.

Can I use my own storyboard instead of generating one?

Yes. Use LoadImage to bring in a custom storyboard (e.g., a contact sheet) and feed that into the Seedance 2.0 group’s Image input. You can bypass the Previs Storyboard Generator group when doing this.

How can I get better control over camera angles and pacing?

Be explicit in your scene description and any prompt text feeding StringConcatenate: include per‑panel cues like “Panel 1: wide establishing shot,” “Panel 2: medium tracking right,” etc. High‑contrast, well‑framed panels improve guidance. If exposed, increase the reference influence in ByteDance2ReferenceNode for closer adherence to the board.

The video doesn’t match my storyboard closely. What should I try?

Check that the Seedance group’s Image input points to the stitched 8‑panel board you generated. Strengthen visual cues in the storyboard (clean silhouettes, consistent style), refine prompt text for composition clarity, and, if available, raise the reference guidance in ByteDance2ReferenceNode. Iterate until framing and motion align.

Storyboard To Video - Seedance 2.0

Back

This workflow turns a plain-language scene description into an 8‑panel storyboard and then animates that plan with Seedance 2.0. Inside the Previs Storyboard Generator group, your text is processed through LLM/image nodes (GeminiNode and OpenAIGPTImageNodeV2), with StringConcatenate assembling structured per‑panel prompts. The generated frames are arranged into a single contact sheet with ImageStitch, previewed via PreviewAny, and saved by SaveImage. This gives you a fast visual blueprint—shot order, camera angles, and composition—before you ever render a video.

When you’re happy with the board, switch to the Seedance Video group. There, ByteDance2ReferenceNode uses the storyboard image (selected through the Image input) as a visual guide to drive motion and style while generating the clip. You can monitor progress with PreviewAny and export the final result with SaveVideo. The workflow relies on ComfyUI’s group bypass toggles so you can iterate on the storyboard first, then hand it off cleanly to Seedance 2.0 for animation—keeping creative control through every step.