Skip to content
BytePlusBytePlusBytePlus ModelArk for ComfyUI

Image to Video - Seedance 2.5 and 2.0

Template
Image to Video - Seedance 2.5 and 2.0, in the ComfyUI template browser under BytePlus
Workflow file
Image to Video - Seedance 2.5 and 2.0.json
BytePlus nodes
Seedance 2.5 First-Last-Frame to Video ×2
Keys needed
ModelArk API key
Also uses
Load Image, MarkdownNote, Save Video (built into ComfyUI)

What it makes

Section titled “What it makes”

Turn your image into a short video with either model. Both branches use the same image as first_frame, with separate prompts and saved results. Seedance 2.5 starts active; Seedance 2.0 starts bypassed so the first run generates only one video.

The workflow

Section titled “The workflow”
One Load Image feeds two First-Last-Frame video nodes, with Seedance 2.5 active and Seedance 2.0 bypassed
Choose one model or enable both branches to compare their results.
  1. Load Image supplies the first frame to both branches.
  2. First-Last-Frame to Video in the upper group runs Seedance 2.5 at 720p for five seconds, with audio.
  3. Save Video displays and saves the Seedance 2.5 result.
  4. The same First-Last-Frame node in the lower purple group runs Seedance 2.0 at 720p for five seconds, with ratio set to adaptive.
  5. A separate Save Video displays and saves the Seedance 2.0 result.

Run it

Section titled “Run it”
  1. Add your ModelArk key in Settings → BytePlus (API keys).
  2. Open Image to Video - Seedance 2.5 and 2.0 and upload your image.
  3. Edit the motion prompt in the Seedance 2.5 node, then run.
  4. To compare models, copy your prompt to the Seedance 2.0 node. Select both purple nodes and press Ctrl+B to enable them, then run again.
  5. Keep the first branch’s inputs and fixed seed unchanged to reuse its cached result. To run only Seedance 2.0, bypass both Seedance 2.5 nodes first.

Settings worth changing

Section titled “Settings worth changing”
  • Prompt: describe subject motion, camera motion and sound. The shipped prompt animates an illustrated character waving with a small head tilt; replace it for your own image.
  • Resolution: both branches start at 720p and also offer 1080p.
  • Duration: both branches start at five seconds; each model exposes its supported range.
  • Seed: both start at 42 with fixed. Change it deliberately for another variation.
  • Last frame: optionally connect a second image for a first-to-last-frame transition.
  • Audio: toggle generated sound on each branch.

Example result

Section titled “Example result”

This example uses the same illustrated character, motion prompt and fixed seed for both models. With the square image and 720p setting, both saved videos are 960 × 960, 24 fps and 5.056 seconds long, with audio. Seedance 2.5 shows a wave and a head tilt; Seedance 2.0 mainly tilts its head in the sampled frames. The models do not produce identical motion from the same seed.

ComfyUI shows successful square videos from Seedance 2.5 and Seedance 2.0 in separate Save Video nodes
Both models animate the same first frame.
Seedance 2.5 on the left and Seedance 2.0 on the right, resized for comparison. Audio is from Seedance 2.5.

Time and cost

Section titled “Time and cost”

Each enabled model is a separate billed generation. The template starts with one branch active. Re-running unchanged inputs with fixed seeds reuses ComfyUI’s cached outputs. Changing the image, prompt or seed requests another generation.

Section titled “Related”