Seedance 2.5 packs some genuinely new tools — 30-second single-shot video, up to 50 references, green-screen and 3D motion control, and the ability to edit a clip without redoing it. But new features only help if you know how to use them. This guide is a plain-English, step-by-step walkthrough of how to use Seedance 2.5 and each of its standout features, plus what it costs and where to get it.
We'll start with the basic workflow, then go feature by feature with real steps, and finish with pricing and a FAQ. No jargon — just what to click and what to type.
The Basic Workflow First
Before the fancy features, here's the core loop that every Seedance 2.5 video follows. Once you know this, the rest is just add-ons.
1.Open the video tool. On Dreamina, open "AI Video" and select Seedance 2.5 as your model.
2.Pick your input type. Text-to-video (start from words), image-to-video (animate a still), or reference-to-video (match a style or subject).
3.Add references if you have them. Click the "+" to upload images, clips, or audio.
4.Write your prompt. Describe the scene in order — subject, action, camera move, lighting, mood.
5.Set length and aspect ratio. Choose your clip length and shape (16:9, 9:16, etc.) for where it'll be posted.
6.Generate, review, export. Check the result, tweak, and download.
That's the whole loop. Now let's make each new feature work for you.
How to Use Long Video (30s, 90s, 180s)
Seedance 2.5's headline feature is length, and it comes in tiers. Standard mode makes a single 30-second clip in one pass. There's also a 90-second draft option and a beta long-video mode that reaches 180 seconds — three full minutes.
Here's how to use it well:
Pick the right length for the job. Use 30 seconds for a finished ad or scene, the 90-second draft to rough out a longer idea cheaply, and 180-second beta only for lower-stakes long content while it matures.
Write the prompt as one evolving shot, not one moment. Because it's a continuous take, describe how the scene starts, develops, and ends. For example: "Camera slowly pushes in as the subject turns from the window and walks toward the door, light shifting from cool to warm."
Name the camera move. Say "slow push-in" or "low tracking shot," not just "cinematic." Specific motion gives you a real camera instead of drift.
Keep one clear action per take. Don't cram a whole story into one prompt. One clear beat per generation holds together better.
The payoff: no stitching seams, and characters and lighting that stay consistent the whole way through.
How to Use 50 References for Consistency
This is the feature that keeps a character or product looking the same across your video. You can load up to 50 references — images, video, audio, even 3D models — in one generation. But the trick isn't just adding them; it's telling the model what each one is for.
Steps to do it right:
Collect clean references first. Gather clear images of your subject from several angles — front, side, detail — plus any style frames or audio you want matched.
Upload them into your reference slots. Add up to 50, mixing images, clips, and audio as needed.
Label each one in your prompt. This is the key step. Say what each reference does: "Use reference 1 for the character's face, reference 2 for the product, reference 3 for the color palette." When the model knows each reference's role, it uses them correctly.
Reuse the same set across clips. For a series or campaign, keep the same reference set for every clip so the look stays consistent from shot to shot.
A simple way to organize: split your references into groups — identity (the character), environment (the set), style (color and mood), motion, and audio. Balanced groups prevent the model from getting confused.
How to Use R2V Motion Control
R2V (reference-to-video) is the feature that gives you director-level control over movement. Instead of only describing motion in words — which rarely works for complex moves — you feed the model a reference that shows the motion. This is where a lot of pros will spend their time.
How to use it:
Prepare a motion reference. This can be a short clip of real movement, green-screen footage, or a simple 3D white-model animation showing the path you want.
Upload it as an R2V reference. Add it alongside your other references.
Use green screen for clean subjects. Green-screen footage helps the model separate the subject and follow its movement precisely.
Use a 3D white model for staging. A rough white-box 3D scene lets you lay out the space, the blocking, and the camera path before spending a full generation. The model follows that structure.
Describe the interaction in the prompt. Add plain-language notes on how characters or objects should interact, and let the reference handle the exact motion.
The result is motion that looks intentional — like a planned shoot — instead of the random drift you get from text alone.
How to Use Local Editing (Fix One Thing)
This feature saves huge amounts of time. Instead of regenerating a whole clip to change one detail, you edit just that part and leave the rest untouched.
Steps:
Start from a clip you mostly like. Local editing is for a clip that's working except for one element.
Select the region or element to change. Point to the specific area — a background, a product, an outfit, a color.
Describe the change. Say what you want it to become — a different background, a new product color, a changed label.
Generate the edit. The model changes only that element and keeps the motion, camera, and lighting intact.
The best way to think about it: treat local editing as a surgical fix at the end, not a shortcut that lets you skip planning. Get the shot mostly right first, then use it to polish one detail or spin off variations — a different product color for each market, for example.
Tips That Apply to Every Feature
A few habits make all of this work better, no matter which feature you're using:
Write like a shot brief, not a caption. Give the model a subject, a motion over time, and a camera — not a still frame described in words.
Lean on references for continuity. The look carries through references, not just words.
Test small before you commit. Rough out a short draft to check camera logic and reference weights before running a full-length, higher-cost generation.
