Seedance 2.5 · Reference to Video

ABOUT THE WORKFLOW

Make a Face Follow Motion With Matched Audio Upload a face image, a motion video, and an audio track. Seedance 2.5 maps the motion onto your face and matches the mouth movement to the audio. The result comes back as a video with the synced sound on the track.

Partner node. This workflow calls an external API, so each run uses credits from your API wallet. No API key needed. Floyo handles the connection.

Model

• Seedance 2.5 by ByteDance. Announced 23 June 2026, launched July 2026. Running in reference-to-video mode with multi-modal inputs: image, video, and audio combined in one generation. Maps motion from a reference clip onto a face image while matching mouth movement to a separate audio track. Closed model with no open weights.

HOW IT WORKS

Step 1. Upload your face image A clear photo of the face that will appear in the output. Front-facing with good lighting works best. Works great with: headshots · character portraits · actor photos · avatars

Step 2. Upload the motion video A clip showing the body and head movements the face should follow. Dance, gesture, talk, nod.

Step 3. Upload the audio The audio track the mouth movement will match. Speech, narration, or dialogue.

Step 4. Hit run and download The model maps the motion onto the face, matches the mouth to the audio, and saves the clip under video/Seedance. Ready for: Premiere · DaVinci Resolve · CapCut · After Effects

First time? Leave every setting as-is. The defaults (720p · 5 seconds · 16:9 · audio on · mp4) are the right starting point for almost everyone.

RECOMMENDED SETTINGS

Quick-start guide. Find the goal that matches yours and copy the settings.

• Standard clip (most people) — 720p · 5 seconds · 16:9 · audio on · mp4. The right starting point for almost everyone.

• Mouth movement is off-beat — Check that the audio timing matches the motion video. The model matches the mouth to the audio, not to the motion, so misaligned timing between the two produces a mismatch.

• Need a different shape — Switch the aspect ratio dropdown. 16:9, 9:16, and 1:1 are available.

• Need a longer clip — Raise the duration. Duration and resolution together drive cost.

• The face does not look like the reference — Try a cleaner, front-facing reference image. Side profiles and extreme angles weaken identity preservation.

• Want to change the prompt — The default prompt tells the model to follow the motion and match the audio. Rewrite it to match your subject and inputs, keeping the numbered references so the model binds each input correctly.

Prompt: Keep the numbered references. "The person in image 1 follows the movement in video 1 and matches the audio from audio 1" is the structure. Change "person" to match your subject, but keep the "image 1," "video 1," and "audio 1" references so the model knows which input is which.

LEARN

📹 Videos

• Intro to Floyo

• ComfyUI 101 Free Course ft. Sebastian Kamph

• Floyo 101 for Team Collaboration

✨ Quick links

• Help docs

• LoRA Training guide

• Discover more workflows

USE CASES

🎤 Talking Head Videos Make a portrait deliver a speech or narration with matched mouth movement from a separate audio track.

🎭 Character Animation Give a still character face natural head motion and matched dialogue from a separate performance.

📢 Multilingual Dubbing Match a face's mouth movement to audio in a different language while the body motion stays the same.

📱 Social and Marketing Turn a headshot and a voice recording into a short talking video for Reels, TikTok, or Shorts.

WHAT WORKS BEST / WHAT TO AVOID

✅ Works great

• Front-facing, well-lit face images

• Motion videos with clear head and body movement

• Audio that matches the motion video's timing

• Clips of 5 seconds

⚠️ May produce softer results

• Side-profile or heavily occluded face images

• Audio and motion that are misaligned in timing

• Fast head turns where the face goes off-angle

• Complex speech with rapid articulation changes

FAQ

What is Seedance 2.5 reference to video? Seedance 2.5 reference-to-video mode takes three inputs: a face image, a motion video, and an audio track. The model maps the motion onto the face and matches the mouth movement to the audio, producing a video where the face follows the performance and speaks the words. Image, video, and audio are processed together in one generation.

How does the audio matching work? The model reads the audio waveform and matches mouth shapes to it while mapping head and body motion from the reference video. The audio drives the mouth; the video drives the body. They are synchronized in the same generation pass.

Can I use any audio file? Yes. Speech, narration, singing, or dialogue. The model matches mouth movement to whatever audio you provide. Quality depends on how clean the recording is and how well the timing matches the motion video.

What is the difference between this and the Seedance 2.5 image-to-video workflow? Image to video animates a still from a prompt with no motion reference and no audio input. This workflow takes all three: a face, a motion source, and an audio track, and produces a result with matched mouth movement. Use image to video when you want the model to choose the motion, and this one when you want to control the motion and the dialogue.

Is Seedance 2.5 open source? No. Seedance 2.5 is a closed model with no published weights. It is reached through the ByteDance Seed API.

What is the difference between this and Kling 2.6 Motion Control? Both map motion from a reference onto a character. Kling 2.6 uses 3D face and body reconstruction for full-body motion transfer with the source video's audio. This workflow adds a separate audio input for explicit mouth matching, so you can control the dialogue independently from the motion. Pick Kling for full-body motion transfer with the original audio, and this one when you need a different audio track matched to the mouth.

How to run Seedance 2.5 reference to video online? You can run Seedance 2.5 reference to video online through Floyo. No installation, no setup, no API key to wire up. Open the workflow in your browser, upload a face, a motion video, and an audio track, and hit run. Free to try.

WHY FLOYO?

Floyo is the only platform with team collaboration for ComfyUI in the browser. You run workflows with no install. You share run history, assets, and models across your team. You pay only when you generate. Floyo supports open-source and closed-source models.

A designer runs an edit and likes the result. A teammate opens that exact run from shared history and keeps going. No file handoffs. No version confusion.

For studios and enterprise teams, Floyo adds private workspaces, pooled resources, and a team usage dashboard. Other ComfyUI cloud tools run for one person at a time. Floyo runs for the whole team, with transparent per-generation costs.

Ready to try it? Upload a face, a motion video, and an audio track, and run it. The settings are already set.

→ Launch Workflow, Free

Questions? Watch the free course or check the FAQ above.