Google is enhancing its Veo 3.1 AI video model to place greater emphasis on the reference images you want your generated clips to be based on. The company is rolling out new visual improvements for the "Video Elements" tool launched last year, alongside expanded native vertical video support and resolution upscaling capabilities.

The "Ingredients to Video" tool enables Veo users to create videos based on up to three reference images, pulling in elements such as character subjects, backgrounds, and textures for better control over the final output. Google states this update will make videos "more expressive and creative," offering "richer dialogue and narrative." There are also consistency improvements that should be more noticeable—Veo 3.1 now ensures characters maintain a consistent appearance across different clips and environments, and allows users to reuse objects, backgrounds, and textures across various scenes.

Clips generated using Ingredients to Video now also support vertical output. This follows Google's move last year to allow developers to generate vertical videos in Veo from text prompts without references. Users can choose to output videos in the native 9:16 aspect ratio, making it convenient for uploading to platforms like TikTok and YouTube Shorts without needing to manually crop results in video editing software.

Google will begin integrating the improved "Video Ingredients" and vertical mode features into the Gemini app starting today, and will also incorporate these tools "for the first time" into the YouTube Shorts and YouTube Create applications.

Finally, this update allows Veo 3.1 users to upscale generated videos to 4K resolution, surpassing the previous 1080p limit. Google notes that 1080p video generation technology has also been refined to deliver "sharper, clearer videos." This is not the native 4K resolution Google claimed Veo could achieve in 2024—we haven't seen such resolution in any publicly released version of Veo yet—but upscaling on the platform is better than nothing.