alibaba-wan（阿里巴巴湾）

Wan 2.6 & 2.5 — AI Video & Image Generation by Alibaba

Generate AI videos and images using Alibaba's Wan 2.6 and Wan 2.5 — featuring text-to-video, image-to-video, video-to-video, text-to-image, and image editing with up to 1080p resolution, 15-second duration, multi/single camera modes, audio-guided generation, and built-in prompt expansion.

Wan 2.6 is the latest flagship model with cinematic motion quality, multi-camera shot types, audio URL input for guided generation, and video-to-video transfer with character-level prompt control. Wan 2.5 offers a cost-effective alternative with 480p support and Flash variants for rapid prototyping.

Data usage note: This skill sends text prompts, image URLs, audio URLs, and video files to the Atlas Cloud API (api.atlascloud.ai) for generation. No data is stored locally beyond the downloaded output files. API usage incurs charges based on resolution, duration, and model selected.

Key Capabilities

• Text-to-Video — Generate video clips from text descriptions with optional audio (2.6 / 2.5)
• Image-to-Video — Animate still images into dynamic video (2.6 / 2.5)
• Video-to-Video — Transfer style or replace characters in existing videos using characterX prompt notation (2.6)
• Text-to-Image — Generate images from text descriptions, 27 preset sizes (2.6)
• Image Editing — Edit images with prompt-based instructions, up to 4 reference images (2.6)
• Audio-Guided Generation — Provide an audio URL to guide video generation with synchronized sound (2.6)
• Multi/Single Camera — multi_camera for dynamic shots, single_camera for stable framing (2.6)
• Prompt Expansion — Built-in prompt optimization for better results
• Up to 1080p — Resolutions: 480p, 720p, 1080p
• Up to 15s — Duration: 5/10/15 seconds (2.6), 5/10 seconds (2.5)
• Flash Variants — Fast, budget-friendly generation for drafts (2.6 I2V Flash, 2.5 Fast)

Setup

1. Sign up at https://www.atlascloud.ai
2. Console → API Keys → Create new key
3. Set env: export ATLASCLOUD_API_KEY="your-key"

The API key is tied to your Atlas Cloud account and its pay-as-you-go balance. All usage is billed to this account. Atlas Cloud does not currently support scoped keys — the key grants access to all models available on your account.

Script Usage

This skill includes Python scripts for both video and image generation. Zero external dependencies required.

List available models

python scripts/generate_video.py list-models
python scripts/generate_image.py list-models

Generate a video

python scripts/generate_video.py generate \
  --model "alibaba/wan-2.6/text-to-video" \
  --prompt "Your prompt here" \
  --output ./output \
  duration=5

Generate an image

python scripts/generate_image.py generate \
  --model "alibaba/wan-2.6/text-to-image" \
  --prompt "Your prompt here" \
  --output ./output

Image-to-video

python scripts/generate_video.py generate \
  --model "alibaba/wan-2.6/image-to-video" \
  --image "https://example.com/photo.jpg" \
  --prompt "Animate this scene" \
  --output ./output \
  resolution=1080p duration=5

Upload a local file

python scripts/generate_video.py upload ./local-file.jpg

Run python scripts/generate_video.py generate --help or python scripts/generate_image.py generate --help for all options. Extra model params can be passed as key=value.

Pricing

Wan 2.6 — Video Models (per second, by resolution)

All video prices are per second of video generated. Atlas Cloud pricing varies by resolution.

Text-to-Video & Video-to-Video

Image-to-Video

Image-to-Video Flash

Wan 2.6 — Image Models (per image)

Wan 2.5 — Video Models (per second, flat rate)

fal.ai pricing sourced from fal.ai/models/wan.

Available Models

Wan 2.6 Video

Wan 2.6 Image

Wan 2.5 Video

Parameters

Wan 2.6 — Text-to-Video

Wan 2.6 — Image-to-Video

Same as text-to-video (without size), plus:

Wan 2.6 — Video-to-Video

Wan 2.6 — Text-to-Image

Wan 2.6 — Image Edit

Same as text-to-image, plus:

Wan 2.5 — Text-to-Video

Wan 2.5 — Image-to-Video

Same as Wan 2.5 text-to-video (without size), plus:

Video Size Options (Wan 2.6)

T2V / V2V (10 presets): 1280*720, 720*1280, 960*960, 1920*1080, 1080*1920, 1280*960, 960*1280, 1920*816, 816*1920, 1280*544

Image Size Options (Wan 2.6 — 27 presets): 1024*1024, 1280*720, 720*1280, 1280*960, 960*1280, 1536*1024, 1024*1536, 1280*1280, 1536*1536, 2048*1024, 1024*2048, 1536*1280, 1280*1536, 1680*720, 720*1680, 2016*864, 864*2016, 1536*864, 864*1536, 2184*936, 936*2184, 1400*1050, 1050*1400, 1680*1050, 1050*1680, 1176*1176, 1560*1560

Workflow: Submit → Poll → Download

Text-to-Video Example (Wan 2.6)

# Step 1: Submit
curl -s -X POST "https://api.atlascloud.ai/api/v1/model/generateVideo" \
  -H "Authorization: Bearer $ATLASCLOUD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "alibaba/wan-2.6/text-to-video",
    "prompt": "A drone shot flying over ancient ruins at golden hour, camera slowly descending toward a central courtyard",
    "size": "1920*1080",
    "duration": 10,
    "shot_type": "multi_camera",
    "generate_audio": true,
    "enable_prompt_expansion": true
  }'
# Returns: { "code": 200, "data": { "id": "prediction-id" } }

