Google Planning to Add AI Image Generator Directly to Bard
Following the Search Generative Experience (SGE) and Slides, Google Bard is set to get its own image generator.
As shared by Dylan Roussel on X/Twitter, an unreleased Google Bard changelog — dated for January 18 — reveals how you can “Create images with Bard.”
Like other tools, you “simply enter a few words to bring your imagination to life” with the ability to download or “Generate more.” It should be powered by Google’s Imagen family of models.
Back to the future? 👀
— Dylan Roussel (@evowizz) January 17, 2024
Here's what's coming next in Bard. . . tomorrow.
Image generation with Bard will use Imagen, Google's Text-to-Image "diffusion technology."https://t.co/FNeiCsS2ar
Important: The content of this changelog may still be changed until officially released. pic.twitter.com/9yRNjHNrLq
In December, Google announced Imagen 2 with “significantly improved image quality” (as seen below). This ability to create “high-quality, photorealistic, high-resolution, aesthetically pleasing images” is aimed at enterprise use cases like logo generation or creating images with text overlays.
Google Search introduced a similar image generator in October. If you’re enrolled in SGE, you can find it at google.com/search/images. “Draw” or “sketch” prompts can be directly entered into the Search field. SGE generates four images at a time, while there are daily limits on generation.
SGE does not generate images of harmful, misleading, or explicit content. Prompts with notable people or photorealistic faces are also blocked. Additionally, every image generated has metadata labeling and embedded/invisible watermarking (using Google’s SynthID).
The image generator in Google Bard should offer similar safety features.
Like Search, Bard is a pretty accessible service. It joins user-facing image generation capabilities in Google Slides and Meet (for custom backgrounds) as part of Duet AI.
