Quick Ideate to Storyboard to Animated Video (GPT Image 2 + Seedance 2.0)
A quick upload of your animated character reference art and description of the scene unlocks a super fast ideation sessions with GPT Image 2 and Seedance 2.0. Great for content on the go.
简介
This workflow turns a single character reference plus a short scene description into a quick storyboard and an animated clip. It chains OpenAIGPTImage1 to generate storyboard stills from your prompt, then feeds your chosen keyframe and reference art into ByteDance2ReferenceNode (Seedance 2.0) to produce motion, finishing with SaveVideo to render a playable file. LoadImage brings your character reference into the graph, while SaveImage lets you export and review the storyboard frames before you animate.
你需要提供
- 正在读取工作流
在 ComfyUI 中运行。 此工作流无法在本页面运行。你可以在 Comfy Cloud 中打开（已预先加载所需内容），或下载后在自己的 ComfyUI 中使用。
在 Cloud 中试用常见问答
You can animate directly from your reference image using ByteDance2ReferenceNode, but generating a storyboard keyframe with OpenAIGPTImage1 helps align composition, pose, and lighting before you commit to motion.
Use a clear, high-quality reference image, keep aspect ratios consistent between the reference, storyboard, and video, and increase the reference strength/weight in ByteDance2ReferenceNode if available. Also keep prompt wording consistent across iterations.
In ByteDance2ReferenceNode, adjust motion length/duration and any motion strength or guidance settings to balance stability and dynamism. In SaveVideo, set fps if exposed. Higher motion strength yields more dramatic movement but may drift from the reference.
This workflow is wired to the OpenAIGPTImage1 node for storyboard generation. Use the node as provided in the graph; if an updated GPT-Image node is available in your setup, you can swap it in while keeping the same overall pipeline.