Mechahearts Windows, Mac, Linux game

So, Mechahearts is a visual novel in steampunk setting. All imagery for this game was created by A.I.: I used Midjourney with a little help of Dall-e 2.
I was very skeptical about neural networks: they are still unable to produce a number of independent drawings connected with one style that I want and technique that I want. A.I. inspires primarily with pictures many pictures that scattered over the network. Let’s take a look at one of them:

Actually, this picture looks awesome, but if you look at the “prompt” that was used to create this, it becomes clear that A.I. didn’t really adhere to the requirements:

Author of this prompt several times specified that the girl should be depicted in “full body” and that she should have cat ears and yellow hai. But all in all, Midjourney gave him an abstract image ignoring some of the words he used in his prompt. This image is still amazing, but it does not meet the requirements of the "customer".

This “feature” of the A.I. makes the work with it very difficult in these cases, when you need to generate a bunch of characters in the same style but with different characteristics. The idea to create a crowd of heroes and stand out among thousands of other visual novels with them repaid immediately: firstly, it is very difficult to get exactly the desired picture through an A.I. Even if I have thouthands of words in my prompt I still have to rely on the luck. Secondly, A.I. also “rewards” all the characters with well-known graphical artifacts: I’m speaking about eyes. Just look at this:

I was happy with this portrait, but I wasn’t able to get beautiful eyes. My goal was to receive a nice-looking result. I had no interest in leaving these artifacts and glitches. People can “forgive” this mess to A.I. in the way we forgive mess in children’s drawings like it always happens: “well, this green daub doesn’t look like anything in this world, but my son draw this, so I will put it on the wall.”

In the same way, many people admire the “paintings” of A.I.: they turning a blind eye to the well-known artifacts like they do with a children’s drawings. Before I realized this simple truth, I sincerely admired this mockup I made in the first day of using Midjourney:

I really thought that this picture looked awesome. Now I understand that this crooked daub doesn’t have any artistic value. It doesn’t matter who drew it - a person, A.I. or a cat. This mockup looks bad.
I did’t want the graphics in the game to have value only for the reason that everything was drawn by A.I. I began to edit it. I understood all in all, the final user won’t care who drew this picture. It is more important how it looks.

This is an example of the most complex retouching I've done on a character. Most of the other characters haven't gone through the same amount of changes. But all of them “visited” Photoshop at least once.

Character’s emotions were another problem: any character in visual novel should have several facial expressions. But any attempts to create several expressions on the face of a character that was already drawn led me to the absurd results:

This screaming face is great, but it has nothing in common with the original character. What did it mean for me? It mean that all in all I should draw by myself again.

During this entire experiment with Mechahearts, I came across another almost impossible task: drawing a character from a different angle. When I needed to portray some of the characters “from behind”, I spent a lot of time on a boring and stupid things: I just spammed with the same request over on over again. All in all I get an image good enough to work with:

A.I. are very inaccurate in details, and as a result, the sleeves, cap and coat hanger disappeared. I had to do it all over again, or finish it, or make collages from other results:

All in all I made it, but this drawing is still much worse than the one that a real artist could create.
But the restrictions imposed by the A.I. doesn’t end there either, and sometimes impenetrable madness of A.I. forces me to give up and simply not even try to generate the scene that I need in the story. It will be easier for me to rewrite the plot.

I needed a beautiful red-haired girl lying on the floor. Along the way, the idea had to be slightly redone, because A.I. have even worse problems with complex poses than with eyes or hands. A complex pose without a broken anatomy can only be obtained by chance.

Working with A.I. is reminiscent of quick blind shooting - it's fun and I was littered with pictures, but 99% of them were rubbish that simply cannot be used in my current project. I used to develop games by myself, not limiting my ideas to other people's capabilities, but now I have to adapt to the peculiar skills of the A.I.. I'm editing the plot because of it!

So the question is: does the use of a neural network make life easier for a developer? It all depends on developer’s own wishes and skills. Without my drawing skills all there characters in Mechahearts would be creepy cross-eyed freaks with three extra fingers. However, I’m not too good in drawing in style like this – art in Mechahearts is too realistic for me. Here is my style:

On the other hand, adjusting a portrait or background is still much faster than painting it from scratch, so I can devote more time to the creating of the story for my game. Of course, I could choose a style that is more similar to mine: with similar style adjusting the drawings would not be so difficult.
But in that case I could lose any reason to use A.I.: what’s the point in using it if I can draw everything by myself in the same style? The idea was to get a result that is as far from my style as possible.

Does A.I. help in technical part of the game development or not is a debatable question. My opinion is that we need to think about the moral part of the question because without the proper fuse and love for our projects, no matter how technically cool and advanced they are, it is very difficult to complete your game project.

Development should be fun: only in that way our brains will perceive the work on the project as a “reward”. Otherwise, our brains will always try to force us to do something else, and creating games will become painful work.

When I draw, even if I do it in vain, I feel like I'm growing as an artist. Without those hundreds of drawings that I crumpled and threw away, I would never earn the skills I wanted and came up with the three games I draw in solo. That feeling of “evolution” brings tons of joy: my brains know that I have to change and improve in order to receive what I want, and it rewards me with dopamine when I draw.

In the case of working with a neural network, I do not develop as an artist at all. The wording “prompt-artist” is, for my opinion, a real nonsense. This word was created to give people an illusion that they gaining skills and developing as an artists while they imagining prompts and repeat the same requests for 50 times.

But this neural network is learning and developing, not a prompt-artist.

The selection of a prompts is a very boring and illogical puzzle, the solution of which does not make anyone better or smarter: it solves a one-time task. Generating pictures is easy at some point, but mentally exhausting for me, because the feeling that I'm wasting my life and not gaining skills is very vile. Probably, for this reason, I edit all these pictures with such pleasure, because that's where creativity begins.

So it’s time for me to finish this short article. I hope you will find it interesting and helpful and will take a look on Mechahearts. You can learn more about the game on game page!