AI images are now part of our daily routine. Something impossible to imagine just a year ago. Stable Diffusion is officially released in August 2022 and in the space of a year and dust, the landscapes of AI images have been forever changed, whether by advances that we find today every week thanks to the Open Source.
And there are so many things coming out that we can no longer keep track of them, so it’s a good opportunity to compare the best AI image generators. And you will be disappointed since I only found 3 that match my criteria. Free enough not to rip me off, good quality images and decent options to control and improve them.
Playground AI
Clearly for me, this is the best AI image generator. I use it daily and its web interface is very intuitive. To think that at one time, idiots gathered on Discord on Midjourney, a terrible thing in terms of ergonomics while Playground AI offers an efficient AI image generator that is powerful enough to integrate all the new features. And what’s more, for a broke “freelance” like me, I especially look to see if it’s free.
And Playground AI allows you to generate 500 images per day (before, it was 1000), but 500 is more than enough for the majority of people. It also has paid subscriptions at $15 and $45 per month giving you more options. In particular, the freedom of image dimensions, but also new functions such as Expand Prompt.
The latter will take your prompt and improve it using AI. With the free version, you have 30 Expand Prompt per month while it is unlimited in the paid version. In addition to having an intuitive interface for classic AI image generation, Playground AI also offers you the Canvas where you can do Inpainting, Outpainting, it’s literally an artist’s space where a lot of Things are made easier by AI.
You also have the filters which are the most popular Custom models in Stable Diffusion. For example, you can use Deliberate, DreamShapper, Juggernaut XL, Startlight XL. And obviously, even in the free version, you can use SDXL, the latest version of Stable whether in base or in Refined mode. They also have their own model called Playground v2 (currently) which is not as beetle-ridden. It also integrates ControlNet (not all) in normal SDXL and in Canvas.
In short, if you are looking for a single AI image generator without the hassle, with enough free content and plenty of options to do what you want, then Playground AI has no equal on the market.
Leonardo A.I
Leonardo AI is also an excellent AI image generator and some will even put it in first position. However, I find it a bit heavy and overloaded. Afterwards, I use these AI image generators on an old PC, but Leonardo AI is mainly intended for semi or professional use. I participated in Leonardo AI when it was available by invitation only and the platform was mainly aimed at video game creators, graphic designers, digital painters, etc.
And above all, the great strength of Leonardo AI is that it is the first AI generator to offer direct training of its own models online. Afterwards, the result was not always up to par, but the training of the Custom models is so boring that it was a real feat.
Another reason why Leonardo AI is in second place is that its free offer is less compared to Playground AI. Thus, the free offer offers 150 tokens per day, 30 upscales per day, 30 background removal per day.
Technically, 1 token = 1 image, but Leonardo has a calculation of tokens depending on the options. If you do an Upscale, then it’s tokens used, if you use their prompt improvement function, then you also consume tokens. So, the 150 tokens would give us more like 10 images per day since the basis of an AI image is that you have to do several tries to find the right one.
Its paid plan starts at $12 (8,500 tokens per month), $30 (25,000 tokens per month), and $60 (60,000 tokens per month). And the guys from Leonardo AI regularly add a lot of options to facilitate and enrich the creation of your AI images and this can be a downside, because we have a ton of options and we don’t know which option has impacted the picture. The big challenge of AI today is not to create beautiful and breathtaking images, but to have control over these images to get what we want.
For example, the most famous option from Leonardo AI is Alchemy which will do something internally to enormously enrich your images and give them some pep, even with simple prompts. But this option is only available in the paid version. This is also why I prefer Playground because it also offers bonus options, but which do not overload the interface. But again, this is a personal criticism of Leonardo AI in relation to my personal use. Others will prefer it to Playground.
Unlike Playground which offers Custom models like filters, Leonardo offers them directly and it also offers its own models like Leonardo Diffusion XL or Leonardo Vision XL in addition to the essentials like PhotoReal, RPG v5, etc.
Bing Image Creator
Less known, but very effective, Bing Image Creator is Bing’s AI image generator which is powered by Dall-E. Recently they even integrated Dall-E. There are no paid offers, but you have daily credits that you can use around 100, so around 70 to 100 images.
It’s a little opaque on how he gives his credits. Dall-e 3 has improved a lot, but it still cannot match Stable Diffusion and its models.
And that’s in the order I use them. As a first choice, I always use Playground, if there is a problem, I switch to Leonardo and if I’m really in trouble, then I go to Bing Image Creator. You can make prompts in French while the other two require English.
We don’t have many settings, the size is only 1024 x 1024. In short, Bing Image Creator is for those looking to quickly generate an AI image, for example, to illustrate an article or a concept.
And that’s just the beginning…of AI image generators
Of course, there are plenty of other AI image generators popping up almost every week. But for me, Playground, Leonardo and Bing Image Creator are must-haves. Often, new generators do less well and they try to charge the user straight away.
Obviously, for the next few years, it will be the AI video generation that will be the main trend and we wonder which platforms will surprise us.
And in pure AI images, I see a trend towards specialization in one area of the image. Platforms that do one thing, but perfectly.
For example, specialized AI generators for logos, photorealism, drawing and painting with AI assistance while giving enough control to the user. We have embryos with Playground’s Canvas, but Leonardo also has his own Canvas (I told you it’s overloaded with options).
0 Commentaires