Seedance 2.5
by ByteDance
Directed moves with exact first and last frames.
Up to thirty seconds, and your song rides on top.
Key features
First and last frame
Set where a move starts and where it ends, and the model walks the camera between your two images.
Up to thirty seconds
Four to thirty seconds per take, or automatic — long enough for a full verse phrase in one render.
Audio-driven motion
Feed the track and the picture follows it — beat, sway and lip movement move with your song.
Your master stays the master
Dimly renders Seedance with its own audio generation off and muxes your original song back, so the track is never reinvented.
Rich references
First/last frames, stills, video clips and audio — up to thirty seconds of each — shape every take.
Honest baseline
480p and 720p output make it the fast, economical pass for exploration before a 2K or 4K finish.
Technical specifications
Seedance 2.5
First/last-frame video model, served through dimly
4 input kinds
First/last frames, images, clips and audio
4 to 30s
Or automatic — the model picks the length
480p / 720p
The economical pass before a 2K/4K finish
Ratios
Vertical-first, matched to the brief
Audio in
Motion follows the track you attach
Silent out
Dimly muxes your original song back after render
Beat-locked
Sway and lip movement land on the beat
Prompt examples
Corridor push
The camera slides down a hotel corridor as a door swings open toward the neon street
Edit promptVinyl spin
The record drops, the groove catches, and the camera pulls back through the den
Edit promptUse cases

Verse-long moves
Thirty seconds covers a verse phrase in one take, so a slow push no longer needs stitching.

Exact endpoints
Match cuts, whip transitions and reveal shots that have to land on a specific frame.

Choreography checks
Block a move between two poses, watch it, adjust the endpoints, watch it again.

Beat-synced performance
Feed the track and the performance sways on it — useful for lip-adjacent chorus work.

Cheap exploration
720p renders keep the explore pass fast, so the doubtful ideas get tried too.

Finish on H3
Explore the move on Seedance, then hand the approved world to H3 for the character-locked master.
Overview
Seedance 2.5 is the first-and-last-frame model in dimly's picker. Give it two images and it walks a camera between them, from four up to thirty seconds, at 480p or 720p. Still images, clips and audio can ride along as references, and when audio is attached the motion follows it — beats, sway and mouth movement land on your track.
Why the silent render is a feature
On dimly, Seedance renders with its own audio generation switched off. Your original song is muxed back onto the finished take afterwards, so what you hear is always your master — not a model's approximation of it. The audio input still matters: it drives the picture even while the output stays silent.
Where MiniMax H3 is the better pick
When identity has to hold across many shots — one artist, one world, three shots around a chorus — H3's reference-to-video is the tool, with native stereo and up to 4K. Seedance is the directed-move specialist and the economical exploration pass: exact endpoints, longer runway, cheaper credits.
Seedance 2.5 on dimly
Seedance is available on every plan at 480p and 720p. Like every dimly generation, the credit cost is quoted and confirmed before the render runs, and takes land in the project workspace next to their H3 siblings for side-by-side comparison.
Simple pricing
Plan your video for free, then pick a plan that fits your release schedule.
Starter
A steady month of chorus cuts, or one shorter song.
- 600 credits, valid for 30 days
- ≈ 8 chorus cuts (15 s)
- or one song up to 2:20
- Every model and resolution · 3 renders at once
Pro
One full song a month with room to retake, and bigger sizes when a release needs them.
Credits a month
- 1,000 credits, valid for 30 days
- ≈ 14 chorus cuts (15 s)
- or 1 full song (3:45)
- Every model and resolution · 4 renders at once
Creator
Several full songs a month, for a steady release schedule.
- 5,000 credits, valid for 30 days
- ≈ 72 chorus cuts (15 s)
- or ≈ 5 full songs (3:45)
- Every model and resolution · 8 renders at once
Full song, chorus cut, lyric video or visualizer · Watermark-free exports on every plan · Export rights depend on your source permissions and provider terms
Credit packs
Pay once, no subscription. Made for one AI effect or one song; if you will make more, a plan above is the better rate.
Single
One AI effect video, like Hotel Lobby or AI Rap Duo.
- 250 credits, valid for 1 year
- 1 AI effect video (Hotel Lobby, AI Rap Duo)
- or ≈ 3 chorus cuts (15 s)
- Exports carry the Dimly watermark
- No subscription, never renews
Song
One full music video (a song up to 3:45), or four AI effects.
- 1,000 credits, valid for 1 year
- 4 AI effect videos (Hotel Lobby, AI Rap Duo)
- or 1 full song (3:45)
- Watermark-free exports
- No subscription, never renews
Studio
A full song in 2K with room for retakes, or two songs.
- 2,000 credits, valid for 1 year
- 8 AI effect videos (Hotel Lobby, AI Rap Duo)
- or ≈ 2 full songs (3:45)
- Watermark-free exports
- No subscription, never renews
Estimates use MiniMax H3 at 768P with 5-second shots (a 15-second chorus cut counts as three), a keyframe per shot, two reference sheets and no retakes; a full song means 3:45, the typical length of the songs people bring. Higher resolutions and the other models cost more per video. You see the exact quote before anything renders, and failed renders are refunded.
FAQs
Dimly's first-and-last-frame video model: two images set where a move starts and ends, and the take runs four to thirty seconds at 480p or 720p with your song muxed back over the render.
By design. Dimly renders Seedance with its own audio generation off and muxes your original track back afterwards, so the audio you ship is your master, never a model's reinvention.
Yes — audio is an input, not an output. Attach the track and motion follows it: beats, sway and mouth movement land where the song puts them.
Every paid plan. Renders come out at 480p or 720p, which keeps the exploration pass fast; finish character-locked masters on MiniMax H3 at up to 2K or 4K.
Four to thirty seconds, or automatic length. That is usually a full verse phrase — long enough for a single continuous camera move without stitching.
First and last frames, still images, video clips and audio, up to thirty seconds each, so a take can be anchored on your existing footage as well as generated frames.
They are excerpts of Seedance 2.5 renders published for the launch: the circus, football and missing-sock films BytePlus released with Dreamina Seedance 2.5, and the text-to-video, image-to-video and multi-language examples on fal. They are shown as examples of the model, not as Dimly renders, and they play with the sound the model generated; a take made in Dimly comes back silent with your song laid over it.
Other models
See all modelsMiniMax H3
MiniMax
Reference-driven video with native stereo sound. Your track drives the picture, 768P up to 4K.
Kling 4.0
Kuaishou
Kuaishou's long-take video model: up to thirty seconds in one pass, ten keyframes and Omni Reference with up to fifteen inputs.
MiniMax H3 Image
MiniMax
MiniMax's open-weights image model: up to sixteen reference images, native 4K and Chinese or English lettering, in about twenty seconds.
Seedream 5.5
ByteDance Seed
ByteDance Seed's image model: natural skin and light, edits that leave the rest of the frame alone, and dense small text that stays legible.











