# Wan-2.2 Image Canonical page: https://fuser.studio/models/wan-2-2-image Wan 2.2 Image in Fuser generates images from text with Alibaba's Wan 2.2, in Lite (5B) and Pro (14B) versions. Refine and upscale on one canvas. ## Specs | Spec | Value | | --- | --- | | Versions | Lite (5B), Pro (14B) | | Inputs | Prompt (text), Negative Prompt (text) | | Output | image | | Cost | 22–35 credits per run | Cinematic film stills, high-end editorial portraits, and complex multi-character scenes with tactile photographic depth Creator: Alibaba Cloud Tongyi Lab [Render cinematic stills](https://app.fuser.studio) ## How Wan-2.2 Image works Wan-2.2's dual-expert MoE architecture translates cinematic scripting into tactile visual assets through a streamlined three-step workflow. ### Frame the cinematic scene Structure your prompt with distinct subject, action, lighting, and camera angle directions separated by hard periods. ### Select model tier and shift Choose Lite 5B for rapid layout iteration or Pro 14B for final texture fidelity, tuning shift for stability. ### Render high-dynamic frames Execute the dual-expert diffusion pipeline to produce filmic imagery with organic grain and tactile surface detail. ## What Wan-2.2 Image is good at Built for filmic fidelity, Wan-2.2 excels at realistic skin pores, volumetric atmosphere, and multi-subject composition across flexible model tiers. ### Dual-expert MoE architecture A high-noise expert establishes compositional balance while a low-noise expert renders fine fabric weaves, skin pores, and natural surface textures. ### Volumetric cinematic lighting Simulates complex film lighting setups with deep ambient falloff, directional rim lighting, and atmospheric dust haze across 2 to 50 inference steps. ### Pro and Lite model tiers Select the fast 5B Lite variant for rapid composition checks, or switch to the 14B Pro active parameter model for pristine anatomical precision and material reflections. ### Tactile material fidelity Renders authentic physical surfaces—lacquered metal, weathered timber, and raw textiles—without synthetic plastic sheen or artificial smoothing. ## Made with Wan-2.2 Image Explore how creators harness organic depth, tangible material textures, and dramatic lighting across film, editorial, and design disciplines. ### Low-light environmental documentary portrait ### High-speed editorial fashion lookbook ### Textured architectural interior study ### Sculptural industrial vehicle hero ### Atmospheric cinematic neo-noir film frame ## What people build with Wan-2.2 Image From cinematic concept development to luxury editorial campaigns, creative professionals rely on Wan-2.2 for uncompromising realism. ### Cinematic Film & TV Pre-Visualization Directors and concept artists can storyboard complex narrative sequences with authentic 35mm film textures, dramatic atmospheric depth, and multi-character staging. ### High-Fashion Editorial Campaigns Art directors and fashion brands create striking digital lookbooks with genuine fabric drape, micro-textures, and bold editorial lighting setups. ### Photorealistic Character Design Game studios and character artists craft emotive portraits with tangible skin pores, realistic weathered features, and natural hair strands without plastic artifacts. ### Architectural & Interior Visualization Interior designers and architectural firms simulate light interaction across complex raw materials, wood grains, and textured surfaces. ### Commercial Product Photography Brand teams and industrial designers produce hero asset imagery emphasizing realistic reflections, machined metal edges, and bespoke material finishes. ## Frequently asked questions ### What is the difference between Wan-2.2 Lite (5B) and Pro (14B)? Wan-2.2 Lite (5B) is optimized for fast, cost-effective prototyping, while Pro (14B) delivers maximum photorealism and fine-grained texture resolution. Lite is ideal for testing compositional layouts and drafting simple subjects. Pro activates 14 billion parameters per step to render realistic skin pores, complex fabric weaves, and coherent multi-subject interactions. ### Which Wan-2.2 variant should I use for commercial production? Choose the Wan-2.2 Pro (14B) variant for high-end fashion lookbooks, detailed character photography, and cinematic stills. Pro utilizes the full dual-expert Mixture-of-Experts pipeline to resolve anatomical details, fine garment textures, and nuanced surface reflections that the smaller Lite model may simplify. ### What types of images is Wan-2.2 best at generating? Wan-2.2 is best suited for cinematic film stills, high-end editorial fashion photography, intricate character portraits, and complex multi-subject scenes. Its dual-expert architecture separates macro compositional layout from micro-textural detailing, producing natural lighting falloff, filmic shadows, and realistic surface grain. ### What kinds of projects should avoid Wan-2.2? Avoid using Wan-2.2 for minimalist flat vector graphics, UI icons, line art, or clean typography and text rendering. It is also not designed for real-time, low-latency web applications requiring sub-second image generation over photographic depth. ### Who created Wan-2.2 and how does its architecture work? Wan-2.2 was developed by Alibaba Cloud's Tongyi Lab. The model uses a Mixture-of-Experts diffusion architecture with 27 billion total parameters and 14 billion active parameters per step. Denoising is split between a High-Noise Expert that locks in overall composition and a Low-Noise Expert that refines organic micro-textures. ## Try Wan-2.2 Image on Fuser Cinematic film stills, high-end editorial portraits, and complex multi-character scenes with tactile photographic depth [Render cinematic stills](https://app.fuser.studio) ## Related models - [Gemini Image](https://fuser.studio/models/gemini-image.md) - [GPT Image](https://fuser.studio/models/gpt-image.md) - [Seedream](https://fuser.studio/models/seedream.md) - [FLUX.3 Image](https://fuser.studio/models/flux-3-image.md) - [FLUX.2](https://fuser.studio/models/flux-2.md) - [ESRGAN Upscaler](https://fuser.studio/models/esrgan-upscaler.md) Node reference: [Wan-2.2 Image docs](https://docs.fuser.studio/docs/nodes/image/wan-2-2-image.md)