Seedance 1 is ByteDance's first-generation video model — smooth natural motion, six aspect ratios including ultra-wide 21:9, native audio, and a unique end-frame option for interpolation between two anchor images. Reliable workhorse for video generation.
Seedance 1 (Seedance 1.5 on the public API) is ByteDance's first-generation video model — smooth natural motion, strong character coherence across the duration of a clip, and a particularly wide aspect ratio range that includes ultra-wide 21:9 cinematic frames.
Text-to-video, image-to-video, and a unique interpolation mode (start image + end image generates motion between them). Durations from 4 to 12 seconds, resolutions 480p or 720p, native audio generation, and a camera_fixed toggle for locked-off shots.
4–12s
Duration range
6
Aspect ratios (incl 21:9)
Yes
Native audio + interpolation
Seedance 1 produces smooth, coherent motion without the jittery / morphing artifacts that plague some other video models. Characters maintain identity across the clip duration, environments stay stable, and motion arcs read naturally.
Six aspect ratios including 21:9 ultra-wide — most video models max out at 16:9. Useful for cinematic landscapes, hero shots, and any output destined for widescreen edits or letterboxed reveals.
generate_audio=true layers a synced soundscape onto the clip — ambient sound, diegetic effects, basic music cues. Toggle off for clean tracks you'll score in post.
camera_fixed=true forces a locked-off shot — no camera motion, only subject and environment animation. Useful for product spots, talking heads, and any composition where camera movement would be distracting.
Each example shows the exact prompt that produced the result. Copy any prompt with one click.
720p · 21:9 · 5s · audio on
A black 1970s muscle car drives through a misty desert canyon at dawn, golden sun breaking over the cliff edge, slow tracking shot from the rear, cinematic warm color grade, distant engine rumble
21:9 ultra-wide is Seedance's standout aspect — most video models cap at 16:9. Pair with cinematic language ("warm color grade", "cliff edge") and explicit sound cue ("distant engine rumble") and the audio pass picks it up convincingly.
720p · 16:9 · 5s · audio on
A woman in a flowing white dress walks slowly across a sunlit Mediterranean rooftop overlooking the sea, a gentle breeze catches the fabric, the camera tracks alongside her at hip height, soft ambient sounds of waves and distant seabirds
"Camera tracks alongside her at hip height" is a clean tracking-shot recipe — name angle + motion together. Seedance maintains character identity and fabric movement coherently across the 5 seconds. End with explicit ambient cues for the audio pass.
720p · 1:1 · 5s · audio on
Macro slow-motion of a cocktail being poured over crushed ice in a crystal coupe, mint sprig floating on the surface, golden bar light catching the glass, premium spirits commercial mood, subtle pouring sounds
1:1 square at 5s is product social sweet spot. Macro slow-motion product shots reward Seedance's smooth motion behavior — droplet physics, ice catching light, fabric draping all read clean. "Subtle pouring sounds" steers the audio toward minimal not overstuffed.
720p · 9:16 · 5s · audio on
Sunset surfer paddles out through gentle rolling waves toward the horizon, vertical handheld shot from waist deep in water, golden hour light bouncing off the spray, atmospheric ocean ambient sound
9:16 vertical at 720p for mobile-social. "Vertical handheld shot from waist deep in water" is a specific perspective — Seedance composes the framing accordingly. "Atmospheric ocean ambient sound" steers the audio to gentle, not crash-heavy.
21:9 is Seedance's distinctive aspect — most video models cap at 16:9. Use it for cinematic landscapes, hero reveals, ultra-wide establishing shots, anything destined for widescreen edit. It composes differently than 16:9 — pictorial / horizon-driven scenes shine here.
Native audio responds well to explicit sound cues at the end of the prompt — 'distant engine rumble', 'soft ambient ocean', 'subtle pouring sounds'. Generic 'with audio' doesn't help. Name what you want and Seedance layers it accordingly.
When camera movement would be distracting (product spots, talking heads, locked-off comps), set camera_fixed=true. Subject and environment still animate; only the camera holds. Cleaner for advertising and interview-style content.
Pass an image_url AND an end_image_url to generate motion between two anchor frames. Useful for narrative beats where you have the start and end shots but need the in-between motion. Few other models offer this pattern natively.
5 seconds is the production sweet spot for prompt iteration — long enough for one beat, short enough to be credit-cheap. Push to 8-12 seconds for multi-beat sequences or atmospheric holds. Stuffing too much into 5s produces compressed motion.
"Walks slowly across the rooftop", "the breeze catches the fabric", "the camera tracks alongside" — concrete verbs and subjects beat 'enjoying the view'. Seedance has a clear motion target to execute when the prompt provides specifics.
| Setting | Values | Notes |
|---|---|---|
| Mode | Text-to-video · Image-to-video · Image-to-image-with-end-frame | End frame mode generates motion between start and end images. |
| Duration | 4 · 5 · 6 · 7 · 8 · 9 · 10 · 11 · 12s | Credit cost scales with duration. 5s is the production sweet spot. |
| Resolution | 480p · 720p | 720p is the default. 480p for cheap iteration. |
| Aspect ratio | 21:9 · 16:9 · 4:3 · 1:1 · 3:4 · 9:16 | Six aspects including ultra-wide 21:9. |
| Camera fixed | Boolean (default false) | Locks the camera off. Subject and environment still animate. |
| Generate audio | Boolean (default true) | Layers synced ambient + diegetic sound. Toggle off for clean tracks. |
| Seed | Integer (-1 for random) | Pin for reproducibility. |
Seedance 1 is the original first-generation model — smooth motion, reliable output, six aspect ratios. Seedance 2 is the newer iteration with stronger fidelity and additional features. Both stay live; Seedance 1 is the dependable workhorse, Seedance 2 is for tasks where the newer aesthetic suits better.