Dall-E 3 vs Midjourney vs Stable Diffusion XL comparison. Which is the best AI image gen tool?

With so many great options out there, which text-to-image generative AI tool should you use at the end of 2023?

[Nephalen]

They interpret prompts differently. These prompts were written to fit Dall-3 better in the way they were structured

[yurijmikhassiak]

Thanks. I see huge differences in quality if you need many people in picture. Tools have issues in counting number of people, and skin/face quality goes down with each added person.

[jugibur]

Thanks for sharing! I think in the next version MidJourney will present a better text creation.

[speedy_o]

Dall-E 3 understands the prompts better, SDXL gives better overall images but you need to be more precise with your prompts. In a year all of this should hopefully have converged nicely

[linda]

I just used Dalle-3 on my iPad for free. Sometimes it worked and sometimes it didn’t. I guess it’s only free for a few images a day. Any way I just want to learn for the fun of it.

[Rehbet]

Are the onscreen mouse movements generated too?

[angelito]

To me the worst fail is that even assuming you'll need to fix some details in photoshop, the results are far away from something you can use professionally. Doesn't matter how much you improve the text prompts and image prompts, there are too many objects, character, poses, situations, etc. that become just impossible to generate, despite being not so 'strange'. I hope in just a few months these three main AI Image Generators will overcome all this.

[schtroumpfyoda]

la blague du chat sans scaphandre ... ouh là ! ;)

[chrisrosch]

Gotta us the correct checkpoint and lora… this video makes no sense