AI art: What is it, and is it really art?
AI art is a new emerging creative field that is transforming the way we think about creativity, the art community and the art-making space. But can it really be classed as art of is it just a bit of fun?
Artificial intelligence has been around for many years, so this type of technology is nothing new, it’s just ever evolving. Heck for many years it has even been used to create drawings, paintings and even music, just on more of a basic level. Over the years companies have perfected the way artificial intelligence reacts with datasets. Due to the increase use in datasets and neural networks we are now at such a creative time for AI. Hence why we have seen such a boom in the news in regards to all of these generative art tools.
Our main focus today is around AI-Generated art, not chatbots (that’s a whole other topic). AI Generated art can create some absolutely amazing images in a matter of minutes or seconds depending on how good your pc is or the service you are using. You could grab a copy of stable diffusion install it on your home pc and then just start creating some amazing masterpieces, this is why there is so much concern around it’s wide use.
AI-generated art has sparked some serious discussions about the way it is creative, some believe it’s just mimicking other artists as it’s datasets are based off of other artists work. In theory you could feed these AI generative art tools a million images of something and it should be able to return that image back to you in it’s own creative way. Wheatear you like AI-generated art or not, this piece of new technology is here to stay and will continue to grow over the next few years. Where it will end up, we are not too sure, but for now it sure is a hot topic of debate.
But, if you have been living under a rock for the past 12 months and have no idea what we are talking about. Or perhaps heard the buzz word “AI art” and would love to find out more, then please let us help explain the piece of Artificial intelligence for you.
So How do AI art generators work?
AI art generators use what is known as machine learning algorithms and deep neural networks to generate images, it’s worth taking note that these generative AI tools don’t just generate art. They can be used for a wide range of reasons. You can create logos, stock images, people, fictitious places or even storyboard ideas using them and of course Art.
The way these AI generative art tools work is by colleting large datasets of images. In the case of using art they will have taken insanely huge amounts of art that already exists. This will form a set of algorithms (rules) on how the tool behaviours. The Ai generated art tool then will find patters and styles when creating new images based off of user prompts.
If you would like to the know the typical process AI goes through to generate these kind of images then here it is:
Data Selection: The first step is the most important. This is where the AI generator is being given a dataset (commonly referred to as models). There are 1000’s of these models online that you can use to create your very own art, stable diffusion allows you to add your own models but other generative AI tools may not. Models for sites like Midjourney are not available to the public.
You could in theory get a model of the Simpsons and reproduce work that looks like it was created by the team. This model would have to have a large amount of images to reference to, since the Simpsons has so many episodes it would not be much of a problem to create this. But this is why some artists are furious about AI Generated art. Basically the AI tool is learning an artists art style which could in turn take their jobs.
Generation: Once the algorithm has been trained, it can be used to generate an image based upon that training. This can be rather in depth for some tools or pretty much automatic for others. Each time an image is created it will add a random seed and use the setting pre configures to create the image. With this information it will grab a user created prompt and try to generate an image based on the training it has received.
Imagine going to a class and learning how to draw someone’s face. Then you go to another class and have to draw that face from memory. It won’t look the same but it might be quite similar. If we stick with the same seed when generating AI art we have the chance to repeat the same image multiple times.
Refinement: This is the fine tuning process. When AI images were first generated they were pretty basic, sure the shape and idea was there but the image quality was pretty poor. With the increase in new technologies we can now refine the whole process.
This involves using additional algorithms and techniques, such as adding context through a positive prompt, removing issues with an image through a negative prompt or image filtering by using more steps. This is all used to create a final image that is more aesthetically pleasing.
AI art generators can use a large amount of algorithms and methods to change the look and feel of their image. It’s basically just trial and error at the moment. Some popular AI art generators include Midjourney, Stable Diffusion and Bing Image Creator.
How can you create art using AI?
If you are keen to create your own artwork using AI there are a handful or ways you can do this. But it really depends on what your needs are. At the moment these are the moist common ways you can create AI Art:
Use an AI Art Generator: If you have a decent enough computer with a good NVidia graphics card (over 8gb of VRAM) then you might be interested in stable diffusion. Stable diffusion is an AI art generator that you can install on your home pc and create your very own art images from home. It won’t cost you anything as it’s open source and free to use.
However if your computer is outdated you might want to check the WEB UI version or try online options instead.
Use an Online AI Art Generator: If your computer won’t run a local installation of stable diffusion then you should check the online alternatives. There are 100’s out there. But if you want to try a free version of this new technology I would recommend Bing Image creator. All you need is a Microsoft account and you can try out this AI generative tool. If you want realistic models then of course head over to Midjourney as mentioned above.
The image above was created using Bing Image creator the prompt is:
“A steampunk red lush fox, in the distance, ATMOSPHERIC GOTHIC HORROR , Midjourney style, Lovecraftian | horror | unreal engine | raytracing| ray tracing, hyper details, very dark lighting, heavy shadows, detailed, ( photo realistic, realistic, dramatic, dark, sharp focus, 8k). wide angle, digital art, realistic painting, dnd, trending on artstation, soft natural lighting, portrait photography, magic photography, dramatic lighting, photorealism, excessive detailing, intricate,“
Why not create your own images and let us know in the comments who well they turned out.
