Wan2.7-R2V - Generate Video (AI Model · By )

API Documentation

API Reference

Videos

Generate Video

Supported Model Code

Loading models...

Wan2.7-R2V (wan2.7-r2v)

wan2.7-r2v0 models support this endpoint

Wan2.7 reference-to-video, enhanced consistency and performance. Delivers superior stability for characters, props, and scenes. Supports hybrid referencing with up to five mixed image/video inputs and audio timbre cloning. Combined with core engine upgrades, it achieves unprecedented cinematic expressive power. https://bailian.console.alibabacloud.com/cn-beijing?tab=model#/model-market/detail/wan2.7-r2v?serviceSite=asia-pacific-china

Generate video

https://api.modelstream.ai

POST/v1/video/generations

Authentication

BearerAuth

AuthenticationBearer <token>

All API requests must be authenticated using a Bearer token in the Authorization header. Please ensure your API key is active.Authorization: Bearer sk-xxxxxx

Parameter Location: Header Param

Request Body

application/json

These parameters come from the selected model form_schema. Switching models updates this list and the request example.

reference_image*string

Required. Upload a reference image of the character (JPEG/PNG/BMP/WEBP, 240-8000px, ≤20MB). Refer to it as 'Image 1' in the prompt.

RequiredExample Value: https://static.modelstream.ai/demo/wan/wan2.7-r2v_reference.png

reference_video?string

Optional. Upload a video to guide motion and expressions (mp4/mov, 1-30s, ≤100MB). Reference images + videos must total ≤5.

reference_voice?string

Optional. Upload audio to clone the voice timbre (wav/mp3, 1-10s, ≤15MB). Only the timbre is referenced, not the spoken content.

first_frame?string

Optional. Specify the starting frame; the video is generated from it. When provided, the aspect ratio follows the first frame and the 'Aspect Ratio' setting is ignored.

prompt*string

Required. Describe the scene and content, max 5000 chars. The reference character is blended into the scene; refer to inputs as 'Image 1' / 'Video 1'.

RequiredExample Value: Reference Identity is walking through a dense, neon-lit cyberpunk market. The camera follows in a low-angle tracking shot. As they walk, they reach out to touch a holographic fish swimming through the air; their fingers ripple through the light, causing a realistic prismatic distortion. Their clothing—a matte black techwear jacket—glistens with realistic water beads and reflections of the neon signs. Atmosphere is moody and cinematic, 1080p, 30fps, 5 seconds duration.

negative_prompt?string

Optional. Describe content to exclude, max 500 chars.

Example Value: low resolution, blurry, distorted, deformed, watermark, out of focus

resolution?string

Output resolution tier, 720P or 1080P; directly affects cost. Default 1080P.

Example Value: 1080P

Enum/Options:

1080P720P

ratio?string

Output video aspect ratio: 16:9 / 9:16 / 1:1 / 4:3 / 3:4. Ignored when a first frame is provided. Default 16:9.

Example Value: 16:9

Enum/Options:

16:99:161:14:33:4

duration?number

Video duration in seconds. 2-15s without a reference video, 2-10s with one. Default 5s; directly affects cost.

Example Value: 5Value Range: 2 ≤ value ≤ 15step: 1

prompt_extend?boolean

When on, an LLM intelligently rewrites your prompt. Enabled by default.

Example Value: true

watermark?boolean

Add a fixed 'AI生成' watermark to the bottom-right of the video. Disabled by default.

Example Value: false

seed?number

Random seed, range [0, 2147483647]. System generates a random seed when unspecified.

Response Parameters

application/json

200apiDocs.responses.successCreateVideoGenTask

task_id?string

Parameter description for Task Id

status?string

Parameter description for Status

400apiDocs.responses.badRequestParams

error?object

Parameter description for Error

message?string

Error Message

type?string

Error Type

param?string

Related Parameters

code?string

Error Code

curl -X POST "https://api.modelstream.ai/v1/video/generations" \
  -H "Authorization: Bearer <token>" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "wan2.7-r2v",
  "prompt": "Reference Identity is walking through a dense, neon-lit cyberpunk market. The camera follows in a low-angle tracking shot. As they walk, they reach out to touch a holographic fish swimming through the air; their fingers ripple through the light, causing a realistic prismatic distortion. Their clothing—a matte black techwear jacket—glistens with realistic water beads and reflections of the neon signs. Atmosphere is moody and cinematic, 1080p, 30fps, 5 seconds duration.",
  "size": "1080P",
  "duration": 5,
  "metadata": {
    "input": {
      "media": [
        {
          "type": "reference_image",
          "url": "https://static.modelstream.ai/demo/wan/wan2.7-r2v_reference.png"
        }
      ],
      "negative_prompt": "low resolution, blurry, distorted, deformed, watermark, out of focus"
    },
    "parameters": {
      "ratio": "16:9",
      "prompt_extend": true,
      "watermark": false
    }
  }
}'

{
  "task_id": "abcd1234efgh",
  "status": "queued"
}

API Documentation

API Reference

Videos

Generate Video

Supported Model Code

Loading models...

