Google Bard adds image generation

You can now use Bard to generate "high-quality, photorealistic" images in English, in most countries around the world, Google announced.

You can now generate images for free using Google Bard. The new capability is available in English, in most countries, Google announced today.

Why we care. AI image generation tools have taken giant leaps forward in the past year, with Midjourney leading the way and other solid options like DALL-E and Bing Chat/Copilot. Now you have a new way to generate visually interesting image content for free.

How it works. Simply enter a prompt for the image you want to create (e.g., “create an image of a dog riding a surfboard”). Bard will present you with four images, with the option of generating more.

What it looks like. Here’s a screenshot Google provided:

Google also shared multiple other examples of Bard-generated images:

Powered by Imagen 2. The technology behind Bard’s image generation is a model called Imagen 2. Google said Imagen 2 “delivers high-quality, photorealistic outputs.” Based on the above images, it seems images generated using Bard should deliver on that promise.

Watermarked. All images created with Bard will be marked with a SyntID digital watermark.

Limits. Google said Bard will limit violent, offensive and sexually explicit content. Google will also try to avoid generating images of named people using filters.