Love it or hate it, AI art generators have already proven that they’re one of the most disruptive new pieces of technology to date. For casual users like me, nothing beats the thrill of having an idea in the middle of the night, turning it into a prompt, and having the artwork be actualized in just a matter of seconds.
Today, I actually juggle between five AI art generators depending on what I need. But if I were to compare all of them, which one comes out on top? That’s what I’m here to find out.
In this article, I’ll introduce you to Midjourney, DALL-E 3, Meta’s Imagine, Adobe Firefly 2, and Stable Diffusion to find out which one you should use on a daily basis.
Midjourney
Since its release, Midjourney has consistently been one of the most popular AI art generators in the market with more than 16 million users and counting. Apart from having a robust model that’s optimized for creativity, Midjourney’s biggest strengths are its consistent developer support and many features like panning and zooming, style and character reference, and image descriptors.
You can read more about Midjourney in our full review which can be found here.
Pricing
Midjourney has flexible pricing options perfect for any use case.
- Basic: $10 per month or $96 per year. 3.3 fast GPU hours per month with no relaxed hours, which means you have a limited number of generations monthly. Maximum of 3 concurrent jobs with 10 in queue.
- Standard: $30 per month or $288 per year. 15 fast GPU hours per month with unlimited relaxed hours. Maximum of 3 concurrent jobs with 10 in queue.
- Pro: $60 per month or $576 per year. 30 fast GPU hours per month with unlimited relaxed hours. Maximum of 12 concurrent fast generations, 3 relaxed generations, with an additional 10 in queue.
- Mega: $120 per month or $1152 per year. 60 fast GPU hours per month with unlimited relaxed hours. Maximum of 12 concurrent fast generations, 3 relaxed generations, with an additional 10 in queue.
You can also purchase additional GPU hours if you run out for $4 per hour.
Personal Opinion on Midjourney
If you were to ask me for AI art generator recommendations, the first thing out of my mouth will always be Midjourney. It’s not just a matter of preference, but absolute dominance when it comes to creating artwork and even realistic images. There’s just no real flaws in Midjourney’s model apart from text generation and prompt understanding, which are both struggles for all image generators anyway.
What’s more is that if you’re only looking for anime artwork, you can also use Midjourney’s anime-optimized model called Niji. We have a separate review for that model here.
Art Showcase
DALL-E 3
Not one to fall behind in the AI space, OpenAI also has an AI art generator called DALL-E, which is now currently in its third iteration. This model promises to be more creative and more nuanced than its previous versions, something that they’ve definitely achieved. Another thing about DALL-E 3 is that you can use it with ChatGPT, so you can directly create images using conversations as context or prompt.
If you’re interested in learning more, here are our complete thoughts on DALL-E 3.
Pricing
You can use DALL-E 3 with ChatGPT Plus for their monthly subscription of $20. Alternatively, you can also use DALL-E 3 for free using Bing Create.
Personal Opinion on DALL-E 3
DALL-E 3’s biggest strength comes in its prompt understanding. Powered by GPT-4, there’s no other AI image generator that comes close to DALL-E’s nuance and uncanny ability to include every piece of a prompt into an image. It’s also one of the first AI image generators to feature text generation in their images, which was a long-time struggle for image models.
My issue with DALL-E 3 stems from its creativity. While it’s capable of creating amazing images, I’ve found that it’s more difficult to have consistent results using this model than other image generators in the market. It’s also not a good generator for realistic images because they tend to generate people that look “too perfect” — sharp jawlines, faces that are too smooth, etc.
Art Showcase
Meta’s Imagine
Meta also has an entry for AI art generators with its Imagine model. This one’s a lot more straightforward than others — it doesn’t have too many fancy features except for image generation and chat history. While it’s certainly capable, people are wary of this model because of an issue stemming from their training set, which featured public images from Facebook and Instagram profiles.
We also have a full review of Meta, which you can read here.
Pricing
Meta’s Imagine is available for free, but it has only been rolled out to the USA, Canada, and the United Kingdom.
Personal Opinion on Imagine
I’ve always found that Imagine’s strengths isn’t with creating art, but generating realistic images. Like I mentioned earlier, it’s a capable model, but it’s still a little behind when it comes to creating artwork. Their artwork sometimes reminds me of mid-2000s DeviantArt, for better or worse.
Art Showcase
Stable Diffusion
Stable Diffusion has been a staple and fan favorite of the AI art community for years. This open-source model has been around since 2022, and has only garnered more attention over the years.
We’ve reviewed Stable Diffusion XL in the past, so make sure to check that one out.
Pricing
It’s free and open-source! However, usage still depends on which site you’re using. For Dream Studio, you have 25 free credits (which is almost equal to 25 generations) but after that, you have to purchase them for $10 per 1000 credits.
Personal Opinion on Stable Diffusion
Stable Diffusion is a crowd favorite for a reason. Sure, we can talk about how its base model is already robust enough, but its main draw is that it’s open-source. That means anyone — and I mean anyone — can make modifications to make the model so much better. This has led to the rise of custom models or LoRAs built on top of base Stable Diffusion, which makes this generator more creative and nuanced.
Art Showcase
Adobe Firefly 2
Lastly, we have Adobe Firefly 2. This AI image generator’s main pull is its direct integration with the Adobe ecosystem, which allows you to use the model while you’re editing images or videos with Photoshop, InDesign, Premiere, and more.
Our full review of Adobe Firefly can be read in this article.
Pricing
When you sign up, Adobe Firefly 2 gives you 25 credits for free, which you can use for one generation each. However, all images will come with an unremovable watermark in this tier. Upgrading to their Premium Plan will cost you $4.99 per month for 100 credits monthly.
Personal Opinion on Adobe Firefly 2
I honestly can only recommend Adobe Firefly 2 if you’re using it for inpainting and outpainting in Adobe Photoshop or InDesign. While it has a knack for realism, I just can’t pick it over any of the four already mentioned here. Photorealistic images look too stock-like and its artwork (especially for anime images) has an amateur feel to them.
Art Showcase
Which Should You Pick
All of these AI art generators have their strengths, but which one should you actually pick?
For me, it’s an easy win for Midjourney. Sure, DALL-E 3 might have better nuance, and Stable Diffusion might be better once you train it with LoRAs, but it’s hard to beat what Midjourney offers at its base model. The only model that comes close to it, in my opinion, is Stable Diffusion, but that’s just because it’s open-source.
Struggling to pick one? Don’t worry — we’ve already compared these five models in-depth previously in this article. Have fun creating AI art!
/origin-imgresizer.eurosport.com/2024/05/26/3975983-80695368-2560-1440.jpg?w=150&resize=150,150&ssl=1 "Armstrong the hero as Southampton secure immediate PL return with Leeds play-off win")
