Artificial Neural Network draws sketches of tattoos - future or reality?
In recent years, tattooing, like any other sphere of modern society, has been actively growing up with technology. Some of them are aimed at providing tattoo artists with high-quality, accurate and reliable equipment, the other is a helping hand to their clients in matters of healing and subsequent tattoo care. The third group can include all modern information technology allowing you to get a maximum of information on the topic of tattoos, both for the tattooists and for their clients. Social media and advertising systems certainly belong to the third group. This is what drives this area and is the turbo of the development of this art, moving it from an underground format to a full-fledged industry.
Today I would like to talk about one very interesting trend. It has already gained enough popularity in other areas of life and is now slowly getting to the tattoo industry. We are talking about the use of artificial intelligence technology in tattooing. I understand that after dozens of apocalyptic films on the topic of AI, rather strange associations involuntarily arise, however, you should not be afraid, everything is more than safe.
A little theory for those who are not at all in the subject of artificial intelligence.
WHAT IS ARTIFICIAL INTELLIGENCE AND ARTIFICIAL NEURAL NETWORKS
Artificial intelligence (AI) is a self-learning, most often highly specialized program built on the principles of operation of a biological neural network. The very word intelligence in this phrase implies the ability of the program to perceive, process, organize and store information. AI is not at all equal to human-like consciousness, since the program does not have any emotional and sensual evaluation of the information received. In other words, it's not worth seeing evil Skynet behind every AI, an AI-based voice processing program won't be able to paint a picture, and an artistic AI won't understand your voice.
TYPES OF Learning in Artificial Neural Network
There are two basic types of learning in Artificial Neural Network (ANN): supervised and unsupervised.
The first method is based on working with an array of training examples that allow the program to understand the main parameters of the sample, comparing them with the standard. For example, let's say we need to build an ANN that can tell if a red apple is in a picture. We load 5000 red apple images and 5000 green apple images into it. We show several images and look at the response of the ANN. When a correct answer is received, it is marked as correct, and if the network is wrong, we supplement the input parameters for determining the red apple so that in the future the ANN understands the difference between red and green. And so on, until the network can accurately identify red apples in any image.
This type of learning is suitable for solving problems in which the required result is known in advance. For example, as in the example described above, when we know for sure that we need the program to find exactly the red apple in the picture.
The unsupervised approach is based on the principles of cataloging and systematizing a large array of input data without intermediate human intervention. That is, an ANN, receiving data, itself tries to highlight the common features of the data received at the input and the relationship between them, cataloging them according to these most identified features. This approach is most often used for clustering problems and statistical models. If we consider this approach from the point of view of applied problems, then the input data has great importance here. The clearer the materials are for the network, the faster and better the learning will be.
We will not delve into algorithms of learning in ANN. All the same, we are talking about art, not about mathematics. So let's move on to the application part. How does all of the above apply to tattoos?
The use of Artificial Neural Networks in tattooing
Let's talk about ANN and their artistic abilities. Again, we will talk about public networks, for communication and work with which highly professional programming skills and knowledge of mathematics are not required. These networks have undergone initial learning and are already ready to process your information with sufficient quality, which allows them to be used in work right away.
And help us to understand all this will be Evgeny Malgin, aka Evgeny Mel - a tattoo artist, one of the first to introduce the use of AI and ANN in his tattoo practice.
«My friend and I trained the first AI about 2 years ago, using the database of tattoos posted on one of the tattoo publics, plus my full portfolio was included there. In total, we have collected about 15,000 images of tattoos. Perhaps even then I could have made a tattoo according to an AI sketch, but we ran into a problem - 2 years ago people simply did not understand what it was. They thought that the AI was just mangling pictures of tattoos, not realizing that it draws everything from 0 by learning through trial and error. Here is his instagram @01tattooAi» - said Evgeny.
According to Evgeny, the use of ANN gives a tattooist a number of additional opportunities, for example, to get several alternative versions of a particular sketch in the shortest possible time, quickly work out shadows, textures, and image detail.
«AI is more of an assistant to a tattooist, not a replacement. It can't draw a masterpiece for you, but it can help refine your ideas. The more precise your request, the more successful the result of the AI will be. For example, I had a client who wanted a crow tattoo. I could sit down and draw it myself, but I decided to entrust this task to AI. The end result exceeded all my expectations. The client was satisfied!» - said Evgeny
Artificial Neural Network for tattooists
Evgeny Malgin shared his list of Artificial Neural Networks suitable for a tattoo artist:
Dall-E is the most sensational and most expensive ANN. I can't imagine how much money it needs to generate something adequate. Lots of features, Google sponsorship, popularity, but those benefits fade when you look at the price of generations. Plus there is one very strange feature. The Dall-E team are LGBT programmers, so they made some kind of joke or something .... A certain percentage of Dall-E generations simply draw kissing men. Is it worth spending money on it? Personally, I didn't. I experimented with free coins and that was enough for me.
Midjourney is the second most popular ANN. It works via Discord (not very convenient, but you can get used to it). $30 per month removes all generation limits. Many use it, but personally it seemed to me not convenient to work via Discord.
Stable Diffusion is a really cool program. But you need, firstly, a powerful computer, and secondly, you need to know the basics of programming in order to simply run it. But the availability of open source code on the Internet allows you to generate everything for free and unlimited.
Draw Things is just the same Stable Diffusion only for the phone. Very cool! You can install and play for free right from your phone. But the minus of this application is that the charge of your phone during generation will melt like ice cream in the sun in the summer. But it's free and you don't need the internet!
Nightcafe is not a very expensive browser ANN with a lot of settings. For those who love browsers it's the best, plus there is a kind of community in it, where you can share pictures, get points and generate something again for free.
Dream is an application for IOS. My favorit. The subscription is inexpensive, generates everything quickly, has a user-friendly interface, and a lot of preset styles. This is very convenient, since I develop sketches on the phone in this program, and then I finish drawing in other editors or ANN if necessary. Of the minuses: few settings, not very "smart" generation.
Artificial Neural Networks are updated every second, because with each new iteration they become “smarter”. And when they have millions of people and trillions of bytes of information at their disposal ... we think that while we were writing this article, all the ANN listed above have taken a few more giant steps towards understanding man and art, so we strongly advise you to check their work yourself. And of course, write in the comments about the results of your experiments!
Comments (1)