AI artists are taking over: Stable diffusion

Less than a month ago, StabilityAI has released an open-source Deep Learning model that can let anyone generate art and images. This is big news since even you can have access to it! I will show you what kind of images it generates and then give you the link to where you can also generate art on your own. Let’s get into it.

Human artists beware!

The amount of detail in the art generated with Stable diffusion is amazing. A few days ago I started playing around with it and it keeps astonishing me with the results. Let me show you a quick example of this. Here is what Stable Diffusion generated for me for the prompt “Ship on a stormy sea”.

Pretty impressive, right? I am extremely talented indeed. Of course, I can paint pictures with a few clicks and my imagination. If you make me draw something on a piece of paper, it will look horrifying.

Considering that I could generate that piece of art using just a few clicks and writing a few words, it makes you really wonder — do we even need painters anymore?

I believe that the next-gen artists are actually people who can use the best tools available to create the most visually amazing art. Artists can use their imagination along with tech to build things up, change things, and create artistic wonders. This is the future.

How to test it yourself

Stable Diffusion is available to everyone, both the code and other, easier-to-use software. I will give you 3 ways in which you can test it, from very easy to code-it-yourself.

The very easy way in which you can test stable diffusion yourself is by using Dream Studio. It’s a free-to-test online tool that you can just click, use an e-mail to register, and get to generating images. The downside is that you only have a limited number of images you can generate since you are running the deep neural network on their servers (which costs them $). Follow this link to test Dream Studio. You can also use Hugging Face’s Stable Diffusion without logging in, at this link.

The second way you can test it is if you have over 8 Gb of dedicated video memory, preferably Nvidia. You can download a simple GUI file at this link. It’s a bit big, but you will have the generator locally, and this will allow you to generate as many images as possible. On my 8GB V-Ram PC, I can generate a maximum of 512x512 images easily.

Now, if you are a coder and want to get your hands down the rabbit hole… the code is open-source, and anyone can take it, tweak it, break it and cry over it. You can find the code at this link. And if you are interested in how Stable Diffusion works, you can find out in this paper. Good luck

Conclusion

With the rise of stable diffusion and the fact that is open-source, we will surely see some awesome generated art around the web. I am really curious about what people’s imagination will bring to the world, assisted by an AI. Remember, you can do this too, you can have fun using it too, maybe even generate a painting, print it, and put it on the wall. You can then brag to everyone who comes to visit that you “painted” it yourself!

Hope you will have fun with Stable Diffusion! Stay safe!