# Step 2: Poll (every 5 seconds until completed)
curl -s "https://api.atlascloud.ai/api/v1/model/prediction/{prediction-id}" \
  -H "Authorization: Bearer $ATLASCLOUD_API_KEY"
# Returns: { "code": 200, "data": { "status": "completed", "outputs": ["https://...video-url..."] } }

# Step 3: Download
curl -o output.mp4 "VIDEO_URL_FROM_OUTPUTS"

Image-to-Video Example (Wan 2.6)

curl -s -X POST "https://api.atlascloud.ai/api/v1/model/generateVideo" \
  -H "Authorization: Bearer $ATLASCLOUD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "alibaba/wan-2.6/image-to-video",
    "image": "https://example.com/landscape.jpg",
    "prompt": "The camera slowly zooms in as clouds drift across the sky and leaves rustle in the wind",
    "resolution": "1080p",
    "duration": 5,
    "generate_audio": true
  }'

Video-to-Video Example (Wan 2.6)

curl -s -X POST "https://api.atlascloud.ai/api/v1/model/generateVideo" \
  -H "Authorization: Bearer $ATLASCLOUD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "alibaba/wan-2.6/video-to-video",
    "videos": ["https://example.com/original-video.mp4"],
    "prompt": "Transform character1 into a cartoon anime character, keep the background unchanged",
    "size": "1280*720",
    "duration": 5,
    "shot_type": "single_camera"
  }'

Audio-Guided Generation Example (Wan 2.6)

curl -s -X POST "https://api.atlascloud.ai/api/v1/model/generateVideo" \
  -H "Authorization: Bearer $ATLASCLOUD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "alibaba/wan-2.6/text-to-video",
    "prompt": "A jazz band performing on stage, musicians playing saxophone and piano",
    "audio": "https://example.com/jazz-music.mp3",
    "size": "1920*1080",
    "duration": 10
  }'

Text-to-Image Example (Wan 2.6)

curl -s -X POST "https://api.atlascloud.ai/api/v1/model/generateImage" \
  -H "Authorization: Bearer $ATLASCLOUD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "alibaba/wan-2.6/text-to-image",
    "prompt": "A cyberpunk cityscape at night, neon signs reflected in rain puddles, photorealistic",
    "size": "1680*720",
    "enable_prompt_expansion": true
  }'
# Returns: { "code": 200, "data": { "id": "prediction-id" } }

Image Editing Example (Wan 2.6)

curl -s -X POST "https://api.atlascloud.ai/api/v1/model/generateImage" \
  -H "Authorization: Bearer $ATLASCLOUD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "alibaba/wan-2.6/image-edit",
    "prompt": "Change the background to a sunset beach scene, keep the person unchanged",
    "images": ["https://example.com/photo.jpg"],
    "size": "1280*720"
  }'

Image-to-Video Flash Example (Wan 2.6)

curl -s -X POST "https://api.atlascloud.ai/api/v1/model/generateVideo" \
  -H "Authorization: Bearer $ATLASCLOUD_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "alibaba/wan-2.6/image-to-video-flash",
    "image": "https://example.com/portrait.jpg",
    "prompt": "The person slowly turns and smiles",
    "resolution": "720p",
    "duration": 5
  }'

Polling Logic

• processing / starting / running → wait 5s, retry (typically takes ~30-120s for video, ~5-10s for image)
• completed / succeeded → done, get URL from data.outputs[]
• failed → error, read data.error

Atlas Cloud MCP Tools (if available)

If the Atlas Cloud MCP server is configured, use built-in tools:

atlas_generate_video(model="alibaba/wan-2.6/text-to-video", params={...})
atlas_generate_image(model="alibaba/wan-2.6/text-to-image", params={...})
atlas_get_prediction(prediction_id="...")

Implementation Guide

1. Determine task type:

- Text-to-video: user describes a scene/action in text → 2.6 T2V or 2.5 T2V - Image-to-video: user provides an image to animate → 2.6 I2V or 2.6 I2V Flash (budget) - Video-to-video: user wants to transform an existing video → 2.6 V2V - Text-to-image: user wants to generate an image → 2.6 T2I - Image editing: user wants to modify existing images → 2.6 Image Edit

1. Choose model version:

- Wan 2.6 (recommended): Latest generation, best quality, multi-camera, audio-guided, V2V - Wan 2.6 Flash: Budget I2V at $0.018/s — ideal for drafts - Wan 2.5: Cost-effective at $0.035/s flat rate, supports 480p

1. Extract parameters:

- Prompt: describe scene, action, camera movement - Negative prompt: exclude undesired elements ("blurry, distorted, watermark") - Resolution: 480p for drafts, 720p default, 1080p for final output - Duration: 5s default, up to 15s for 2.6, up to 10s for 2.5 - Shot type (2.6 only): multi_camera for dynamic shots, single_camera for stable framing - Audio: provide URL for audio-guided generation, or set generate_audio: true - Prompt expansion: set enable_prompt_expansion: true for auto-optimized prompts

1. Execute: POST to generateVideo/generateImage API → poll result → download

1. Present result: show file path, offer to play/open

Prompt Tips

• Scene + Action: "A samurai draws his sword in a bamboo forest at dawn, mist rising from the ground"
• Camera direction: "Camera slowly dollies forward...", "Aerial tracking shot of...", "First-person POV walking through..."
• Multi-camera: Use shot_type: "multi_camera" with prompts like "Cut between close-up and wide shot..."
• V2V character control: Reference characters as character1, character2 — e.g., "Transform character1 into an anime character"
• Audio-guided: Provide an audio URL to sync video with music, dialogue, or sound effects
• Negative prompts: "blurry, low quality, distorted faces, watermark, text overlay"

Model Comparison