Google has released an exciting new capability for its AI chatbot Bard – the ability to generate images from text prompts for free. This positions Bard as a direct competitor to OpenAI‘s DALL-E and Bing Image Creator, which already offer AI image generation.
The launch comes after months of anticipation over when Google would release its own take on the viral image creation trend sparked by DALL-E 2 last year. While still in beta testing, the feature marks a major expansion of Bard’s features as Google seeks to catch up to its AI rivals.
So how does Bard’s new skill work exactly? And how good are the images it creates? Here’s a hands-on look at what users can expect.
Putting Bard’s Image Creator to the Test
I gave Bard’s image generator a test run across a variety of prompts to see what it can do. The results show creativity and photorealism still have some room for improvement compared to DALL-E 2 and Midjourney. But as a free tool, Bard produces quality images that capture the essence of prompts fairly well.
Behind the Scenes: How Bard’s Image Tech Works
Bard’s new functionality comes courtesy of Google’s updated Imagen 2 model – specially designed to create quality images quickly from text descriptions. According to Google research scientist Andrew Goodman, Imagen 2 generates “our highest-quality images yet” through several under-the-hood improvements:
- Understanding semantics better – Images stay truer to prompt meanings
- Increased accuracy in rendering fine aspects and textures
Tips to Craft Prompts for Better Bard Images
Bard can paint broad strokes given sparse prompts like “a cute puppy.” But taking a little time to add vivid details helps it capture what you actually want to see. Let Your Imagination Run Wild with Bard
While still developing, Bard’s new talent opens unlimited creative avenues at no cost. It makes visualizing ideas, characters or products easier than ever. Students can quickly mock up art history projects, graphics for school presentations; hobbyists can manifest works matching their mental pictures. Budding creators in any domain now have an intuitive launchpad for taking visions from minds to screens.
Urooj is a freelance writer of some repute, even if she says so herself. Her goal in life is to be published in every international magazine and visitor-load-weary website. She can be contacted at kaziurooj [at] gmail [dot] com.
AI Terms 101: An A to Z AI Terminology Guide for Beginners
Leave a Reply
You must be logged in to post a comment.