Users may upload personal photos or files to influence the artwork. The system supports multiple illustration styles, including pixel art, comics, clay models, crochet, and outlines for coloring. Each book reflects the user’s original input. The stories can explain basic concepts, teach lessons, or turn a child’s sketch into a digital narrative. Audio narration is included, and the feature works in all languages.
The company recently added other functions to Gemini. One tool converts photographs into short videos using Veo 3. Another, called Deep Think, allows the model to weigh multiple interpretations before answering. These updates aim to improve both visual output and reasoning ability.
On a separate track, Google DeepMind has released Genie 3. It is a new version of its AI model for creating virtual environments. This version supports longer sessions and remembers objects across user interactions. When a user looks away from a detail, such as a written note or color pattern, the system keeps it in place when viewed again.
Users can input text prompts to produce these virtual spaces. Genie 3 accepts instructions for generating characters or adjusting the environment, such as changing weather. The scenes run at 720p and 24 frames per second.
Earlier versions of this model could only manage 10 to 20 seconds of user interaction. Genie 3 now handles a few minutes of continuous engagement. The system creates interactive spaces from scratch rather than using prebuilt assets.
Access to Genie 3 is limited. Google is offering it to selected researchers and developers. The company plans to study performance, risks, and use cases before expanding access. Some limitations remain, including difficulty generating readable in-world text unless it is explicitly described in the input.
These changes in Gemini and DeepMind suggest an ongoing effort to improve AI for personal, educational, and creative use. The tools are being tested for reliability, memory handling, and user-guided interaction. The company is updating its platforms in phases, with more features expected in future releases.
Notes: This post was edited/created using GenAI tools.
Read next: OpenAI Releases Two Open-Source Models to Regain Ground Against Global Competition