As the AI wars intensify, Google is racing to upgrade its AI capabilities across the board. This includes enhancements to its conversational AI Bard and the release of ImageFX, a new AI image generation tool.
Bard Gets Built-In AI Image Generation
The biggest news is that Google is upgrading Bard to allow AI image generation directly within the chatbot. Previously, Bard could only generate text responses during a conversation.
Now, it can also produce images from text prompts, bringing it into tighter competition with ChatGPT and other rival AI assistants.
Also Read: Microsoft quietly releases Copilot AI Chatbot app for Android users
This image generation capability comes through integrating components of Imagen, Google’s leading text-to-image model. It greatly expands what Bard can do, allowing more visual and interactive conversations.
Google Unveils Standalone ImageFX
In addition to supercharging Bard, Google also launched ImageFX, a dedicated AI image generator like Midjourney and Stable Diffusion. ImageFX gives creators yet another way to turn language into imagery.
Users simply type a text prompt, and ImageFX produces a high-quality image through diffusion models. Google says ImageFX often requires fewer tries to hit the desired result compared to other AI image generators.
ImageFX also introduces “expressive chips,” which display similar images that capture different nuances. This allows creators to easily explore variations on their original idea.
Under the hood, ImageFX runs on Imagen 2, Google’s latest text-to-image model. Imagen 2 comes from DeepMind’s cutting-edge advancements in diffusion models. It generates images with exceptional detail and minimal visual artifacts.
The same Imagen 2 also fuels the new built-in image generation capabilities of Bard. By leveraging Imagen 2, both Bard and ImageFX deliver some of the best AI-generated images available today.
Trying Out ImageFX Myself
After testing ImageFX, I’m genuinely impressed with how the images turned out. The expressive chips also work well to produce variations. One limitation I noticed is that ImageFX still falters at generating realistic hands and fingers, as shown below.
This is an issue that plagues all AI image generators, and this is largely because within AI datasets, human images display hands less visibly than they do faces. To successfully depict hands and fingers, AI would need more reference photos with hands as the main focus.
Updates Across Google’s AI Offerings
Alongside ImageFX and Bard, Google is improving MusicFX too for better quality and faster music generation. All these products fall under Google’s AI Test Kitchen, where they develop and trial new AI innovations.
Currently, ImageFX and MusicFX are available in English to users in the US, Kenya, New Zealand, and Australia. With Google making big bets across AI generations, we can likely expect wider availability soon.
Have you given ImageFX a try yourself? What has your experience with it been? Let us know your thoughts and reactions in the comments below!
