Which AI Art Generator Will Become The Standard?

At this point, there’s no stopping AI art, despite the controversy in the art world. There are several different AI art generators already gaining a lot of traction. The question now is: which of them will become the gold standard of AI image platforms (as Photoshop did years ago for editing?)

I have tried a variety of AI platforms. I will give a short summary of each, and choose the “winner.” In other words, I am going to predict which AI platform will become the most popular in the months to come — the one people automatically associate with AI art.

I rated each platform in the categories of ease of use/user experience (UX), imaging power/results, flexibility/range, and value.

DALL-E 2

This AI generator from OpenAI is the first that I joined via waitlist, and it was worth it.

Ease of use/user experience (UX): DALL-E 2 is quite simple to use, with an interface that’s not intimidating. You don’t have to be a pro in order to use this platform. You can even easily upload your own photo as a starting point.

Imaging power/results: In terms of how impressed I am by the results, DALL-E 2 might be at the top. My draw dropped when I typed in the simple prompt of “steampunk dogs”, and it gave me this (as one example):

DALL-E has also introduced “outpainting,” which lets you imagine an image extended beyond the frame using AI.

Flexibility/range: I’ve found DALL-E to be one of the more “prudish” platforms. It rejects prompts that seem like adult content. It also seems to reject requests for images of known people, for example Taylor Swift.

DALL-E allows personal/commercial use for the images you create. (However, commercial use and copyright are not the same. While you can use AI images quite liberally in most cases, it doesn’t stop others from using it.)

Value: Aside from the 50 free credits I got from DALL-E upon joining, it’s quite economical. You can buy 115 credits for $15 USD, and that usually lasts me a 3–4 weeks at my pace. It awards you 15 free credits each month.

Midjourney

This is the second platform I tried, and with great results. However, it’s not quite as straightforward to use as some of the alternatives.

Ease of use/user experience (UX): This may be the only downfall of Midjourney, at least for newbies. You generate the images through Discord, and there are a few essentials to learn (for example, typing /imagine to tell the machine you’re ready to prompt it.)

Imaging power/results: Midjourney is endlessly impressive with its results. It faithfully produces detailed images across many styles. Like, here is a fantasy princess in a medieval setting:

Flexibility/range: There are a lot of “modifiers” you can add to Midjourney prompts to influence the output, if you care to master them. It also recognizes many celebrity names, but is mildly restrictive in terms of censorship.

Value: I opted to subscribe to Midjourney, choosing the basic plan for $10/month. This gives around 200 images per month. If you exceed that, you can also set a limit for spending and keep going until the subscription renews. For $30/month, you get unlimited generations.

If you don’t want your creations to end up in the public Discord stream, you can pay a fee for private visibility. (I have been able to learn from prompts used by others in the “general” chat.)

DreamStudio/StabilityAI

This is — you guessed it — the third platform I tried. While some of the results are impressive, I probably will not be buying any more credits, as I like the other platforms better.

Ease of use/user experience (UX): DreamStudio has a fairly straightforward input method. There is a nice sliding scale that lets you control features such as image output size, and you can also upload a photo as a base. The platform provides a brief prompt guide to help you along.

Imaging power/results: This AI platform can produce photorealistic results. However, I find that the prompt and the result are often at odds, even when I slide the Cfg scale higher.

I’ve also noticed that it likes to put multiple people into the image when I’ve clearly asked for only one, as shown in the sample below. The man in the image seems as confused as I am about the result.

Flexibility/range: DreamStudio will respond to your style preferences, for example, a cartoon instead of a realistic figure. However, I find there’s a lot of “oddities” in the results, even the ones that aren’t supposed to be realistic.

Value: I find the credits a bit confusing for this platform. The number of credits per image varies with your output settings — the bigger the output and the more detailed, the more it costs. I have bought some additional credits (for $10), but looking at my subscription page, I’m not sure exactly where I stand.

NightCafe Creator

This may be the friendliest AI art generator for rookies. As soon as I joined, I could easily create stunning images without learning fancy prompts.

Ease of use/user experience (UX): As noted, this platform is simple to use. You can enter a plain English prompt, and then choose one of the modifier presets to influence the style. For example, you can choose everything from b/w portrait to cyberpunk with the click of a button.

There’s also an advanced mode that lets you control output size (as an example), and also allows you to upload an existing photo to work from.

Imaging power/results: NightCafe produces great results in almost all of the styles. However, the b&w portrait mode blows me away. With very little prompting, it gives me an image that looks almost real.

However, the standard output size (512x512 pixels) is lower than DALL-E and Midjourney (at 1024x.) However, you can increase the size using more credits.

Flexibility/range: You can create many different styles easily using the presets. The platform also allows you to input modifiers (like adding f2.8 depth of field for portraits) to add to the realism.

Value: One of the best things about NightCafe is that it gives you 5 free credits each day to play around with, along with occasional bonuses. I usually end up using the credits as soon as I get them, but you can bank them as well.

You can also buy a subscription starting at $9.99 for 100 credits, which should keep you busy for a while. I would say in terms of cost to imaging power, this may be one of the best to try.

Imagen

This is Google’s latest foray into the AI image world. Unfortunately, while the samples look incredible (what platform has more image information than Google?), it doesn’t seem to be open to the public yet.

Imagen claims to be above the others in terms of realism, but until it lets me join, I will not attempt to prove this (because I can’t.) However, I expect Imagen to be a top contender based on the sheer power of Google.

While Meta (Facebook) doesn’t seem to have an AI image generator, it has launched a text-to-video tool that I have not tried.

So… which AI art generator wins?

It’s clear that AI generators are here to stay. Traditional artists should accept this, and learn to benefit from the tools instead of fighting them.

It’s tough to say who comes out on top, at least for now. AI imaging is really just in its infancy. However, I predict one of these platforms will emerge as the standard for AI image generating.

If I were to make a guess right now based on ease of use and image quality, I would likely choose DALL-E as the winner. It continues to blow me away with its results using simple prompts.

In second place I would put NightCafe, for its impressive power and accessibility.

What do you think? Who wins among the AI generators you’ve experimented with so far?