Creating an avatar can be both easy and hard, depending on how complex that avatar is going to be.
Wan wants to help with that, with the release of a tool called 'Character X' in the platform's playground section. The feature is designed to let users create entirely original AI characters without uploading a reference image. Announced through Wan's official channels, the feature allows users to generate unique character identities by selecting attributes such as gender, ethnicity, age, facial structure, hairstyle, eye shape, accessories, and other visual traits.
Instead of uploading or modifying an existing photo, Character X builds a new face from scratch, producing a ready-to-use character that can serve as an avatar, story protagonist, or recurring digital actor.
What makes Character X notable is its integration with the broader Wan video-generation ecosystem.
Once a character is created, it can be assigned as a starring role and reused across animation, character replacement, and reference-driven video workflows.
The release extends Alibaba's ongoing efforts to improve character consistency in AI-generated videos, an area where the Wan family of models has already focused heavily through technologies for character animation, identity preservation, expression transfer, and multi-scene storytelling.
A demonstration shared alongside the announcement shows users selecting facial attributes from a dedicated playground interface before generating a fully rendered portrait. In one example, the system creates a detailed fantasy-style female character with white hair, facial scars, and armor, illustrating how Character X can produce stylized identities that are immediately available for use in subsequent video-generation tasks.
Rather than functioning as a standalone image generator, the tool appears designed as the first step in a larger workflow where original characters can be animated, inserted into scenes, and maintained consistently across multiple outputs.
New feature alert on Wan!
Meet Character X: Create a unique face instantly.
Whether you need a custom avatar, a brand new character, or a completely unique identity, Character X brings your vision to life with ease. pic.twitter.com/zuesqVxrgf— Wan (@Alibaba_Wan) June 8, 2026
Alibaba’s Tongyi Lab develops the Wan series of AI models for image and video generation.
These tools form part of the company's broader work on multimodal visual systems under the Tongyi brand, also known as Tongyi Wanxiang in Chinese references. The platform supports text-to-image, image-to-video, text-to-video, editing, and reference-based generation, with versions such as Wan 2.1 through 2.7 released over time.
Some components have been made available as open source on platforms including GitHub and ModelScope, allowing researchers and developers to access model weights and code.