Motion Reference Guide
+ Timeline Control
This guide focuses on dance creators, product demo teams, sports editors, VFX artists, performance directors, and character-video creators. It explains how to prepare reference videos, green-screen motion clips, pose examples, action beats, timing notes, identity references, product operation clips, and performance descriptions — and how Seedance 2.5 interprets them.
What Problem Does Motion Reference Solve?
Traditional production tools often stop at preparation. Dance creators, product demo teams, sports editors, VFX artists, and performance directors can create blocking, images, scripts, or references, but still need time, budget, and specialist labor to turn those assets into a video that communicates motion, mood, and timing.
Seedance 2.5 changes that by making the reference material part of the generation workflow instead of treating the prompt as the only input. The value is speed plus direction: Dreamina can help convert guiding dance, product operation, performance clips, or pose references into usable AI video without requiring a full production team.
What Source Materials to Prepare
- Reference videos for motion: dance clips, product operation videos, sports footage, performance recordings — mp4/mov, [2-30]s, max 200MB
- Green-screen motion clips: isolated subject on green background for clean motion transfer
- Pose examples: static images showing key poses or body positions
- Action beats: timestamped descriptions of when each key action happens
- Identity references: face or full-body reference images for character consistency
- Product operation clips: how a product is held, used, or interacted with
- Performance descriptions: emotional arc, energy level, rhythm
How R2V, White Model, and Timeline Control Work Together
R2V (Reference-to-Video) uses your reference assets to guide generation. The model extracts motion, identity, style, or scene structure from each reference and applies them to the output.
White/Clay model renders strip away texture and lighting, giving the model a clear motion structure to follow without visual noise from the original footage.
Timeline control tells the model when each key action should happen within the clip, creating a beat map the model follows.
Step-by-step workflow
- Prepare and organize all reference materials (motion, identity, scene)
- Create clay/white render if needed from 3D blockout
- Define the timeline: list each key action with its timestamp range
- Write the prompt using the reference structure (@Video1 for motion, @Image1 for identity)
- Generate, review against the beat map, refine one variable at a time
Prompt Structure for Motion Reference
Common Mistakes to Avoid
- Too many references without roles: Each reference needs a declared purpose. If you upload 5 videos without explaining what each provides, the model guesses — and guesses wrong.
- Motion reference as style reference: If your motion video has strong lighting or color, the model may use it as a style guide instead of a motion guide. Use clay renders to isolate motion.
- No identity anchor: Without a face or body identity reference, character consistency breaks between shots.
- Judging without a goal: Review each generation against a specific checklist, not just personal taste. Define what "success" looks like before you generate.
Why Does Timeline Control Matter for Longer AI Videos?
If you are asking how to control timing and scene order in Seedance 2.5, you are really asking whether AI video can follow a plan instead of improvising every beat. Timeline control matters because longer AI videos need structure: when the subject appears, when the action happens, how the camera moves, and how the scene reaches its payoff. Without that plan, even a beautiful video can feel random.
The practical answer starts with a brief. Users should define the subject, action, timing, references, and what must not appear. This makes Dreamina Seedance 2.5 easier to evaluate because every output can be compared against a clear standard.
How Timeline Control Fits the Seedance 2.5 Upgrade
Seedance 2.5 is built for: longer generation, richer references, localized editing, R2V, multimodal inputs, and better control. Timeline control fits that story because production users need more than visual quality. They need scenes to happen in the right order, avoid off-brief elements, preserve character logic, and stay usable after the first generation.
How to Plan Timing Before Generating
Start with one focused use case. A good first test should not include every feature at once. Choose a short product video, a simple story beat, or one reference-driven scene. Define:
- What should happen (key actions)
- What should NOT happen (exclusions)
- How long each important moment should last (timestamps)
Then generate a first result in Dreamina and write down what succeeded and what failed.
If problems occur, diagnose one at a time:
- Timing off: Adjust scene order and action language
- Unwanted elements: Strengthen exclusion terms, simplify competing details
- Character drift: Add clearer identity references, reduce scene complexity
- Audio mismatch: Specify what should be removed or avoided explicitly
Timing Mistakes to Avoid
- Only positive instructions: Describe everything you want but also say what should NOT happen
- Overloading the prompt: Too many styles, camera moves, and references at once overwhelm the model
- Judging without a goal: Professional users need repeatability, not magic. Set a standard before you generate
When Do Creators Need Timeline Control Most?
- Social creators: Need a clean clip without random text, correct timing for a reveal
- Marketers: Need timing that matches a product reveal or call-to-action
- Filmmakers: Need a scene to follow storyboard order
- Game teams: Need trailer beats to land at specific moments
In each case, the user is asking Seedance 2.5 to follow direction, not simply produce motion. This makes timeline control one of the most important workflow skills for professional use of Seedance 2.5.