Skip to main content
Seedance 2.5 Draft mode renders a fast, low-cost 480p preview before you commit to a full 1080p render, so you can iterate on a prompt, a first frame, or a set of reference images without paying for full-resolution generations. Draft mode is available on three Seedance 2.5 workflows. Select Seedance 2.5 Draft as the model on the Text to Video, First-Last-Frame to Video, or Reference to Video node:
  • Text to video (T2V): iterate on the prompt and the duration
  • Image to video (I2V): iterate on the first frame
  • Reference to video (R2V): iterate on the reference images and the storyboard
Resolution is locked to 480p in draft mode, and each run returns both the preview and a draft_task_id that the final render uses.
To use the Partner Nodes, you need to ensure that you are logged in properly and using a permitted network environment. Please refer to the Partner Nodes Overview section of the documentation to understand the specific requirements for using the Partner Nodes.
Make sure your ComfyUI is updated.Workflows in this guide can be found in the Workflow Templates. If you can’t find them in the template, your ComfyUI may be outdated.If nodes are missing when loading a workflow, possible reasons:
  1. You are not using the latest ComfyUI version (Nightly version)
  2. Some nodes failed to import at startup

​
How the two stages work

Every draft workflow runs in two stages:
  1. Create the draft: run the Seedance 2.5 node with the Seedance 2.5 Draft model to get the 480p preview and its draft_task_id, which the Save Text node writes to a .txt file in the output folder. Rerun with a different prompt or input until the draft looks right.
  2. Render the final: copy the draft_task_id from that .txt file into the draft_task_id field on the ByteDance Seedance 2.5 Draft to Final Video node and run it. The node renders the 1080p final and reuses what the draft was created with, including the prompt, the input, the duration, the aspect ratio, and the audio setting, so there is nothing to re-enter.
Each template puts the two stages in separate node groups, named Stage 1: Create and preview the draft video and Stage 2: Render the final video from the draft task ID. Leave Stage 1 bypassed once you are happy with the draft, because a second run of the draft node creates a new draft instead of reusing the one you reviewed.

​
Workflow templates

Run Draft Text-to-Video on Cloud

Try the Draft Text-to-Video workflow instantly on Comfy Cloud.

Download Draft Text-to-Video workflow

Download the workflow JSON.

Run Draft Image-to-Video on Cloud

Try the Draft Image-to-Video workflow instantly on Comfy Cloud.

Download Draft Image-to-Video workflow

Download the workflow JSON.

Run Draft Reference-to-Video on Cloud

Try the Draft Reference-to-Video workflow instantly on Comfy Cloud.

Download Draft Reference-to-Video workflow

Download the workflow JSON.

​
Input materials

The draft image-to-video and reference-to-video templates use these sample inputs:

First frame

Download sample first frame

Reference image

Download sample reference image

Storyboard

Download sample storyboard

​
Tips

  • Run only the stage you need: select the Stage 1 nodes and press Ctrl+B (Cmd+B on macOS) to bypass them, then select the Stage 2 nodes and press Ctrl+B again to enable them, so only the final render runs
  • Keep the draft task ID: the Save Text node writes it to a .txt file in the output folder, and Stage 2 reads it from there
  • Drafts stay renderable for 7 days after they are created, so render the final within that window
See Seedance 2.5 for the model overview, the standard workflows, and video editing.