Name: Wan 2.2 Vace Pose API
Brand: Alibaba Cloud

Overview

Model Overview

Wan 2.2 Vace Pose is an advanced video generation model developed by Wan Labs using the innovative VACE (Video Adaptive Compositional Editing) technology. It specializes in controllable video synthesis with fine-grained parameter support such as Canny edge detection, Depth maps, Pose estimation, MLSD (line segment detection), and more. This model empowers creators with enhanced control over video content generation, enabling high-quality, consistent, and fluid video sequences ideal for professional and creative applications.

Technical Specifications

Performance Benchmarks

Produces high-quality cinematic video sequences with smooth motion transitions and stable, flicker-free outputs.
Significantly improved temporal consistency and semantic accuracy over Wan 2.1 models.
Efficient memory use with optimized VAE compression enabling faster generation.
Real-time video editing and generation workflows supported on modern GPUs (e.g., RTX 4090).
Outperforms earlier versions in handling complex dynamic scenes involving multiple characters and camera movement.
Superior fine-grained control for precise creative direction (lighting, composition, motion).

Key Features

Advanced Video Control: Supports Canny, Depth, Pose, MLSD maps, and trajectory parameters for fine-grained video content manipulation.
Multi-Resolution Support: Generates videos efficiently at 512, 768, and 1024 pixel resolutions.
Temporal Frame Handling: Trained on sequences of 81 frames at 16 FPS for smooth, coherent video generation.
Multilingual Input Support: Enables video generation with prompts in multiple languages.
Generative Flexibility: Capable of subject-specific video generation using tailored control inputs.
Open and Extensible: Fully open-source model weights and workflows available for customization and integration.

API Pricing

360p: $0.065;
540p: $0.0975;
720p: $0.13

‍

Code Sample

Comparison with Other Models

vs Veo 3.0: Veo 3.0 focuses on fast text-to-video conversion with efficient rendering speeds but features limited direct control over spatial video elements. Wan 2.2 VACE Fun A14B Pose, while slower and more VRAM-intensive, allows detailed video editing and precise motion handling, making it better suited for high-quality production workflows.

vs KLING 2.0: Both models provide advanced video generation capabilities with open-source codebases. Wan 2.2 VACE excels in adaptive compositional editing with multi-condition inputs (pose, depth, canny edges), while KLING 2.0 matches in raw video quality but lacks the fine-grained editable controls present in VACE 2.0.

vs Wan 2.1 VACE: Wan 2.2 VACE Fun A14B Pose shows improved temporal consistency and visual fidelity over Wan 2.1 VACE, particularly in complex scenes involving multiple moving subjects. It reduces identity loss in face and pose representation more effectively and supports higher native resolutions (up to 1080p).

API Integration

Accessible via AI/ML API. Documentation: available here.

Use cases

Professional video production and animation with precise pose and motion control.
Cinematic storytelling and concept video creation.
Digital art animation with enhanced stability and visual coherence.
Real-time compositional editing for film, advertising, and multimedia projects.
Interactive video content generation with multi-modal input triggers.

‍

How to use Wan 2.2 Vace Pose API

Install any OpenAI-compatible SDK, point it at api.aimlapi.com/v1, and set the model to wan-22-vace-fun-a14b/pose.

import requests, time

headers = {"Authorization": "Bearer " + AIMLAPI_KEY}
job = requests.post(
    "https://api.aimlapi.com/v2/video/generations",
    headers=headers,
    json={
      "model": "wan-22-vace-fun-a14b/pose",
      "prompt": "A serene timelapse of clouds over a mountain range"
    },
).json()
gid = job["id"]

while True:
    res = requests.get(f"https://api.aimlapi.com/v2/video/generations?generation_id={gid}", headers=headers).json()
    if res.get("status") in ("completed", "error"):
        break
    time.sleep(5)
print(res["video"]["url"])

const headers = {
  Authorization: `Bearer ${process.env.AIMLAPI_KEY}`,
  "Content-Type": "application/json",
};
const job = await (await fetch("https://api.aimlapi.com/v2/video/generations", {
  method: "POST",
  headers,
  body: JSON.stringify({
    "model": "wan-22-vace-fun-a14b/pose",
    "prompt": "A serene timelapse of clouds over a mountain range"
  }),
})).json();

let res;
do {
  await new Promise((r) => setTimeout(r, 5000));
  res = await (await fetch(`https://api.aimlapi.com/v2/video/generations?generation_id=${job.id}`, { headers })).json();
} while (!["completed", "error"].includes(res.status));
console.log(res.video.url);

# submit the generation — the response contains the job "id"
curl -X POST https://api.aimlapi.com/v2/video/generations \
  -H "Authorization: Bearer $AIMLAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"wan-22-vace-fun-a14b/pose","prompt":"A serene timelapse of clouds over a mountain range"}'

# then poll until status is "completed" — video.url holds the result
curl "https://api.aimlapi.com/v2/video/generations?generation_id={id}" -H "Authorization: Bearer $AIMLAPI_KEY"

OpenAI-compatible — swap the base URL and it works with your existing SDK.