What Seedance 2.5 does
Seedance 2.5 is one model with four modes. Frames animates a start image and reaches an end image if you give one. Reference builds a video from your reference images, videos and sounds. Edit retouches an existing video. Text starts from the prompt alone. On the canvas, the mode follows what you connect to the node.
4–30 s
per clip, in 1-second steps
720p
by default, or 480p
30 images
as references, plus 10 videos and one audio clip
The four steps
Step 1
Pick the mode that fits what you have
In Generate, choose “Seedance 2.5”, then the Frames type (the default), Reference, Edit or Text. On the canvas, add a model node, choose Seedance 2.5 and connect your inputs: a start image for Frames, references for Reference, a video on the video input for Edit.
Step 2
Add the images, videos and sounds
In Frames: the start image, and an end image if needed. In Reference, in Generate: up to 30 images, 10 videos and one reference audio clip; audio on its own is refused, it needs at least one image or video with it. In Edit: the video to retouch, 30 seconds at most in Generate; only the first one is edited.
Step 3
Write the prompt and cite your images
The prompt is required in every mode. In Generate, type “@” to insert a reference image in the sentence: the app turns it into “the first reference image”, “the second”… when it sends the request. Describe the action, the camera move and the sound you expect.
Step 4
Set length, resolution and format, then launch
Length from 4 to 30 seconds (Generate starts at 4 s), 720p or 480p, format auto, 21:9, 16:9, 4:3, 1:1, 3:4 or 9:16. Auto gives 16:9 in Text and Reference; in Frames the framing follows the image, in Edit the source video. In Generate, sound is on by default and can be turned off. The cost is shown before you launch.
The settings at a glance
| Setting | Values | Tip |
|---|---|---|
| Mode | Frames, Reference, Edit, Text | Frames by default in Generate; on the canvas, the mode follows the connected inputs |
| Length | 4 to 30 s | In 1-second steps; in Edit, the length follows the source video |
| Resolution | 720p, 480p | 720p by default; 480p costs less while you iterate |
| Format | auto, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 | Auto = 16:9 in Text and Reference |
| Sound | On or off | On by default in Generate; turning it off does not change the price |
| Video bitrate | Standard, high | High by default, in advanced settings; mp4 or mov output |
Ideas to try
A packshot that comes alive
Frames mode: the product photo as start image, a prompt describing the camera move, 8 to 10 seconds at 720p.
A transition between two images
Frames mode with a start and an end image, generated earlier in the same graph.
A character held for the whole shot
Reference mode: the character, outfit and set as reference images, up to 30, and a prompt that stages them.
Retouch an existing clip
Edit mode: the source video and the edit instruction. The output keeps the length of the source.
Frequently asked questions
- Seedance 2.5 or Seedance 2.0: which one should I use?
- Seedance 2.5 goes up to 30 seconds, takes 30 reference images and edits video, at 720p at most. The Seedance 2.0 family stops at 15 seconds: 2.0 Fast and 2.0 Mini are the fast, economical versions, at 720p at most, and Seedance 2 Reference to Video reaches 1080p. For a long or heavily referenced shot, 2.5; to iterate fast, 2.0 Fast or Mini.
- What resolution does Seedance 2.5 output?
- 720p by default, or 480p. For 1080p in the same family, Seedance 2 Reference to Video and Seedance 1.5 Pro offer it, on clips of 15 and 12 seconds at most.
- Does Seedance 2.5 generate sound?
- Yes, by default: the video comes out with a generated soundtrack. You can turn it off before launching; it does not change the price. In Reference mode you can also give one reference audio clip in Generate, with at least one image or video.
- How is Seedance 2.5 billed?
- Per second of video produced, by resolution: 480p costs less than 720p. When a video goes in (reference or Edit), its seconds are added at a lower per-second rate. Sound and bitrate do not change the price. The cost is shown before you launch.
- Can I fix the camera or the seed with Seedance 2.5?
- No: Seedance 2.5 has neither a fixed-camera option nor a seed (the node’s Seed field has no effect on this model). Describe the move you want in the prompt (“static camera”, “slow dolly in”). In the Seedance family, only 1.5 Pro offers a fixed camera, on the canvas.
- Where can I use Seedance 2.5 on Beemm Vision?
- In Generate (“Seedance 2.5”), on the canvas as a model node, in batches with Scale, and from an AI assistant through the MCP server. There is no ByteDance account to open.
Try Seedance 2.5
Open the studio, pick Seedance 2.5 and launch a first 5-second clip at 480p to check the motion.