Wan 2.6 is now available on Atlas Cloud: A New Standard for Long-Form, Multi-Shot Video Generation
We are proud to introduce Wan 2.6, a powerful upgrade to our video generation capabilities, now available on Atlas Cloud. This release focuses on extending video duration, enhancing narrative control through multi-shot consistency, and providing flexible resolution options for professional creators.
Wan 2.6 Snapshot:
| Model | Specifications | Price |
|---|---|---|
| Wan 2.6 text-to-video | 720p Tier: 1280 × 720; 720 × 1280; 960 × 960; 1088 × 832; 832 × 1088. 1080p Tier: 1920 × 1080; 1080 × 1920; 1440 × 1440; 1632 × 1248; 1248 × 1632 | Start from 0.08/sec(\<0.08/sec(\<0.08/sec(\<0.1/sec on wavespeed.ai) |
| Wan 2.6 image-to-video | 720p; 1080p | |
| Wan 2.6 reference to video |
Introduction of Wan 2.6
Wan 2.6 introduces significant optimizations for production workflows. It moves beyond simple animated clips to support longer, narratively complex video generation. By integrating advanced temporal processing, it allows users to tell complete stories with consistent characters and scenes.
Key Features
15-Second Long Video Generation
Wan 2.6 significantly expands the temporal capacity of generated content. Users can now generate videos up to 15 seconds in length.
- Why it matters: Previous models were often limited to 5-10 seconds. The 15-second capability allows for more complete narrative arcs, richer spatial-temporal content, and sufficient runtime for social media formats like YouTube Shorts or TikTok without needing to stitch multiple clips together.
Multi-Shot Narrative Control
This feature is designed for storytellers. Wan 2.6 can generate sequences that mimic professional editing with multiple camera angles (shots) within a single generation task.
- How it works: The model maintains a high consistency of key information (characters, environment, lighting) across different shots.
- Smart Storyboarding: It supports simple prompts that the model intelligently breaks down into a storyboard, ensuring the visual flow makes sense cinematically.
Video Reference Generation (Video-to-Video)
Wan 2.6 introduces a robust "Video Reference" capability.
- Functionality: You can input a reference video to guide the appearance and tone of the new generation.
- Flexibility: It supports using a person or any object as the main subject.
- Interaction: The model supports complex scenarios, including single-subject focus or dual-subject interactions (e.g., co-starring/pairing), making it ideal for recreating viral trends or specific motion patterns using text prompts to modify the content while keeping the reference vibe.
Application of Wan 2.6 with Cases
With the extended 15-second duration and multi-shot consistency, Wan 2.6 is suitable for various professional applications:
- Social Media Content: Creators can produce ready-to-post 15-second clips for Reels or TikTok. The native audio synchronization and high frame rates (24fps) ensure the output feels premium and engaging.
Prompt:
A cinematic sci-fi trailer. Shot 1: Wide shot, a lonely explorer in a battered spacesuit walking across a desolate red Martian desert, a massive derelict spaceship in the distance. Shot 2: Close-up, the explorer stops and wipes dust off their helmet visor, eyes widening in shock. Shot 3: Over-the-shoulder shot, revealing a glowing, bioluminescent blue flower blooming rapidly in front of them. 8k resolution, highly detailed, consistent character.
- Commercial Advertising: The multi-shot feature allows marketers to show a product from different angles (close-up, wide shot) in a single consistent video generation, reducing the need for complex post-production editing.
A fluffy, adorable British Shorthair kitten with huge round eyes, cream color. TikTok viral video, bright commercial lighting, high saturation, energetic, wide-angle lens.
Shots:The Rhythmic Crash Zooms.The kitten sits on a soft rug looking innocent. The camera performs rapid, rhythmic crash zooms (snap zooms):Sudden Extreme Close-up on the kitten's pink nose (filling the screen).Snap back to Wide Shot. Snap in to close-up of one eye.Snap back out.This mimics a "beat-sync" effect. The kitten looks confused but cute.
- Music & Creative Visualization: Artists can use the Video Reference feature to transfer the "vibe" or motion of a dance or performance onto new animated characters, maintaining the beat and energy of the original reference.
Style: High-quality 2D pixel art, 16-bit retro game aesthetic, side-scrolling animation, neon noir, flat design, black background.
Subject: A silhouette of a little girl running continuously to the right side of the screen.
[00s-05s] The Spark: The screen is mostly pitch black. As the girl runs, her footsteps ignite neon geometric grids and pixelated streetlights that instantly fade in behind her. The area in front of her (right side) remains a total dark void.
[05s-10s] Cyberpunk City: The trail behind her expands into a glowing cyberpunk cityscape. Holographic billboards, skyscrapers, and data streams glitch into existence as she passes. Purple and cyan color palette. The contrast between the lit city behind and the darkness ahead is sharp.
[10s-15s] Deep Sea Dream: The city morphs seamlessly into a bioluminescent deep sea. The pixel buildings turn into glowing jellyfish and coral reefs. She leaves a trail of bubbles and light. Dreamy, ethereal atmosphere matching the lofi beat.
Negative prompt：
3D render, realistic, photograph, detailed face, sunlight, daytime, messy, blurry, noise, full background ahead, light in front, stopping, turning back, static image, complicated details, vector art, smooth lines.
Conclusion
Wan 2.6 expands the capabilities of video generation on Atlas Cloud by focusing on duration and narrative consistency. The update moves from short clips to 15-second sequences with multi-shot support, allowing for more detailed storytelling. With the simplified resolution structure for Text-to-Video and the introduction of Video Reference tools, the model provides practical solutions for both individual creators and enterprise workflows.
👇Experience Wan 2.6 on Atlas Cloud today.👇
FAQ
How does the Text-to-Video resolution selection differ in Wan 2.6 compared to previous versions?
In Wan 2.5 and earlier models, users selected specific pixel dimensions. In Wan 2.6, the process mimics the Image-to-Video workflow. Users select a general quality tier (720p, or 1080p), and the system automatically applies the appropriate resolution from the supported list (e.g., 1920×1080 or 1440×1440 for the 1080p tier).
What is the maximum video duration supported?
Wan 2.6 supports generating videos up to 15 seconds in length. This applies to both Text-to-Video and Image-to-Video tasks.
Can Wan 2.6 handle complex character interactions?
Yes. The Video Reference feature supports single-subject focus as well as dual-subject interactions (such as two people co-starring). The model uses the reference video and prompt to maintain the appearance and interaction logic of the subjects.
Does the "Multi-Shot" feature require complex prompting?
No. The system supports simple prompts. It uses intelligent storyboarding to break a simple prompt into multiple shots while ensuring the key visual information remains consistent across the sequence.
Where can I access Wan 2.6? Wan 2.6 is currently available for use on Atlas Cloud.
