Seedance Series - QWave API

POST

tasks

Seedance Series

curl --request POST \
  --url https://www.qingbo.dev/v1/tasks \
  --header 'Content-Type: application/json' \
  --data '
{
  "model": "<string>",
  "action": "<string>",
  "prompt": "<string>",
  "aspect_ratio": "<string>",
  "resolution": "<string>",
  "duration": 123,
  "image_urls": [
    "<string>"
  ],
  "first_frame_image": "<string>",
  "last_frame_image": "<string>",
  "video_urls": [
    "<string>"
  ],
  "audio_urls": [
    "<string>"
  ],
  "callback_url": "<string>",
  "callback_events": [
    "<string>"
  ]
}
'

{
  "task_id": "task-wave1775285160b950328499",
  "model": "doubao-seedance-2.0",
  "action": "reference",
  "status": "queued",
  "created_at": 1775285160040,
  "progress": 0
}

ByteDance Doubao Seedance video generation series, covering the full range from flagship fast (Pro Fast) → flagship high-quality (Pro Quality) → audio-enabled 1.5 → multimodal 2.0.

Generation	Positioning
1.0 Pro Fast	Flagship fast tier, balanced quality and speed; about 3x faster than Pro Quality
1.0 Pro Quality	Flagship high-quality tier, 1080P multi-shot storytelling, suited for final delivery
1.5 Pro	New-generation audio-enabled video, joint audio-video generation, multilingual dialogue + lip sync
2.0	Multimodal flagship, unified architecture for text + images (≤9) + videos (≤3) + audio (≤3); `@image1`/`@video2`/`@audio3` reference syntax
2.0 Face	2.0 with enhanced face/identity reference, suited for digital-human shorts and lip-sync content
2.0 Fast / Fast Face	2.0 fast tier, no 1080P, trades quality for faster output and lower cost

Pricing

Billed by resolution × duration, in $/sec.

Model	480P	720P	1080P
`doubao-seedance-2.0-fast-face`	`$0.085`	`$0.18275`	—
`doubao-seedance-2.0-fast`	`$0.06205`	`$0.13345`	—
`doubao-seedance-2.0-face`	`$0.1054`	`$0.22695`	`$0.53125`
`doubao-seedance-2.0`	`$0.077095`	`$0.16592`	`$0.374`
`doubao-seedance-1.5-pro`	`$0.021675`	`$0.04675`	`$0.11475`
`doubao-seedance-1.0-pro-quality`	`$0.021675`	`$0.04675`	`$0.1105`
`doubao-seedance-1.0-pro-fast`	`$0.00935`	`$0.02125`	`$0.0442`

2.0 series video-reference subprice (when video_urls is present and triggers reference_video / reference, the table below replaces the base price above, in $/sec):

Model	480P	720P	1080P
`doubao-seedance-2.0-fast-face`	`$0.051`	`$0.10965`	—
`doubao-seedance-2.0-fast`	`$0.036975`	`$0.0799`	—
`doubao-seedance-2.0-face`	`$0.06375`	`$0.13685`	`$0.31875`
`doubao-seedance-2.0`	`$0.04675`	`$0.1003`	`$0.22695`

Video reference is actually cheaper — generations with video_urls are easier because they have a motion-rhythm reference, so the subprice is significantly lower than the pure-generation base price (e.g. 2.0 720P base $0.16592 vs video reference $0.1003, about 60% of the base). Do not interpret it as a markup.

Mode Routing

The whole series shares a field-routing convention — the backend automatically determines the action from the media fields you pass in, so you usually don’t need to set action explicitly.

Fields passed	Routed mode	Notes
`prompt` only	generate (T2V)	Text-to-video; aspect controlled by `aspect_ratio`
`+ first_frame_image`	image2video (I2V)	First-frame driven; follows the first frame’s aspect
`+ first_frame_image + last_frame_image`	first_last_frame	Interpolation constrained by first + last frame
`+ image_urls`	reference	Multi-image character / style consistency
`+ video_urls` (2.0)	reference_video	Video clip reference (2.0 only)
`+ audio_urls` (2.0)	reference_audio	Audio-driven (2.0 only)

2.0 multi-asset reference syntax — inside prompt you can use placeholders like @image1 / @video2 / @audio3 to reference image_urls[0] / video_urls[1] / audio_urls[2]; indices are 1-based.

