TLDR: Vidu, a product of Chinese firm ShengShu Technology, has launched a significant update to its AI image generation platform, introducing a ‘reference-to-image’ feature. This new capability allows users to upload up to seven reference images, which the AI model then interprets using ‘semantic understanding’ to compose highly consistent and realistic generated pictures. The update aims to provide users with unprecedented control over visual elements, enabling quick and precise editing of photographs and the creation of imaginative realism.
Vidu, the flagship artificial intelligence offering from Chinese technology firm ShengShu Technology, has rolled out a transformative update to its AI image generation platform. The new feature, dubbed ‘reference-to-image,’ is poised to ‘reinvent photography’ by enabling users to upload multiple image references, which an AI model then seamlessly integrates into vivid, highly consistent generated pictures. This advancement, released on September 8, 2025, builds upon the company’s expertise in generative AI video and foundation models, extending similar capabilities to static image creation.
The ‘reference-to-image’ functionality allows users to provide up to seven distinct reference images. Vidu’s underlying AI model employs ‘semantic understanding’ to interpret the intricate relationships between these multiple inputs, resulting in generated content that boasts superior consistency and imaginative realism. This level of control and coherence has historically been a challenge in AI image generation, with recent breakthroughs from models like Google LLC’s Gemini 2.5 Flash Image (also known as ‘Nano Banana’) making such capabilities more accessible.
The practical applications of this update are extensive. Users can leverage the feature to generate entirely new images from a text prompt combined with several reference pictures. This facilitates rapid and highly consistent editing of existing photographs. For instance, a photographer could modify a wedding picture by adding a bouquet, altering the style of flowers on tables, or adjusting the lighting to compensate for a gloomy day. Individuals can also use the function to refine a selfie, change a logo on their attire, or transpose themselves into a different setting. Marketers and advertisers are expected to benefit significantly, as the tool allows for the swift composition of AI-generated ‘photographs’ featuring products with precise visual control.
Also Read:
- Amazon Unveils ‘Lens Live’: Real-Time AI Visual Shopping Experience
- AI Revolutionizes Collectible Creation: PicLumen and Google’s Gemini 2.5 Flash Drive Transform 2D Art into 3D Figures
Vidu’s latest offering enters a competitive landscape, vying with solutions such as Google’s Nano Banana and Black Forest Labs Inc.’s Flux Context in the realm of generative image editing and production. However, Vidu asserts that its model distinguishes itself by delivering ‘unmatched image and character consistency, along with natural image blending for richer and more realistic details.’ A notable claim is its ability to accurately carry over both visuals and embedded text from reference images with remarkable clarity – an area where modern generative AI image models often struggle.
