Bytedance Text to Video
fal-ai/bytedance/seedance/v1.5/pro/text-to-video
Generate videos with audio with Seedance 1.5
Inference
Commercial use
Partner
Input
Additional Settings
Customize your input with more control.
Result
Idle
What would you like to do next?
Each 720p 5 second video with audio costs roughly $0.26. For other resolutions, 1 million video tokens with audio costs $2.4. Without audio, the price is 1.2 per millition tokens. tokens(video) = (height x width x FPS x duration) / 1024.
Logs
Seedance 1.5 Pro Text-to-Video
Generate broadcast-ready video clips with synchronized dialogue, sound effects, and music. Seedance 1.5 Pro generates all from a single text prompt. Seedance 1.5 Pro uses a dual-branch diffusion transformer to render video and audio in the same latent space, producing tight lip-sync and natural foley without post-production.
Use Cases
| Use Case | Why Seedance 1.5 Pro fits |
|---|---|
| Short-form drama | Generate 5–12 second scenes with dialogue, emotional performances, and cinematic camera moves — ready for TikTok, Reels, or YouTube Shorts. |
| Ad spots with voice-over | Create product or brand videos with synchronized narration, ambient sound, and polished visuals. No separate VO recording or audio sync needed. |
| Social teasers & trailers | Produce scroll-stopping previews with dramatic tension, music, and sound design baked in. |
| Product demos | Animate a product hero shot into a reveal sequence with motion, lighting shifts, and spatial audio. |
| Talking-head avatars | Generate realistic speaking characters with accurate lip-sync for explainers, virtual hosts, or localized content. |
| Storyboarding & pre-vis | Quickly visualize scenes before committing to live-action shoots — test framing, pacing, and dialogue delivery. |
| Music videos & lyric visuals | Pair generated visuals with synchronized audio for lo-fi music content, lyric videos, or ambient loops. |
Key Features
| Feature | Description |
|---|---|
| Native audio generation | Dialogue, sound effects, and ambient audio generated alongside video. Lip movements stay locked to speech; foley stays locked to action. |
| Cinematic camera work | Full camera grammar: pan, tilt, zoom, dolly, orbit, tracking shots, rack focus. Describe the move in your prompt and the model executes it. |
| Character consistency | Faces, clothing, and expressions stay stable across the clip, even when camera distance or angle changes. |
| Narrative coherence | The model understands story beats: it can hold emotional arcs, maintain scene logic, and keep multi-character blocking consistent. |
| High resolution | Output up to 1080p at 24 fps with smooth temporal consistency. |
Controls
| Parameter | Options | Notes |
|---|---|---|
`prompt` | Text (required) | Describe scene, action, dialogue, camera, and sound |
`aspect_ratio` | `21:9` · `16:9` · `4:3` · `1:1` · `3:4` · `9:16` | Default: `16:9` |
`resolution` | `480p` · `720p` | `480p` for faster iteration; `720p` for final output |
`duration` | `4`–`12` seconds | Default: `5` |
`generate_audio` | `true` / `false` | Default: `true` — set `false` for silent video |
`camera_fixed` | `true` / `false` | Lock the camera in place (tripod shot) |
`seed` | Integer | Set a value for reproducibility; use `-1` for random |
Start Frame / End Frame (only for Image-to-Video)
Anchor your video with specific start and end frames. Upload reference images to define the opening and closing compositions — the model generates all the motion, camera movement, and audio in between.
- Start frame: Sets the initial composition, lighting, and subject placement.
- End frame: Defines where the shot lands — useful for precise transitions or product reveals.
- Motion is generated, not interpolated: The model renders physics-aware movement in latent space, not frame blending.
Prompting Tips
Write your prompt like a shot description on a call sheet:
| Element | Example |
|---|---|
| Scene | "Rainy Tokyo alley at night, neon reflections on wet pavement" |
| Action | "A woman in a trench coat turns and walks toward camera" |
| Dialogue | `"I told you — we don't have much time."` (use quotes) |
| Camera | "Slow dolly-in ending on a close-up" |
| Audio/Foley | "Rain on metal, distant traffic, her heels on concrete" |
More tips:
- Be specific about camera behavior: "locked tripod," "handheld with subtle shake," "smooth orbit right," "drone pull-out"
- Include ambient sound: "room reverb," "crowd murmur," "wind through trees"
- For best coherence, keep to one location and 1–2 characters per clip
Specs
| Spec | Value |
|---|---|
| Max duration | 12 seconds |
| Max resolution | 1080p |
| Audio | Mixed dialogue + foley + score, 48 kHz AAC |
| Output format | MP4 (H.264) |
| Inference speed | ~30–45 s for a 5-second clip (varies by hardware) |