Inference Logoinference shell
apps/bytedance/seedance-2-5-studio

seedance-2-5-studio

Professional multimodal video generation with private asset library support. Automatically uploads reference images to BytePlus virtual portrait library for enhanced character consistency. Supports up to 4K (10-bit color), durations up to 30s, MOV output, and multimodal reference-to-video with synchronized audio.

run in browser →run via API
run with your agent
# install belt
$curl -fsSL https://cli.inference.sh | sh
# view schema & details
$belt app get bytedance/seedance-2-5-studio
# run
$belt app run bytedance/seedance-2-5-studio

api reference

about

professional multimodal video generation with private asset library support. automatically uploads reference images to byteplus virtual portrait library for enhanced character consistency. supports up to 4k (10-bit color), durations up to 30s, mov output, and multimodal reference-to-video with synchronized audio.

1. calling the api

install the client

the client provides a convenient way to interact with the api.

bash
1pip install inferencesh

setup your api key

set INFERENCE_API_KEY as an environment variable. get your key from settings → api keys.

bash
1export INFERENCE_API_KEY="inf_your_key"

run and get result

submit a request and wait for the final result. best for batch processing or when you don't need progress updates.

python
1from inferencesh import inference23client = inference()456result = client.run({7        "app": "bytedance/seedance-2-5-studio",8        "input": {}9    })1011print(result["output"])

stream live updates

get real-time progress updates as the task runs. ideal for showing progress bars, partial results, or long-running tasks.

python
1from inferencesh import inference23client = inference()456# stream=True yields updates as they arrive7for update in client.run({8        "app": "bytedance/seedance-2-5-studio",9        "input": {}10    }, stream=True):11    if update.get("progress"):12        print(f"progress: {update['progress']}%")13    if update.get("output"):14        print(f"output: {update['output']}")

2. authentication

the api uses api keys for authentication. see the authentication docs for detailed setup instructions.

3. files

file inputs are automatically handled by the sdk. you can pass local paths, urls, or base64 data.

automatic upload

the python sdk automatically detects local file paths and uploads them. urls are passed through as-is.

python
1# local file paths are automatically uploaded2result = client.run({3    "app": "bytedance/seedance-2-5-studio",4    "input": {5        "image": "/path/to/local/image.png",  # detected & uploaded6        "audio": "https://example.com/audio.mp3",  # url passed through7    }8})

manual upload

you can also upload files manually and use the returned url.

python
1# upload and get a hosted URL2file = client.files.upload("/path/to/file.png")3print(file.uri)  # https://cloud.inference.sh/...

4. webhooks

get notified when a task completes by providing a webhook url. when the task reaches a terminal state (completed, failed, or cancelled), a POST request is sent to your url with the task result.

python
1result = client.run({2    "app": "bytedance/seedance-2-5-studio",3    "input": {},4    "webhook": "https://your-server.com/webhook"5}, wait=False)

webhook payload

your endpoint receives a JSON POST with the task result:

json
1{2  "id": "task_abc123",3  "status": 9,4  "output": { ... },5  "error": "",6  "session_id": null,7  "created_at": "2024-01-15T10:30:00Z",8  "updated_at": "2024-01-15T10:30:05Z"9}
idstring— task id
statusnumber— terminal status (9=completed, 10=failed, 11=cancelled)
outputobject— task output (when completed)
errorstring— error message (when failed)
session_idstring— session id (if using sessions)
created_atstring— iso timestamp
updated_atstring— iso timestamp

5. schema

input

promptstring

text prompt describing the video content (not needed with draft_task_id). use @image1, @image2, @video1, @video2, @audio1 to reference inputs in order (e.g. '@image1 is the style reference, @video1 provides the motion'). for video editing, use editing verbs ('edit @video1 to remove...'). supports english, spanish, indonesian, portuguese, japanese, malay, thai, arabic, vietnamese, and korean.

default: ""example: "@Image1 is the character reference. @Video1 provides the motion. A marble statue stands in a sunlit hall, camera slowly orbiting."
imagestring(file)

first-frame image for image-to-video generation. ratio must be 'adaptive'. mutually exclusive with reference inputs.

end_imagestring(file)

last-frame image for first+last frame video generation. requires image to be set as the first frame. ratio must be 'adaptive'.

reference_imagesarray

reference images for multimodal generation (up to 30). use prompt to describe how to use each, e.g. 'image 1', 'image 2'. mutually exclusive with image/end_image.

default: []
reference_videosarray

reference videos for multimodal generation (up to 10). each 2-30s, total max 30s. formats: mp4/mov (mov recommended for best quality). mutually exclusive with image/end_image.

default: []
reference_audiosarray

reference audios for multimodal generation (up to 10). each 2-30s, total max 30s. formats: wav/mp3. audio-only input is supported.

default: []
resolutionstring

video resolution. 1080p for high quality, 720p for balanced, 480p for fastest.

default: "720p"
options:"480p""720p""1080p"
ratiostring

aspect ratio. must be 'adaptive' for image-to-video, first+last frame, video editing, and video extension tasks. 'adaptive' auto-selects based on input content.

default: "adaptive"
options:"adaptive""21:9""16:9""4:3""1:1""3:4""9:16"
durationinteger

duration in seconds (4-30), or -1 for auto-select. must be -1 for video editing tasks.

default: -1
output_formatstring

output video format. mov provides higher color precision (yuv 4:4:4); recommended for professional workflows and when using video references.

default: "mp4"
options:"mp4""mov"
generate_audioboolean

whether to generate synchronized audio with the video.

default: true
seedinteger

seed for reproducibility (-1 for random).

default: -1
watermarkboolean

whether to add watermark to the output video.

default: false
task_typestring

guide the generation task type. 'auto' lets the api detect from prompt content. set explicitly to avoid misdetection errors — e.g. when the api wrongly classifies a reference task as video editing and rejects duration/ratio settings.

default: "auto"
options:"auto""video_editing""video_extension""video_reference"
safety_filterboolean

enable input safety filtering. set to false to disable nsfw content filtering on inputs.

default: true
draftboolean

draft mode: render a fast, low-cost 480p preview instead of the final video. the output's draft_task_id renders the chosen take at 1080p later. iterate on drafts, then render only the one you want.

default: false
draft_task_idstring

render the final 1080p video from a draft (the draft_task_id of an earlier draft run, valid for 7 days). prompt, images, references, duration, ratio, seed and audio are reused from the draft; those inputs are ignored here. resolution is always 1080p.

safety_identifierstring

unique identifier of end user for platform safety policy. must be fixed and unique per user, max 64 chars. recommended: hash of username, user id, or email. also used to namespace asset groups.

output

draft_task_idstring

set on a draft run: pass it as draft_task_id to render this take as the final 1080p video.

videostring(file)*

the generated video file.

ready to run seedance-2-5-studio?

try in browserbrowse all tools

we use cookies

we use cookies to ensure you get the best experience on our website. for more information on how we use cookies, please see our cookie policy.

by clicking "accept", you agree to our use of cookies.
learn more.