Artificial intelligence (AI), i.e. machine intelligence, tries to simulate human intelligence through computer systems. It can be applied to many tasks, such as computer vision, speech recognition, robotics, and video games. It has been used for simple tasks in some industries, such as industrial automation and data mining (extracting useful information from large amounts of data), and can also be integrated into devices such as smartphones or tablets.
Today, many graphics and art tools are based on artificial intelligence, and industry professionals and enthusiasts make use of them for their projects. Midjourney is one of them.
What is Midjourney?
Midjourney is a technology based on artificial intelligence capable of creating new images starting from textual descriptions as well as other AI programs, such as DALL-E by OpenAI or Imagen by Google. These AI-based computer systems have been “trained” to process billions of images and texts to return an artistic result.
How do I access MidJourney?
There are two ways to access the Midjourney service: through the official web application or using the Midjourney Bot on the official Discord server. Both can be reached from the homepage of the official website, respectively by clicking on the “Sign in” or “Join the beta” button.
Thanks to the latter you will be able to access the official Midjourney Discord server where you will be able to take advantage of a free trial period which can be quantified based on the computing effort required from the Midjourney system; we are talking about around 25 uses of the program (jobs) understood as sending the command to request image creation, variations, enlargements or requests for high definition in terms of image quality.
At the end of the free-trial period, Midjourney will send the user a communication asking if they are interested in subscribing to one of the subscription plans. Here you can view the official information on Midjourney’s subscription plans.
How does Midjourney work?
How Midjourney works is very simple: you describe the scene you want to view through text in the appropriate input field (preferably in English), you send the request to Midjourney and the system returns four images in a 2×2 grid, four interpretations of artificial intelligence of what you want the machine to represent.
To request the system to create images, after logging in to the Midjourney Discord server, you need to select one of the “#newbies” channels in the list in the left column of the screen. The section dedicated to the creation of images will then appear, which looks like a common Discord chat, in which the user and other users connected at that moment can send the input text.
In the input field, at the bottom, you need to type the first command: “ /imagine “, and “imagine” in English. Immediately after typing the “/” sign, a list of commands including “image” will appear above the text field; you can select it or continue typing the entire word and press the space bar on the keyboard. In both cases, the ” prompt ” field will be created in which you can add everything you want the bot to generate.
What should the basic structure of the request look like within the prompt? After “/image prompt:” we could have a first optional part that recalls one or more images because, in addition to working on the texts, Midjourney also considers other starting images, followed by the descriptive text of the image to be created and finally the third component concerning the technical parameters.
There is no need to worry too much if this formula may be tedious, in reality, artificial intelligence can bring images to life even with simple indications: the prompt in the image below is “giant robot in London”.
Obviously, the more exhaustive you are with the prompt, the greater the chance of receiving a faithful representation of what you imagined from the system.
Various enthusiasts have thought of helping the community of Midjourney enthusiasts in drafting the perfect prompt, including the creator of an automated document that, through an intuitive screen, allows the user to produce a complete set of keywords and technical parameters without any effort. This is Prompter, a Google spreadsheet shared by author Shane McGeehan to copy to your Google account.
Another resource on the web that allows you to compile very effective prompts is Promptomania. Its peculiarity and also its most interesting aspect is the large library of technical and artistic parameters made available which, once selected, gradually and automatically updates the prompt to be inserted into Midjourney.
In addition to the possibility of indicating reference images and textual descriptions, you can in fact choose with a simple click which pictorial style of a particular artist of the past to imitate, which light and which colours to use for the image, decide the type of shot by acting on the camera and add many other details that will profoundly influence the processing result.
Once all the values have been set, the prompt is ready to be copied and used in Midjourney.
Let’s go back to the midjourney Bot commands: in addition to “/imagine”, useful for sending the image creation request, the ” /info ” command is very useful for viewing the credits or “jobs” not yet used, the ” / settings ” to set the basic parameters that will influence both the qualitative rendering of the image and the consumption of credits and the “ /help ” command which will bring up a list of useful information for deepening your knowledge of the system. The other commands may also be interesting, it is best to read the short description below each command item at least once.
After sending the input text, within one minute, the grid of four 2×2 images with the four proposals of the artificial intelligence will be visible within the Discord chat, together with those of the other connected users.
Below the 2×2 grid, you will see two rows of buttons:
U1, U2, U3, and U4, in the top row, respectively refer to the upscaling (enlargement of approximately 1024×1024 pixels + additional details) of the image above left, top right, bottom left, and bottom right.
V1, V2, V3, and V4, in the bottom row, are the buttons for creating variants similar in general style and composition to the chosen image, always following the position order as happens for “upscaling”.
How to use mid-journey in your workflow?
Learning to know the settings and commands that the AI computer system makes available and studying which elements, keywords, and parameters work best in a prompt allows the user to create almost perfect images, significantly mitigating the natural “naive” of the system. But it’s not simple. Although the images produced by artificial intelligence are surprising for the imagination in the composition and colours, for the number of details in a single illustration, for the ability to almost perfectly simulate the style of a Renaissance painter or multiple artistic styles at the same time, at present currently still require human intervention to be complete. Unfortunately, it is not uncommon to notice anatomical imperfections when there are characters, dissonant elements in some areas of the image, and similar and repeated details in many creations.
