The generation of content using artificial intelligence has evolved a lot in recent years, with people loving to put machines to “imagine” unusual situations. Because it was using one of these tools, Midjourney, that a guy managed to do something very interesting: create an experimental game.
Known on Twitter as Nao_u, in a long sequence of publications the game designer explained how he developed the Shoona horizontal progression shoot ’em up in which the visual part was created with the help of artificial intelligence.
It can be tried for free here, Shoon works like a typical game of the genre, in which we’ll have to dodge a lot of shots and destroy the enemies that advance towards us. Made in the Unity engine, the title has a very simple mechanics, with our shots being automatic and the player only having to move the spacecraft.
However, the main highlight of Shoon it’s really in your visual part. From the logo that appears on the opening screen to the ships, going through the destroyed city that appears in the background, everything was generated by Midjourney and in the video below we can see the result.
As the tool generates images randomly according to the terms entered by the user, Nao_u says that in the end it all comes down to a stroke of luck. For him, it’s like we’re using one of those machines that release balls with a toy inside.
The positive side of an artificial intelligence like this is that variations of the same image can be created with the press of a button, but patience will be essential to find exactly what we want.
In the thread published by the author, he showed some of the variations he found, leaving him “only” with the task of filtering what could be used and everything that would be discarded. He also reported some difficulties he encountered, such as not being able to implement a variation on the scenario due to lack of time.
メカ 系 の は 「Starwards X-Wing SpaceFighter Flying From Left to Right in Black Space, Armored Core For Answer, High Quality」 の ガチャ 回し て たら 「背景 黒 で 横向き」 」」 」」 」」 」」 」」 」」 」」 」」 」絵がすべての祖先になってる (赤丸のやつ)
— Nao_u (@Nao_u_) August 14, 2022
The use of artificial intelligence in game creation is nothing new. Being able to help improve image quality in older titles, generate stages and balance the level of difficulty, several companies have ventured into this area to make their projects more attractive.
However, what Nao_u achieved with this shoot ’em up goes to show just how much game development could change in the future. If today some people are afraid to start creating a game because they don’t have artistic skills, something like Midjourney, DALL-E or Google Imagen can serve as strong allies.
For some, letting an artificial intelligence create the visual part of a game can be considered cheating, the product of a synthetic creativity that could never be treated as art. I understand such an argument, but as long as the result is something beautiful, does it really matter whether it was born from the mind of a man or the bits of a machine?
Let’s take as an example the images that illustrate this article. All of them were created by me, or rather, by Midjourney under my request and I confess that I felt a certain pride in how they turned out. Yes, I recognize that it is strange to consider myself the author of something that was born after the roll of the dice, having my influence only in the directions I gave.
When using the tool, I wanted to see what I would achieve if it suggested a war between spaceships on a post-apocalyptic Earth and that’s exactly what artificial intelligence delivered to me. Even though the illustrations aren’t much use as backgrounds for a game, they could be a great starting point, beautiful concept art that could serve as a reference for competent artists.
How to generate images using Midjourney
What you might be asking yourself now is: how do I play with Midjourney? The answer to that couldn’t be simpler. The first thing you will need is a Discord account, as the tool works there. Then, just visit the project website and click the Join Beta button.
Once inside the Midjourney server, you need to know that you will only have access to a trial period, where you can generate a limited number of entries (around 25). If you liked the tool, you can purchase a license, which on the cheapest plan costs $10 per month.
After that clarification, just head to one of the channels for newbies located in the sidebar. With names like “newbies-40”, that’s where you, I mean, artificial intelligence will make the magic happen. Go to the text box located at the bottom of Discord and type the command /imagine followed by the words you want to create the arts, apart from enter and wait a few seconds.
The program’s bot will start generating the images, with them gradually gaining details and when the work is done, you will see some letters and numbers just below. They mean the following:
- U (upscale) – creates a more detailed version of the image.
- V (variation) – creates an image based on the one you selected.
- Refresh – generates four new images based on the same words entered.
As for the numbers present in each option, they represent the four generated images, with the first being the one on the left in the top line and the fourth being the one on the right in the bottom line.
If you choose to increase the resolution of an image, a new creation process will start and at the end you can download it to your computer. Finally, if you prefer the generated images to have a specific measurement, just add the command –ar 1280:720 (change the numbers to the desired measurement) at the end of the text line.
On the Midjourney website it is possible to see a gallery with several images generated by users.