In the world of Generative Image AI, there are many applications to choose from. But two of the most popular in my book are MidJourney and Dall-E 3. Each has their strong suits, and each has their downfalls. But in the past year, both have advanced way beyond what I would have ever thought a computer could create from nothing, and in the year, months, even weeks to follow, they will keep getting better.
Each of these images was created using the same prompt. Dall-E 3 on the left, and MidJourney on the right.
While both create stunning images, you can notice some differences.
Dall-E 3 tends to ‘listen’ to a prompt a bit better than MJ; especially complex ones. An elongated cartoon character in Dall-E, actually looks like what was asked, Whereas MidJourney did the opposite. Both are in New York and have donuts, but MJ had some interesting ideas of where to put the donuts.
 |  |
 |  |
MidJourney tends to create more photo-realistic images. Able to create fine details, and more realistic features.
MJ also creates more dynamic lighting. Understanding complex lighting structure and shadows.
 |  |
 |  |
Where Dall-E shines, however, is with the text. As seen in the bottom two images, the one on the left actually was able to create the requested words. While the MidJourney images, did produce letters, but only a random string of gibberish.
 |  |
So which is better? Who wins the competition? it really depends on your needs.
EDIT: With the release of MidJourney 6.0, MJ is catching up with text and even more realism, so the competition rages on.
Comments