Furthermore, in the cases of creations closer to realism, it is possible to find inconsistencies in the physics of light, which translate into random effects on colours, illuminated areas, and those in shadow.
Despite this, Midjourney can be used within an artist’s workflow along at least two paths:
Midjourney’s creations as inspiration to create a new work, through any other technique;
the processing of the material created with Midjourney to create a finished project with photobashing, i.e. the technique of fusing photographs, 3D graphics elements, digital painting, and in this case also the images proposed by artificial intelligence.
Human creativity within Midjourney
Midjourney’s AI creates the image for us, so what is our artistic contribution in making the illustration?
I believe that the human contribution can be found precisely in communication with artificial intelligence: in the ability to summarise clearly and in keywords a concept that is still visually unexpressed. Only by understanding the functioning of the computer system will it be possible to communicate effectively with the artificial intelligence and, in the input phase, guide the program in creating an image.
Does Midjourney penalise artists?
On the contrary, artificial intelligence applied to art is a very powerful tool that can help the artist by proposing multiple suggestions useful for giving inspiration, adding stylistic variety to the work, and quickly providing different variations of the same concept.
Probably, in the coming years, artificial intelligence will make giant strides in all sectors, and in the artistic sector, it will produce images of increasingly higher quality in terms of stylistic rendering, complexity of details, and conceptual coherence. It is possible that in the future several people will make use of AI like Midjourney to obtain images for their projects, even those who until now would have turned to artistic professionals because they were not able to create images independently. This eventuality could certainly have some impact on the work of illustrators, designers, photographers, and the like, effectively decreasing professional opportunities, but opposing technology, and feeling threatened by it is not the way to go. The artist should in fact learn to know the technological innovations present in his time and decide whether to take advantage of them or not.
If we think about it, something similar happened with the widespread diffusion of computers and related graphics, digital painting, and 3D processing programs: some artists have decided to take advantage of the new digital tools to speed up their workflow and obtain a greater variety in their works, while others have continued and continue to use traditional tools.
So let’s ask ourselves: “Does the digital artist take away the work of the traditional one?” Yes and no, the answer depends on the many variables relating to the individual project: the request for a very specific style, the preference for one technique over another, sometimes it can simply strike a certain expressive sign, etc…
The path to pursue, for the artist, is the one that allows you to show your personality, and your artistic soul, regardless of the tools used, be they digital or traditional. Only in this way is it possible to stand out from others and from machines and therefore get noticed professionally.
Copyright issues
Who owns the images created with Midjourney? Is it possible to sell the creation made with AI? Midjourney’s terms of service specify that the images produced, while acknowledging their authorship, become public domain in the case of a free account. In the discord chat, in fact, there are images and prompts of all the users present, and therefore anyone can use someone else’s creation or prompt without being legally prosecuted, but it is not possible to use the work to earn money.
It is only in the case of paid accounts that Midjourney grants permission to use the creation for commercial purposes and the possibility of generating the images in a personal space, in private mode, away from the eyes of other users. Even in this case, however, if someone reused the same work, it would not be legally prosecutable, because United States legislation regarding copyright does not allow requesting copyright for creations generated by AI. The US Copyright Office recently rejected, once again, a copyright claim for a work of art generated by artificial intelligence, because it believes it lacks the “human authorship” needed to claim copyright.
If creation by Midjourney is used to rework through photo bashing, digital overpainting, or a traditional technique, copyright would be recognized, provided that the “post-creation intervention ” is evident, “heavy” and allows to point out a substantial difference with the original image.
In any case, if you choose to use the images generated with AI for commercial purposes, it is always important to check that there are no other copyrighted materials present in the creation; In fact, Midjourney uses an artificial intelligence system trained on very large public datasets to produce its assets and, although it is unlikely that such assets could be unintentionally similar to materials protected by copyright or trademark, this could still happen.
Conclusions
Personally, I find the collaboration between humans and artificial intelligence very fascinating, between the infinite imagination of a human being and the machine’s attempt to live up to our imagination through the processing of an immense amount of data.
Midjourney, as well as other similar artificial intelligence technologies in the artistic field, is a very powerful and useful tool for artist who can take advantage of it by speeding up and adding variety to their work. However, this does not mean that the aspiring artist can afford to be completely replaced by the automatic work of a machine thinking that he can avoid the study of the fundamental disciplines of art and without training consistently in the many artistic applications to choose from. Technologies change over time, they can improve or disappear to make room for something else and it is therefore better to have solid artistic foundations, so as not to be left off guard.
But for those who have not had the opportunity to improve their artistic skills or do not feel any need to be an art professional, for non-artistic/AI applications enthusiasts, or for non-artists, what does Midjourney offer? It offers the rewarding sensation of visualising your imagination on the screen, with a few digitised words; offers hours of entertainment thanks to its playful peculiarity “Let’s see what happens if I write like this?”. It simply offers to be a very fun experience!