Request Examples

curl -X POST https://www.qingbo.dev/v1/tasks \
  -H "Authorization: Bearer $WAVE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "doubao-seedance-1.0-pro-fast",
    "prompt": "A Shiba Inu in a spacesuit walking on the moon, cinematic lighting",
    "duration": 5,
    "resolution": "720p",
    "aspect_ratio": "16:9"
  }'

{
  "task_id": "task-wave1775285160b950328499",
  "model": "doubao-seedance-2.0",
  "action": "reference",
  "status": "queued",
  "created_at": 1775285160040,
  "progress": 0
}

After submission, poll status with GET /v1/tasks/{task_id}. See Task System for details.

Available Models

Model ID	Series	Supported actions	Resolution	Duration (sec)
`doubao-seedance-2.0-fast-face`	2.0	Same as 2.0	480p / 720p	4-15
`doubao-seedance-2.0-fast`	2.0	Same as 2.0	480p / 720p	4-15
`doubao-seedance-2.0-face`	2.0	Same as 2.0	480p / 720p / 1080p	4-15
`doubao-seedance-2.0`	2.0	generate / image2video / first_last_frame / reference / reference_video / reference_audio	480p / 720p / 1080p	4-15
`doubao-seedance-1.5-pro`	1.5	generate / image2video / first_last_frame	480p / 720p / 1080p	4-12
`doubao-seedance-1.0-pro-quality`	1.0 Pro	generate / image2video / first_last_frame / reference	480p / 720p / 1080p	2-12
`doubao-seedance-1.0-pro-fast`	1.0 Pro	generate / image2video / reference	480p / 720p / 1080p	2-12

Common Parameters

model

string

required

Model ID; see the “Available Models” table above

action

string

default:"generate"

Operation type; usually no need to pass explicitly — the backend routes automatically based on media fields. Valid values:

generate — text-to-video (T2V)
image2video — image-to-video (I2V); requires first_frame_image
first_last_frame — first/last-frame constraint; requires first_frame_image + last_frame_image
reference — reference-image driven; requires image_urls
reference_video — video clip reference (2.0 series only); requires video_urls
reference_audio — audio-driven (2.0 series only); requires audio_urls

prompt

string

Video description text. Required for T2V; optional as guidance for other modes. The 2.0 series supports @image1 / @video2 / @audio3 reference syntax

aspect_ratio

string

default:"16:9"

Frame aspect ratio. Valid values:

16:9 — landscape widescreen
9:16 — portrait
1:1 — square
4:3 — landscape
3:4 — portrait
21:9 — ultrawide
adaptive — adaptive (2.0 series only, follows the reference asset)

resolution

string

default:"720p"

Output resolution; valid values vary by model:

480p
720p
1080p (not supported by Lite-i2v / 2.0-fast / 2.0-fast-face)

duration

integer

default:"5"

Video duration in seconds. Range varies by series:

1.0 series: 2-12
1.5 Pro: 4-12
2.0 series: 4-15

image_urls

string[]

Array of reference-image URLs; triggers reference mode. 1.0 / 1.5 support 1-9 images; 2.0 supports up to 9

first_frame_image

string

First-frame image URL; triggers image2video or first_last_frame mode

last_frame_image

string

Last-frame image URL; combined with first_frame_image triggers first_last_frame mode

video_urls

string[]

Array of reference-video URLs (2.0 series only), up to 3 clips; triggers reference_video mode. Billing uses the “video-reference subprice” table above in place of the base price (significantly lower than pure generation)

audio_urls

string[]

Array of reference-audio URLs (2.0 series only), up to 3 tracks; triggers reference_audio mode

callback_url

string

Webhook callback URL, invoked when the task reaches a terminal state. See Callback Mechanism

callback_events

string[]

Subscribed callback event types; defaults to terminal states only (completed / failed)

