Skip to main content
  1. Home
  2. Computing
  3. News

Google’s Nano Banana 2 promises to end the AI text rendering problems

The end of squiggly, unreadable AI text? Google seems to think so.

Add as a preferred source on Google
Collage of images created by Nano Banana 2
Google

Google Nano Banana was a hit sensation at its launch, with millions of users creating images using Google’s latest AI image generation service. One of our team members tested Nano Banana’s capabilities by creating social media posts, and the results were surprisingly good. 

That said, like all other generative AI models, including ChatGPT, Nano Banana struggled with creating images containing text. The poor text rendering was always a dead giveaway that the images were generated by AI. 

Recommended Videos

Google claims that it has solved this problem with its latest Nano Banana 2 engine. It delivers improved text rendering along with several other meaningful upgrades.

A smarter, faster way to turn ideas into images

One of the biggest upgrades coming with Nano Banana is how much the model “knows.” Nano Banana 2 can draw from Gemini’s broader world knowledge and pull real-time information and images from web searches.

Google says this helps Nano Banana 2 become more accurate, enabling it to create infographics, transform notes into clear diagrams, and generate accurate data visualizations.

Nano Banana 2 offers vastly improved text rendering and can not only generate legible and accurate text but also translate text to other languages.

More control and enhanced consistency

Nano Banana 2 brings several other upgrades, allowing users to create consistent and production-ready images. Some of the major improvements include:

  • Subject consistency: You can now maintain character resemblance up to five characters and 14 objects. This makes storyboarding and multi-scene narratives far more practical.
  • ‌ Precise instruction following: Another big improvement is instruction following. Nano Banana 2 sticks to your complex prompts, capturing specific nuances and producing an accurate version of your idea.
  • ‌ Production-ready specs: You can finally generate images in any aspect ratio or resolution, making it easier for you to create assets for social media and other projects.
  • ‌ Visual fidelity upgrade: Nano Banana 2 also promises richer textures, sharper details, and more vibrant lighting, delivering significantly better images than before.

More than just an upgrade

If what Google promises is even close to reality, Nano Banana 2 will mark the next evolution in image generation. With high-fidelity images, enhanced text rendering, and character consistency across entire projects, Nano Banana 2 is shaping up to be a game-changing update.

Rachit Agarwal
Rachit is a seasoned tech journalist with over ten years of experience covering the consumer technology landscape.
Layr is a new macOS clipboard manager that replaces hotkeys with trackpad gestures
This new Mac app opens clipboard history with a four-finger tap instead of a keyboard shortcut
Cursor open on Mac

macOS users already have several clipboard manager options, including Paste and Maccy. Most of them work well, but they are usually built around keyboard shortcuts. That is useful for keyboard-heavy users, but it can feel out of place for users who rely on the trackpad for most of their work.

Layr, a new clipboard manager from the developer behind Declutr, takes a different approach. Rather than assigning a keyboard shortcut to open the clipboard history, the app lets users bring up a clipboard overlay with a four-finger tap on the trackpad.

Read more
YouTube’s AI content labels are getting a much-needed makeover
No more hunting through descriptions. YouTube's AI labels are finally moving front and center.
YouTube ai declaration longform video

This year’s Google I/O marked the transition of Google from a search company to a fully AI-focused company. The company launched several AI tools, but the one that matters the most for YouTubers is Google Omni, built for video generation and editing. 

While tools like Omni lower the barrier for creators, which is a good thing, it also results in the platform being inundated with low-effort AI content. The company understands that this will annoy a large percentage of its users, so it has been asking creators to disclose AI-generated content since 2024. 

Read more
AI models have a religion favoritism problem, and new research exposes it
AI models are subtly steering users toward certain religions, and most people have no idea it's happening.
Artificial Intelligence

A new research consortium has found something worth paying attention to: when you ask AI about grief, love, loss, or moral decisions, it almost never brings religion into the conversation.

The Consortium for Evaluation of Faith and Ethics in AI (CEFE-AI), a collaboration among researchers at Brigham Young University, Baylor University, the University of Notre Dame, and Yeshiva University, published its findings this week at the Summit on AI Ethics in Athens, Greece.

Read more