Midjorin Creative artificial intelligence that will revolutionize the world of art

Artificial intelligence finds new aspects in human life day by day and now it has the ability to create attractive artistic images inspired by human written instructions. A process that an artificial intelligence system called Midjourney, by making it available to people, has helped to expand it in a short period of time.

Artwork produced by artificial intelligence is quietly beginning to reshape culture. Over the past few years, the ability of machine learning systems to generate images from text messages has dramatically increased in quality, accuracy, and expressiveness.

Now, these tools are slowly leaving the research labs and reaching the hands of ordinary users; Where they create new visual languages ​​and, most likely, new kinds of problems.

It is believed that there are currently only a few dozen AI image generators. They are difficult and expensive to build and require access to millions of images used to train the system as the algorithm looks for patterns in the images and copies them. On the other hand, this work has a high computational cost, which varies depending on the application, but one million dollars seems to be a common figure in this field.

For now and as long as the output of these artificial intelligence systems is seen on the covers of magazines or used to generate memes, it is mostly considered as a new experience. But artists and designers are still integrating this software into their workflows, and before long, AI-generated and augmented art will be everywhere.

In this way, questions related to the right of ownership of the work, such as who made it and who owns it, as well as about potential risks, such as intentional output or false information generated by artificial intelligence, must be quickly resolved.

However, as the technology becomes more mainstream, companies can gain special credit by launching their products in this field. For example, a 10-person research lab called Midjourney has been designing a product that makes an image-generating artificial intelligence available through the Discord chat server.

Although the name may be unfamiliar, you’ve probably seen the output of the Midjourney system on social media. To create your own, just join the Midjourney Discord, type in a command, and the system will create an image for you.

What is MidJourney?

In fact, MidJourney can simply be considered a new text-to-image converter based on artificial intelligence that, based on the user’s text request, produces images that are strangely artistic and abstract.

The team describes itself as: “an independent research laboratory for exploring new intellectual media to expand the human imagination.”

This artificial intelligence has quickly taken over the internet, and social media users are comparing it to Dall-E Mini, which has become famous for performing the same function.

The background of this artificial intelligence

“David Holz”, who was working as a NASA and Max Planck Institute expert during his doctoral studies, suddenly left it and went to San Francisco around 2011 to start “Leap Motion” company in the field of The construction of a kind of hardware for reading hand gestures has been launched.

But according to him, in order to create something different, after about 12 years, he abandoned this activity and launched Midjourney, which currently consists of only 10 people. They work on different projects but there is a common background in all their activities: reflection of reality, imagination and harmony.

The goal of this team is not to create deep fic or even art, but to expand the imagination of the human species. According to themselves, just like engines, including in airplanes and cars, which have made transportation easier but humans have not stopped walking, they also see this artificial intelligence as an engine for imagination.

Midjourney’s position among other artificial intelligence

Many labs and companies are working on similar text-to-image technology. Among them, Google has “Imagen” and “Dall-E” belongs to “Open AI” and there are even smaller projects such as “Craiyon”.

For AI to have such power, two aspects of understanding language and the ability to create images must be combined, and this is where extraordinary things can be done.

“In the next year or two, you’ll be able to create content in real time: 30 frames per second in high definition,” Holz said. It will be expensive, but possible. Then in 10 years, you can buy an Xbox with a powerful AI processor, and all the games will be your dreams.”

Xbox has artificial intelligence with the ability to simulate dreams
Credit: The Verge/MidJourney

From a technological point of view these are facts and there is no way around it, but what about from a human point of view? There must be a way to make this artificial intelligence practical, and that’s why Midjorney has focused on socializing this technology.

According to the team, after the first tests, they soon realized that people do not request individually in a coherent structure and when writing text, they aimlessly move from one topic to another.

But in a social environment like Discord, it is possible to complement each other’s descriptions and create more practical things with group imagination.

On the other hand, this artificial intelligence has an artistic background and therefore involves human emotions.

According to Holz, if by saying a word such as “dog” it shows its picture, it is the same thing that Google search also does, but the artistic look involved in the produced works is a special feature of this artificial intelligence.

Compared to other AI trends, while Dal-e images are often more realistic, Midjourney’s focus on creating works of art is different. In other words, it can be said that Dal-E creates very clear scenes, but Midjourney offers a more creative and surprising texture.

Comparison of the image produced by Dalai (left) and Midjourney (right) with the joint command of plastic bag layout.
Credit: Fabian Stezler

Man, controller of artificial intelligence

According to Midjourney, every time you ask this AI to make a photo, it doesn’t remember or know anything else it’s ever made. It has no will, no goal, no tendency, and no ability to tell stories; Therefore, the whole ego, will and story of man is commanding to it.

Artificial intelligence security

With such potential, some are concerned that this AI could be used to generate harmful, obscene, violent, or misinformed content.

According to this team, there are such actions in all societies and naturally, it is not something they want. For this reason, Midjourney is developing solutions to deal with such uses, but it already has solutions.

First, the name of each designer is recorded on it, so they are more cautious in using it. Also, the focus of this artificial intelligence has a specific framework for which the use of unconventional words is limited.

Different from deepfake

This artificial intelligence has the ability to create images of famous faces, but due to the fact that it is based on artistic and aesthetic concepts, it is currently not possible to use it to produce deep fakes.

According to Holz, “Maybe if you try 100 hours, you can find the right combination of words to make the image look realistic, but you have to work really hard to make it look like a real photo.”

Educational data source

The training data for this AI is collected almost entirely from the internet, and like any large AI model, it gets all the data it needs for text and images from the web.

Considering that this project is still in its early stages, it has a lot of data in general, and according to the creators, this is an experimental process, and with the passage of time and considering people’s desire, it will be possible to make better use of the available data.

The view of artists

Currently, according to this team, artists who have used this project call it interesting and believe that this tool has no fear. They have stated that Midjourney is like an art student trying to create something inspired by the original artist.

On the other hand, Holz believes that artists can use this tool as a part of their artistic flow.

A group of humans whose imagination has been enhanced by artificial intelligence.
Credit: The Verge/MidJourney

Enrollment in Midjourney

If you are fascinated by Midjourney’s ability to produce abstract works and you have the ability to express suitable expressions to create a work of art, you can register in this system. Joining Midjourney’s Discord to use its AI is done using an invite code.

There are two main ways to get this invitation code. You can go directly to the MidJourney website and fill out the related form by clicking on Join the beta and wait for the invitation code, the time of which is unknown. Another way is to use an invitation code that one of the non-free users can provide you. Each user with a $10 or $30 subscription will receive 5 invitation codes, each of which can be used to invite one user.

After joining Midjourney Discord, it is possible to use 20 to 25 free commands to test the Midjourney bot, and then to get the ability to create more images, you need to pay for the basic or standard subscription.

To create images, just send your statement (Prompt) to the robot and after some time, the robot will deliver the generated image to you.

Abstract design of a Parthian king in golden armor made with Midjorney
Credit: Mohammad Rasoulipour/MidJourney

Time will tell if Midjourney can live up to the fame that Dal-E has achieved, but it’s clear that AI technology is advancing dramatically to create new and different artwork than we’ve seen before. Is.

Cover photo: Advancing human creativity with artificial intelligence
Credit: The Verge/MidJourney

Sources: The Verge, Dexerto, ViViPic