As I reported last week, Google is upgrading its Gemini AI assistant to the updated Gemini 1.5 Flash model, which aims to provide faster performance and expand its capabilities with a larger 32k-long context window.
An APK Teardown of the latest Google app for Android (15.29.34.29) by Android Authority reveals that Google is working on a feature to streamline the image generation process of Gemini. Currently, users need to start from scratch if they want to modify any aspect of an AI-generated image, which can be a lengthy task if the initial output is not satisfactory.
The beta release of the Google app introduces a new tool, enabling users to fine-tune details in AI-generated images. With this tool, users can make specific adjustments to an already created image using natural text prompts, thus avoiding the need to recreate the entire image.
Enhancing User Experience
This addition offers two main advantages. Firstly, it maintains the fundamental elements of an image while allowing users to modify specific sections through text prompts. Secondly, it eliminates the cumbersome process of generating a completely new image for minor tweaks.
Users will have the flexibility to make image adjustments either via text prompts or by manually circling the areas they want to edit. Using a manual option is particularly beneficial for devices equipped with a stylus, simplifying the specification of which parts of the image require changes.
Google Gemini 1.5 Flash
Gemini 1.5 Flash aims to elevate both response time and cognitive capabilities. A notable enhancement is the model’s expanded context window, quadrupled to accommodate up to 32,000 tokens. This significant increase empowers the AI to process more complex and extensive prompts, thereby augmenting its utility for users.
To enhance user experience and factual accuracy, Google has introduced several refinements. The AI now appends links to relevant content for fact-based queries, indicated by a gray arrow at the paragraph’s conclusion. The feature expedites users’ access to credible information, thereby enhancing the AI’s efficacy as a research and verification tool.
Last Updated on November 7, 2024 3:29 pm CET
