Google Imagen vs. DALL·E 2: Which One is Better at AI Art?

We live in an AI art race era now. After OpenAI released DALL·E 2 in April 2022, Google released their version of text-to-image generator Imagen a few weeks ago in May 2022. For folks who want to learn more about DALL·E2, you can read my article Everyone Can Be An Artist with AI. Like DALL·E, Google Imagen is an AI system that creates photorealistic images from input text. In this article I will do a comparison of images generated by Imagen vs DALL·E.

Google Imagen is currently not available to public. DALL·E 2 is not available to public either, but it is available to a limited number of users who have signed up for their waitlist. So some folks who have access to DALL·E 2 are comparing Imagen with DALL·E 2 by generating images with DALL·E 2 using the same prompts that some of Imagen’s example images use on the Imagen website! Let's see a few side-by side comparisons now. The images below are from these 2 twitter posts: 1) Post by @joeyliaw (the link of the post) 2) Post by @benjamin_hilton (the link of the post).

For all images below, left is Imagen and right is DALL·E 2.

1. "A photo of a Corgi dog riding a bike in Times Square. It is wearing sunglasses and a beach hat."

2. "A dragon fruit wearing karate belt in the snow"

3. "A robot couple fine dining with Eiffel Tower in the background."

4. "A dog looking curiously in the mirror, seeing a cat."

5. "A brain riding a rocketship heading towards the moon."

After seeing these 5 comparisons, do you think if Imagen is better or DALL·E 2? To be honest, I think they are about the same quality. Both are amazing in generating images from text inputs. What do you think? Feel free to comment below.