Wan2.7-R2V (wan2.7-r2v)

wan2.7-r2v0 models support this endpoint

Generate video

https://api.modelstream.ai

POST/v1/video/generations

Authentication

BearerAuth

AuthenticationBearer <token>

All API requests must be authenticated using a Bearer token in the Authorization header. Please ensure your API key is active.Authorization: Bearer sk-xxxxxx

Parameter Location: Header Param

Request Body

application/json

These parameters come from the selected model form_schema. Switching models updates this list and the request example.

reference_image*string

Required. Upload a reference image of the character (JPEG/PNG/BMP/WEBP, 240-8000px, ≤20MB). Refer to it as 'Image 1' in the prompt.

RequiredExample Value: https://static.modelstream.ai/demo/wan/wan2.7-r2v_reference.png

reference_video?string

Optional. Upload a video to guide motion and expressions (mp4/mov, 1-30s, ≤100MB). Reference images + videos must total ≤5.

reference_voice?string

Optional. Upload audio to clone the voice timbre (wav/mp3, 1-10s, ≤15MB). Only the timbre is referenced, not the spoken content.

first_frame?string

Optional. Specify the starting frame; the video is generated from it. When provided, the aspect ratio follows the first frame and the 'Aspect Ratio' setting is ignored.

prompt*string

Required. Describe the scene and content, max 5000 chars. The reference character is blended into the scene; refer to inputs as 'Image 1' / 'Video 1'.

negative_prompt?string

Optional. Describe content to exclude, max 500 chars.

Example Value: low resolution, blurry, distorted, deformed, watermark, out of focus

resolution?string

Output resolution tier, 720P or 1080P; directly affects cost. Default 1080P.

Example Value: 1080P

Enum/Options:

1080P720P

ratio?string

Output video aspect ratio: 16:9 / 9:16 / 1:1 / 4:3 / 3:4. Ignored when a first frame is provided. Default 16:9.

Example Value: 16:9

Enum/Options:

16:99:161:14:33:4

duration?number

Video duration in seconds. 2-15s without a reference video, 2-10s with one. Default 5s; directly affects cost.

Example Value: 5Value Range: 2 ≤ value ≤ 15step: 1

prompt_extend?boolean

When on, an LLM intelligently rewrites your prompt. Enabled by default.

Example Value: true

watermark?boolean

Add a fixed 'AI生成' watermark to the bottom-right of the video. Disabled by default.

Example Value: false

seed?number

Random seed, range [0, 2147483647]. System generates a random seed when unspecified.

Response Parameters

application/json

200apiDocs.responses.successCreateVideoGenTask

task_id?string

Parameter description for Task Id

status?string

Parameter description for Status

400apiDocs.responses.badRequestParams

error?object

Parameter description for Error

message?string

Error Message

type?string

Error Type

param?string

Related Parameters

code?string

Error Code

curl -X POST "https://api.modelstream.ai/v1/video/generations" \
  -H "Authorization: Bearer <token>" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "wan2.7-r2v",
  "prompt": "Reference Identity is walking through a dense, neon-lit cyberpunk market. The camera follows in a low-angle tracking shot. As they walk, they reach out to touch a holographic fish swimming through the air; their fingers ripple through the light, causing a realistic prismatic distortion. Their clothing—a matte black techwear jacket—glistens with realistic water beads and reflections of the neon signs. Atmosphere is moody and cinematic, 1080p, 30fps, 5 seconds duration.",
  "size": "1080P",
  "duration": 5,
  "metadata": {
    "input": {
      "media": [
        {
          "type": "reference_image",
          "url": "https://static.modelstream.ai/demo/wan/wan2.7-r2v_reference.png"
        }
      ],
      "negative_prompt": "low resolution, blurry, distorted, deformed, watermark, out of focus"
    },
    "parameters": {
      "ratio": "16:9",
      "prompt_extend": true,
      "watermark": false
    }
  }
}'

{
  "task_id": "abcd1234efgh",
  "status": "queued"
}