Using AI text-to-image Generation In 3D Modeling Pipeline

Good news for 3D artists: it looks like we’ll survive the AI madness for at least a couple more years.
I wish I could say the same for concept artists, though. Image generation technologies are evolving at an incredible pace. Tools like Stable Diffusion, Midjourney, Firefly, and ChatGPT have completely changed the rules of the creative process. They are immensely powerful and beneficial across nearly every aspect of art production.

Of course, AI isn’t limited to image generation—it has already integrated into nearly every moment of our daily lives. But that broader topic isn’t what I want to focus on here.

How I Use AI in My Workflow

In my professional life, I actively use ChatGPT and Stable Diffusion. These tools have largely replaced the countless hours I used to spend browsing Pinterest, collecting reference materials, and gathering inspiration.
Now, I can generate dozens of concept ideas within minutes, all tailored to the brief I receive. It has dramatically shortened the time I spend on the pre-production phase.

Below, I’ve shared a few AI-generated concept images created for a game my team is currently working on.
The brief was for a neutral NPC character—a large, friendly robot with a massive trash-container body that helps players recycle the waste they collect. I simply provided this description to the AI as a text prompt, and within minutes, I had multiple visual interpretations of the idea.

The results were surprisingly useful. The AI’s interpretations helped us quickly visualize proportions, silhouettes, and surface details that guided our modeling process. Among the concepts, we chose the first one as the foundation for the final design.

AI as a Creative Partner, Not a Replacement

That said, as an artist, I don’t use AI-generated results as they are.
I see them as starting points—a way to ignite creativity and guide visual direction. Beyond that, I prefer to build upon them manually, adapting and reshaping the designs to fit our world and artistic style.

AI-generated art still sits in a gray area when it comes to intellectual property—its legal and ethical boundaries remain unclear. That’s another reason why I treat these outputs as references rather than final assets.

For example, in the design below, I used the general shape of the AI-generated concept but made significant changes:

  • I redesigned the face to match our unified color palette and texture style.
  • I added an excavator-like hand to emphasize its functionality.
  • I introduced ladders and ventilation details on the back to highlight the scale and industrial feel of the character.

These refinements make the final result truly ours while still benefiting from the efficiency AI tools bring to the creative pipeline.

Closing Thoughts

AI is definitely changing how we create, and that’s okay. It’s fast, practical, and sometimes surprisingly inspiring — as long as we remember to stay in charge. At the end of the day, the real work still comes from us: turning those generated ideas into something that actually makes sense, looks good, and feels alive.

Tenth Year As A Game Artist

Ladies and gentlemen, a new huge milestone has been achieved in my life. While we bitterly celebrate the 100th year of foundation of the Republic of Turkey, in this August 2023 I am celebrating the 10th year of working as a game artist in the gaming industry.

I cannot really track down the actual dates but I have started making frame by frame animations with an application that I do not remember in 2000 when I was around 11 years old. Then I have started to use Macromedia Flash in around 2002. It was a huge step for me because it had tween animations and vector drawing that took my mind in that time. What I can really track down is the first model I made in 3Ds Max. It dates 8/8/2004. A car rim. It makes sense because probably I made it under the effect of Need for Speed Underground. When I was learning the app, we had no internet. Maybe we had but it was very limited. I studied it from a book my elder brother once brought, known as the book with dinosaur. I still keep that book with the blood, sweat and tears I have shed among the pages.

Book With The Dinosaur

Before starting to write this article I have checked my previous works and articles and I have seen that how my perspective has changed in the course of time. What I used to think was very shallow as it should be according to my experience in that time. The longer I have stay on gaming industry, I have discovered more of it. I feel like I used to look everything very closely which makes you see the details but unable to see the whole picture. Experience is basically increasing peripheral view of yourself in any area. Besides being a game artist I am also practicing Kendō almost as long as my professional activity. Looking at the whole picture and increasing the peripheral view is very similar to the term we use enzan no metsuke in Kendō . It means “looking somewhere far away”. This technique brings awareness by standing unfocused, alert and calm. I think what I have learned the most is the ability to look far away even when I make my art. It is really difficult to succeed if you look into details a lot.

Working in the game industry is actually surprisingly similar to playing a game. There are hundreds of descriptions about what a “game” is. These descriptions usually do not contradict with each other. Maybe only disagreement is the property about “being fun”. However my favourite and the most inclusive, clear and simple description was made by Bernard Suits as ” overcoming unncessary obstacles.” What I do is actually basically overcoming the obstacles by destroying them, fakeing them and finding a way around them. Every game comes with a puzzle and the duty of the developers or artists is untangle this puzzle with the most efficient way. So when I work on a game, I feel like I am untangling a puzzle. Every game is unique and the challenge never ends.

Video games should not be seperated from the other art disciplines. Playing a game is actually interactively watching a movie. Recently Netflix’s Black Mirror: Bendersnatch has made a direct connection between these two disciplines. Video games can be considered as audio/visual discipline that combine auditory art and visual art. Since art is a way of communication, video games deliver a message to the players. Video game artists main role is neither only making good looking models nor technical advancements but delivering the message correctly to the players. We are narrators eventually. Video game artists must be curious about surroundings and also observe a lot. They need to know literature, history, movies, fictional universes, popular culture etc. Because the message can only be delivered correctly if it is convincing enough. It will also increase the immersion in the game. Lets say in a zombie apocalypse theme, if the character has a well groomed hair and clean cloths, this would mislead the message. Even its behave, walk and the way of talk may change the message completely. I feel like studying in painting improved me a lot on creating convincing models. I think even if a game artist who create only 3d models should be able to draw as well. Drawing is great to understand the concept of anathomy, composition and color. It also helps how an object would deform in time, or how a cloth folds which is essential for creating convincing models.

There are too many dragons to slay and dungeons to be cleared ladies and gentlemen. Celebrating a decade of growth and exploration while my journey continues, I am both the traveler and the cartographer of this boundless realm of creativity.

Female Warrior

This is a new 3D model for a new idea of a game that the player manages a tavern in a medieval fantasy setting. Player gives quests to the adventurers and sends them to the journeys.
I like to finish one of each difficult task in order to see how long they will take. This way i can see the possible problems and i can find ways to overcome them before the production starts. It really helps me to plan the scope more accurately.

Viking Village

This is the village model for upcoming Viking game. This is going to be a screen that player will upgrade his army or buy something from the shop etc. Since there was a key visual that was provided to us, we have followed an art rule. Models have to be stylized but not childish. Atmosphere is dark and gloomy. I believe we achieved a nice look for the game.

The picture above is the arena. It is created for PvP fights in the game.

Viking Warrior – Character Model Armour Variations

Character models for an upcoming video game. They have no texture, rig or animation yet.

The game will be set in Vikings. After the concept research I decided to move along with the fit and agile characters like in the series “Vikings” instead of bulky body type that is used in many popular culture pieces.

Modeled in 3Ds Max, Rendered in Marmoset

After some hand painted texture work on the characters this is how they looked in Unity.