88Z AI, 1 April — Tongyi Lab has released the Wan2.7-Image model, focusing on human realism, text stability, and color accuracy. The model supports deep customization of virtual avatars to avoid homogeneous faces, stably generates images with ultra-long text including multiple languages, tables, and formulas, and introduces a Hex value palette for precise control of brand colors.
At a glance
- Supports deep customization of avatars and stable generation of multi-language text up to 4000 characters
- Introduces Hex value palette for precise control of brand colors
The update
Tongyi Lab has officially launched the Wan2.7-Image AI model for image generation and editing, aiming to enhance performance in human representation, text rendering, and color fidelity. For character generation, the model supports deep customization of virtual avatars, effectively avoiding the common "standard AI face" phenomenon and making characters more personalized and realistic. Regarding text processing, the model can stably generate images containing multiple languages, complex tables, and mathematical formulas, supporting text inputs of up to 4000 characters to resolve previous issues with garbled text. Additionally, the new "Color Control Palette" feature allows users to specify colors precisely via Hex values, ensuring output images strictly adhere to brand visual identity standards and meet professional design requirements.
Sources & editorial note
This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
- 通义实验室Source published
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.