업데이트
What is SeeDance 1.0?
SeeDance 1.0 is ByteDance's first-generation AI video generation family, served on the UnificAlly API as three models with different jobs. bytedance/seedance-1.0 handles text-to-video and image-to-video at 480p or 720p. bytedance/seedance-1.0-pro does one thing: first-and-last-frame video between reference images, at 480p, 720p, or 1080p. bytedance/seedance-1.0-pro-fast runs text-to-video and image-to-video faster, also up to 1080p. Every model in the line outputs 4 to 12 second videos.
The family stands out for physical realism — motion that respects weight and momentum, even in large-scale movements like sports or crowd shots — and a wide dynamic range in lighting. An integer seed makes runs reproducible: re-sending the same inputs with the same seed returns the same result, which is useful for A/B variants and locking in a video you want to deliver. A camera_fixed flag keeps the virtual camera still when you want motion in the scene but not in the framing.
Key features of SeeDance 1.0
Four features cover what teams build with the 1.0 line.
Text-to-video and image-to-video
The base model turns a prompt of up to 5,000 characters into a video across six aspect ratios (1:1, 3:4, 4:3, 9:16, 16:9, 21:9), or animates a single reference image passed as first_frame_url with an optional motion prompt.
First-and-last-frame interpolation on Pro
bytedance/seedance-1.0-pro generates the motion between two reference images. Pass first_frame_url, last_frame_url, or both; an optional prompt describes the transition. This is the only mode Pro runs, and the only first/last-frame path in the 1.0 line.
4–12 second videos, 480p to 1080p
Every model takes any whole-number duration from 4 to 12 seconds, converted to frames at 24 per second. The base model outputs 480p or 720p; Pro and Pro Fast add 1080p.
Seed and fixed-camera control
Pass an integer seed to make runs reproducible: same inputs, same seed, same result. Set camera_fixed to true to keep the virtual camera still while the scene moves. Both fields work on all three models.
What SeeDance 1.0 is best for
Short ads with locked aspect ratios
Six aspect ratios match each social channel native. 9:16 for Reels, 1:1 for feed, 16:9 for YouTube.
Hero shots via interpolation
First-and-last-frame mode on Pro produces controlled motion between two known compositions, up to 1080p. Useful for product reveals.
Quick concept videos
480p drafts at 4 seconds are cheap enough for client review and concept testing in volume.
Storyboard animatics
Feed two boards to Pro as first and last frames and get the motion between them. Useful pre-production for film and animation.
Social-first content at low cost
Good when 720p is enough and the placement does not need a hero render.
A/B testing motion treatments
Hold the prompt and seed constant and vary one field at a time to see how a concept reads in vertical vs widescreen.
SeeDance 1.0 variants
Three models make up the 1.0 line, each with its own generation path.
SeeDance 1.0
The base model. Text-to-video from a prompt across six aspect ratios, or image-to-video from a first_frame_url. Output is 480p or 720p, and it is the cheapest way into the family. It does not take a last frame.
SeeDance 1.0 Pro
First-and-last-frame only. At least one of first_frame_url or last_frame_url is required, and a prompt-only call is rejected. In return it goes up to 1080p and is the only model in the line that interpolates between two frames.
SeeDance 1.0 Pro Fast
Text-to-video and image-to-video like the base model, tuned for speed, with the same 1080p option Pro has. Use it for quick turnarounds and batch jobs. Like the base model, it does not take a last frame.
SeeDance 1.0 use cases
Generate a short ad for a single platform by picking the right aspect ratio at the call site (9:16 for Reels, 1:1 for feed, 16:9 for YouTube) and running a 5-second text-to-video on the base model. Build a controlled product reveal by sending your opening and closing stills to bytedance/seedance-1.0-pro as first_frame_url and last_frame_url and letting the model handle the in-between motion. Animate existing key art by passing it as first_frame_url on the base model with a short motion prompt. Lock a chosen result for delivery by re-sending the same inputs with the same seed on the final render.
SeeDance 1.0 API examples
Call SeeDance 1.0 from any language by POSTing to /v1/tasks. The same endpoint serves all three models; Pro swaps the bare prompt for first_frame_url and last_frame_url. Full parameter docs live at docs.unifically.com/models/video/bytedance/seedance-1.0.
curl -X POST https://api.unifically.com/v1/tasks \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"model": "bytedance/seedance-1.0",
"input": {
"prompt": "A butterfly landing on a flower in slow motion",
"aspect_ratio": "1:1",
"duration": 5,
"resolution": "720p"
}
}'
Successful submission returns a task_id. Poll GET /v1/tasks/<task_id> or set a callback_url on the request to receive the finished result.
FAQ
SeeDance 1.0에 대해 자주 묻는 질문
SeeDance 1.0 is ByteDance's first-generation video family. UnificAlly serves three models: bytedance/seedance-1.0 for text-to-video and image-to-video at 480p or 720p, bytedance/seedance-1.0-pro for first-and-last-frame video at up to 1080p, and bytedance/seedance-1.0-pro-fast for faster text-to-video and image-to-video at up to 1080p. All three output 4 to 12 second videos.
First-and-last-frame video only. Send at least one of first_frame_url or last_frame_url; a prompt-only text-to-video call is rejected. Output is 480p, 720p, or 1080p at 4 to 12 seconds, with an optional prompt describing the transition. For plain text-to-video or image-to-video, call bytedance/seedance-1.0 or bytedance/seedance-1.0-pro-fast.
Pass first_frame_url, last_frame_url, or both, and the model generates the in-between motion. An optional prompt of up to 5,000 characters describes the transition. Useful for hero shots where you know the open and close beats but want the middle handled. Input images can be png, webp, jpeg, or gif up to 30 MB.
Six: 1:1, 3:4, 4:3, 9:16, 16:9, and 21:9. The aspect_ratio field applies to text-to-video calls and to Pro's first-and-last-frame calls, with 16:9 as the default. Image-to-video has no aspect_ratio field; the input frame sets the shape.
Any whole number from 4 to 12 seconds, on all three models. The API converts your duration to frames at 24 per second. For longer output, SeeDance 2.5 handles 4 to 30 second videos.
No. Every SeeDance 1.0 model is video-only. SeeDance 1.5 Pro and SeeDance 2.0 generate synchronized audio in the same call.
There is no free tier. New UnificAlly accounts get a $0.50 signup balance: enough for a short 480p test on the base model (from $0.015 per second), while bytedance/seedance-1.0-pro-fast at $0.007 per second stretches it to about 71 seconds of output across a few runs. After that, billing is pay-per-use with no subscription, and your balance never expires.
ByteDance writes it Seedance (one word); UnificAlly styles it SeeDance. Searches for "seeddance 1.0", "seed dance 1.0", or "seadance 1.0" all refer to the same ByteDance AI video generation model family documented on this page.