AI Art Face-Off! DALL-E vs Mage vs Bing!

By Lou Wallace

With all the news and interest in all things AI today, I thought it would be interesting to see how each compares when given the same prompts. (There was one exception, see below…)

Just to be clear none of these images represents the best any of these can do. I just entered specific textual prompts and let them do their stuff. In every case, each image generator can produce better images by fine-tuning the settings and text used to describe the desired scene. However, just jumping in and typing a description off the cuff is how most people use these. 

So having said that here we go!

Prompt: scifi robot walking down a city street, photorealistic and ray traced

DALL-E

Bing

Mage

Prompt: roses growing in abandoned urban lot at night during full moon, fantasy

DALL-E

Mage

Bing

Prompt: beautiful woman sunbathing on a beach while her dog plays in the surf, photorealistic

DALL-E

Mage

Bing (Bing required beautiful woman reclining on a beach while her dog plays in the surf, photorealistic)

Prompt: a modern open design office, photorealistic and ray traced

Bing

DALL-E

Mage

Prompt: children’s birthday party with cake and ice cream, photorealistic

Bing

DALL-E

Mage

Prompt: SpongeBob and Patrick in the gothic style of Castlevania

Bing

DALL-E


Mage

Bing seemed to do the best for human forms (but it did not allow sunbathing as a keyword).

Mage makes some good images but human form isn’t good with out finetuning it.

DALL-E was good at the robot and humans, but it didn’t follow the guideline for the urban rose prompt.

If you want to try these AI art generators yourself, here are the links:

DALL-E

https://labs.openai.com/

Mage

https://www.mage.space/

Bing

https://www.bing.com/images/create