Reuse what works. Once a reference set gives a good result, keep it for the whole series.
If you want the full list of entry points and rollout timing before you dive in, our how to access Seedance 2.5 guide covers every route.
Seedance 2.5 Pricing and Free Credits
Now the money question. As of now, ByteDance hasn't published official pricing for Seedance 2.5. Here's what's known and what to expect.
Official pricing: not yet public. The model is billed per second of generated video, but exact rates go live when the Volcano Engine / BytePlus API opens (targeted for mid-July 2026). Until then, treat any specific figure as an estimate based on Seedance 2.0.
What to expect on Dreamina. Access will most likely run through Dreamina's existing subscription tiers, using a credit system. Based on Seedance 2.0, all paid plans tend to unlock the same features, with the main difference being how many credits you get. Since 2.5 makes longer 30-second and 4K clips, expect each generation to use more credits than 2.0 did.
Free credits. The Dreamina Seedance 2.5 page mentions free daily credits, so you'll likely be able to try it without paying. Just check the fine print — free tiers often come with queue limits, watermark rules, or export caps, so confirm those before using it for real work.
The practical move: start with the free daily credits to learn the workflow, then move to a paid tier only once you know it fits your needs.
Where to Access Seedance 2.5
Quick version, since the focus here is usage. Seedance 2.5 rolls out in stages: ByteDance's own apps first (Dreamina internationally, Jimeng in China), then CapCut for editors around mid-July, and the BytePlus API for developers, targeted for mid-July. If you're setting up, create your Dreamina account now so you're ready to use everything above the moment access opens.
If you'd rather skip the wait and the separate accounts, this site updates with new models as soon as they go live and offers free trial access — so keep an eye on seedance25.tools to try Seedance 2.5 the moment it's ready.
A Full Example, Start to Finish
Let's tie it all together with one realistic project: a 30-second product ad for a coffee brand, with a consistent product and a clean camera move.
Step 1 — Gather references. Collect three clear photos of the coffee bag (front, side, label detail), one brand color swatch, and one short clip of a slow push-in you like the motion of. That's five references, well under the 50 limit.
Step 2 — Upload and label. Add all five, then in your prompt say what each is for: the three bag photos pin the product, the swatch sets the palette, and the clip guides the camera motion via R2V.
Step 3 — Write the shot as one take. Something like: "A 30-second continuous shot. Slow push-in on a coffee bag on a marble counter in warm morning light. Steam rises from a fresh cup beside it. Camera holds, then gently pulls back to reveal a cozy kitchen." Notice it describes motion over time, names the camera moves, and keeps one clear beat.
Step 4 — Set length and shape. Choose 30 seconds and 16:9 for a website, or 9:16 if it's for Reels.
Step 5 — Generate and review. Check that the bag looks identical to your references, the motion is smooth, and the audio fits.
Step 6 — Fix one thing with local editing. Say the client wants the bag in a different color for a second version. Instead of redoing the whole ad, use local editing: select the bag, describe the new color, and generate. The motion, lighting, and kitchen stay exactly the same — you just get a second variant in a fraction of the time.
That single project uses long video, 50 references, R2V, and local editing together — which is exactly how these features are meant to work as a set, not in isolation.
Frequently Asked Questions
How do I use Seedance 2.5 for the first time? Open the AI Video tool on Dreamina, select Seedance 2.5, pick your input type (text, image, or reference), add any references, write a clear scene prompt, set your length and aspect ratio, then generate. That's the basic loop; the advanced features build on it.
How do I make a longer video in Seedance 2.5? Choose your length tier — 30 seconds standard, a 90-second draft, or the 180-second beta. Write the prompt as one continuous, evolving shot (describe how the scene changes over time) rather than a single moment, and name your camera move.
How do I keep a character consistent across clips? Use the reference system. Upload clear images of your character from several angles, label each reference's role in your prompt, and reuse the same reference set for every clip in the series.
How does R2V control work? You upload a motion reference — a real clip, green-screen footage, or a 3D white model — and Seedance 2.5 follows that path for movement and staging. It's far more precise than describing complex motion in words.
Can I edit a Seedance 2.5 clip without redoing it? Yes. Use local editing: select the one element you want to change, describe the new version, and generate. The model changes just that part and keeps the rest of the shot intact.
How much does Seedance 2.5 cost? Official pricing isn't published yet. It's billed per second, with exact rates coming when the BytePlus API opens in mid-July. Dreamina is expected to offer free daily credits and subscription tiers, with longer 4K clips using more credits.
Is Seedance 2.5 free to try? Likely yes. The Dreamina page mentions free daily credits, so you should be able to test it without paying — just check queue, watermark, and export limits on the free tier first.
Final Thoughts
Seedance 2.5's new features are powerful, but the difference between a rough clip and a professional one is knowing how to drive them: write for one continuous take, label every reference, guide motion with R2V instead of words, and use local editing as a final polish. Start with the free daily credits, learn the workflow on short drafts, and scale up once it clicks. The tools are here — the skill is in how you use them. Want to jump in and start creating with Seedance 2.5?