We have added four new text-2-image models to Integrail, and couldn’t be happier with this. I am sure you know Midjourney — it is still probably the leader — and we have compared it with Dall-E and a basic Stable Diffusion model in the first part of this article. There, midjourney was (and still is) the best, but unfortunately, they do not allow access to them via API, which means if you want a good text-2-image option in your complex AI agents you are out of luck.

This is not true anymore! The four models we tested all provide amazing quality across the wide range of prompts — see for yourself below, and start using on Integrail today! We have tested stylizations, portraits, fantasy settings, aliens, people, toys, book illustrations, game design elements — everything so that you can make a judgment for yourselves!

Traditional Bonus: Grumpy cat in “Scream” by Munch as the last picture for your viewing pleasure :)

Every image in this article out of our 45 prompts is structured the same way, left-to-right, top-to-bottom as shown below:

The caption for the image is always a literal prompt, e.g. above it is “Your representation of the generative text-to-image model”.

Let’s start with the NEW 15 prompts (compared to the previous article), and then we will discuss the 30 prompts from the previous article but generated with these four new models.

Part 1: New 15 prompts

1. Girl with a pearl earring in New York

2. The Witcher

3. Fantasy Landscape

4. Post-Apocalyptic AI Agent

5. French Sailor

6. Blues Guitarist

Only one model is able to understand that there is a woman in the image, but the mood is excellent in all.

7. Isometric Game Art

8. Icons Game Art

9. Drifting Nissan

10. Kentucky Miner

11. Japanese Landscape

This is one of the rare cases where with the exception of Playground 2.5 the quality lacks that of Midjourney quite a bit.

12. Hyppopotamus

13. Cat by Chagal

14. Martian Octopus

15. Portrait of a RedHead

Part 2: 30 “classic” prompts from the previous article

1. Felt Toys Stylization

2. Old Book Stylization

3. Whiskey and Smoke

4. Woman Playing Cello

Models still struggle with people doing complex tasks, such as playing musical instruments (wait till you see the hands playing piano…)

5. Porcelain

6. Bathroom

This is pretty amazing from all the models. I actually generated a wooden house once with different rooms and made a “for sale” ad with it. Had so many inquiries…

7. Sushi

8. Alien Breakfast

9. Mischievous Ferret

10. A Redhead Portrait

The portrait in the 1st part is much better — so it pays off to give more detailed descriptions in the prompts.

11. 60s Kodachrome Sci-Fi

You may need a couple of tries to get things right with more complex stylizations, but they work out quite nicely at the end.

12. Japanese Girl

13. More Landscapes

14. Japanese Ink Stylization

Some misunderstanding of what a “bow” means, otherwise — beautiful.

15. Hands

Aaaa, my eyes, my eyes!! Midjourney is your choice if you want good hands. Playground 2.5 tries.

16. Helmut Newton

17. Monet and Penguins

Take your pick.

18. Motorboat

19. Jewelry

20. Underwater Swimming

This has improved a lot.

21. Vintage Commercial

22. Chef

23. RPG Map

Generating art for games is a popular GenAI application. Together with a couple of examples from the first part of the article, this is something you can work with.

24. RPG User Interface

25. Fantasy Landscape

26. Children’s Book Illustration

27. Fantasy Skunk

Another prompt from DALL-E 3 paper, where all models are excellent.

28. Another Illustration

Another DALL-E 3 paper prompt.

29. As Promised, Grumpy Cat

That’s it for today! Judge for yourself, but we think that all of these models can serve as at least a good starting point in your image projects.

If you like this work, do subscribe to our updates or join Integrail and start building your own AI Agents and experiment with tens of other GenAI models! Here is how:

Go to “Agent Studio”, “Design Agents”, pick the “Text to Image” models that you like, connect them to inputs and outputs — and start experimenting!