There are plenty of other ways you can create AI generative art. On this list there is over 190 AI tools that can perform this for you. So you are kind of are spoilt for choice with this one. When it comes to making art with AI it is quite experimental and it may take some time before you find a style you like. So remember to play around a little and try more than one application. Here at So Artificial we run stable diffusion locally, we love that it is open source, customizable and free to use.
Can you create your own AI art generator?
Well yes, you sure can with some art generators being open sourced you can just take someone else’s code do a bit of re programming and re-release it under your own name. But why would you want to do that?
Oh you weren’t kidding you really want to? ok well it’s going to be a bit of a challenge. Ypou need to know about machine learning, programming well Python not just the basics, we mean advanced python to be specific, data and image processing. Once you know all this only then can you delve deep into the process of building your own AI generative art tool
There is other ways though, you could use API systems of other AI tools and “pretend” it’s your own. Or you could even train your very own models to use with stable diffusion. These are the best options for someone starting off in this space. But if you are still eager to jump onto creating your very own AI generative tool then we suggest you look up TensorFlow good luck!
Does AI art serve a purpose?
Well we could reply with “does art serve a purpose?” to some it might not to others it brings joy, happiness and all kinds of floods of emotions. For others it’s just something good or interesting to look at or admire. Let’s face it “art is in the eye of the beholder” this too also rings try for AI generative art.
AI art is often created to test out the potential of machine learning technology and to push it’s limits to see how far this new tech can go. By pushing these limits researchers and devlopers of this technology can learn so much more about what these systems are capable of and perhaps develop even bigger and better ideas.
Another point of view to take on the whole “what is AI generated art and what purpose does it serve” well think of it being a new way to create forms of artistic expression. Ever seen the pope on a tight rope? yeah thought not. Well here is one:
One of the biggest plus sides of AI generative art is that it gives people a brand new way of experiencing art. What’s not to love about that? more people are entering the art space everyday because of this new medium style of art. Let’s face it generative art is nothing new, people have been creating this style of art for years, it’s just humans were doing the creating now we rely on robots.
F AI art can provide us with a unique and interesting way to interact with the world around us. This can be done through interactive installations, generative pieces that change over time, or immersive experiences that blur the line between art and technology.
Is AI-generated artwork art?
This is a hot topic up for debate across the art community. Let’s face it art in generally is a bit hit and miss. Apparently someone taping a banana to a wall is considered art these days, so why not AI generative art? it’s much better to look at and at times can send a powerful message.
You may or may not heard of the popular board game Warhammer 40k, it involves miniature models fighting one another, well you roll dice and hope you kill the other team. Well you can hand paint these is this considered art? Well the golden demon award thinks it is!
The issue with the Art community is that many people assume that art has to be human, meaning that true art must involve human creativity, intention and expression. So from this point of view AI generated art would be missing that vital element, the human touch. But can’t we say the human touch is from writing the prompt? technically a human has come up with the concept and the AI is just creating something from that concept.
Personal we say AI generated art and a new form of artistic representation. Now that does not mean we feel the same way about all AI generated images. If you are aiming to clone other people work then that’s a big no from us, but to be creative and push new boundaries, well we would love to see that.
Let’s face it AI-generated art is still in it’s early days so we still have a long way to go before we see it’s full impact on the art world as we know it.
Is AI art copyrighted?
Ok this is a bloody tough questions. The main problem with AI generative art is that we are at a conflict. When we use these tools we are bound to an agreement with the developers. Some of these developers allow us to use the work freely and commercially where as others don’t, fair enough.
But the huge conflict we have is the original art owners, the ones these datasets were trained on. If these images were taken without prior permission from those artists then all images created using them images will be in breech of copyright claims.
The above is one argument to have, but let’s see that’s all fine everything has been sourced through creative commons licences and all above board. Well that’s good for one but two in order for something to be copywritten it has to be created by humans. So we need to ask ourselves is the copywrite for the person who created it i.e. their prompt or the software owners who provided this amazing piece of kit for them? confusing right?
As we have said before AI generated art is in it’s early days so this is all going to be a bit of a grey area but ultimately, as technology changes, copyright ownership for art made by AI most likely is going to be part of ongoing legal debates far into the future. We are already aware of new legal frameworks being developed to address these new forms of creative images.
AI art is a rapidly evolving field or Artificail intelligence. It has really strecthed the definitinon of art to the point it’s causing alot of stress in teh community. Some people might be skeptcial or machines creating art and replacing the jobs of artists, which is understandbale. But we have governmnets working on new legislation to help protect those postiitons.
It is clear that AI-generated art works have already had a huge impact on the art community as a whole and even the general public. There have even been instances where AI art has fooled professionals and won art competitions. So is it art? we think so. We just hope that people use this form of technology professional and don’t break any laws while using this software.
Here at So Artificial we use AI Generative art tools on a daily basis and love it, so our views might be a little biased. Let us know if you are for or against AI art in the comments below.