Model-Specific Parameters

2.0 Fast / 2.0 Fast Face
2.0 / 2.0 Face
1.5 Pro
1.0 Pro Quality
1.0 Pro Fast

Parameters identical to 2.0 / 2.0 Face (generate_audio / tools / image_with_roles / return_last_frame).Differences:

Only 480P / 720P, no 1080P
Fast tier; slight quality trade-off for faster output and lower cost
Fast Face variant retains face-consistency capability
Duration 4-15 seconds

Typical use cases:

Bulk production of digital-human short videos
Social-media lip-sync content
Rapid asset iteration

generate_audio

boolean

default:"false"

Whether to generate audio (audio-enabled video). The 2.0 series uses the field name generate_audio

tools

array

Tool list; e.g. [{"type": "web_search"}] enables web-search augmentation

image_with_roles

array

Reference images with role tags; takes priority over image_urls + first_frame_image / last_frame_image. Element shape:

url — image URL
role — first_frame / last_frame

return_last_frame

boolean

default:"false"

Whether to additionally return the last-frame image, useful for chained sequel generation

Features (2.0):

Multimodal unified architecture: any combination of text + images (≤9) + videos (≤3) + audio (≤3)
@image1 / @video2 / @audio3 prompt reference syntax
aspect_ratio supports adaptive
Duration 4-15 seconds

2.0 Face additions:

Enhanced face / identity consistency; stable features and skin tone across multi-shot
Suited for digital-human shorts / lip-sync

audio

boolean

default:"false"

Whether to generate audio (audio-enabled video). The 1.5 series uses the field name audio — different from 2.0’s generate_audio

camerafixed

boolean

default:"false"

Whether to lock the camera

Features:

Joint audio-video generation, lip sync
Multilingual dialogue: Chinese / English / Japanese / Korean / Spanish
Dialects: Sichuanese / Cantonese
Duration 4-12 seconds

No model-specific parameters; uses common parameters only.Features:

Flagship fast tier, about 3x faster than Pro Quality
Full resolution: 480P / 720P / 1080P
Duration 2-12 seconds
Does not support first_last_frame

Resource Limits

Item	Limit
Reference images (`image_urls`)	Up to 9, ≤ 30MB each, JPG / PNG / WEBP
First/last frame (`first_frame_image` / `last_frame_image`)	≤ 30MB each, JPG / PNG / WEBP
Reference videos (`video_urls`, 2.0 only)	Up to 3 clips, MP4 / MOV, ≤ 100MB per clip
Reference audio (`audio_urls`, 2.0 only)	Up to 3 tracks, WAV / MP3, ≤ 15MB per track
Output	MP4, link valid for 24 hours

Task System Reference — task state machine / polling cadence / async push
Request and Response Format — common error codes / headers / rate limits
Authentication — API key application and usage

Kling SeriesKuaishou Kling video generation — v2.6 / v3 / v3-omni / video-o1, four generations covering text-to-video, image-to-video, first/last frame, multimodal reference, and reasoning-enhanced

Seedance Series

curl --request POST \
  --url https://www.qingbo.dev/v1/tasks \
  --header 'Content-Type: application/json' \
  --data '
{
  "model": "<string>",
  "action": "<string>",
  "prompt": "<string>",
  "aspect_ratio": "<string>",
  "resolution": "<string>",
  "duration": 123,
  "image_urls": [
    "<string>"
  ],
  "first_frame_image": "<string>",
  "last_frame_image": "<string>",
  "video_urls": [
    "<string>"
  ],
  "audio_urls": [
    "<string>"
  ],
  "callback_url": "<string>",
  "callback_events": [
    "<string>"
  ]
}
'

{
  "task_id": "task-wave1775285160b950328499",
  "model": "doubao-seedance-2.0",
  "action": "reference",
  "status": "queued",
  "created_at": 1775285160040,
  "progress": 0
}

Documentation Index

​Pricing

​Mode Routing

​Request Examples

​Available Models

​Common Parameters

​Model-Specific Parameters

​Resource Limits

​Related Docs

Pricing

Mode Routing

Request Examples

Available Models

Common Parameters

Model-Specific Parameters

Resource Limits

Related Docs