How to Make Magical Hand-Painted Crystal Sword in 3D
See how Swayam Sontakke modeled and textured crystal swords inspired by League of Legends' Katarina Du Couteau in this breakdown, where the artist showed us how he dealt with UVs and explained how he made lighting play on the crystal edges.
Introduction
Hi everyone! I’m Swayam Sontakke from India. I'm currently in my 3rd year, pursuing a degree in game art and design. I'm a 3D Environment Artist focusing on hand-painted textures. I also love game design. I have worked on a 3D horror game with a small team of 3, where I was a Game Designer and Lead Artist.
It was lockdown time when I started my 3D journey; I had just passed my 10th grade and was exploring games and the industry to pursue a career. At that time, I decided that I wanted to be in the game industry; I wanted to learn how games are made and designed.
I found a college that had a bachelor's degree in Game Art and Design and started with 2D art, and I was doing very well. I even made a small 2D illustration of League of Legends.
Ahri birthday illustration
After that, I started with 3D, exploring more techniques and art styles, and I found my goal: the hand-painted/stylized art style. From there, my actual journey as a 3D Environment Artist began. Learning 2D texturing was very beneficial for me, as my fundamentals were already clear and I knew how to render which materials; it was a really great starting point for me.
Before joining the college, I had various skills like graphic design, video editing, matte painting, and more. I built these skills by watching tons of tutorials on YouTube.
After joining college, I did not stop self-learning because even if there is faculty for you, you can't learn things unless you do self-study. Self-study and practicing consistently are very important in this field. By following these techniques, I learned my skills, and I’m still learning from YouTube, websites, etc.
Room 104 is my game project. It is a psychological horror puzzle-based game in which I worked as Game Designer and Lead Artist at my college. It was my first group project, and it went decently, not as well as expected, as it was my first big project. I learned a lot of new things about the pipeline and workflow. I am currently working on other games as a 3D Artist.
Katarina Du Couteau Sword
When I saw a knife in Valorant called Mystbloom, I got very excited and inspired; then, I decided to make a knife or sword.
At that time, I found a really nice concept of a sword made by Yudai.Nari.
It is a League of Legends sword, so after looking into it, I changed my direction to the game, which was my best decision because it took me out of my comfort zone. So, without a second thought or wasting time, I started grabbing visual references of League of Legends 3D models and their textures.
So my thoughts were clear that I would make fully hand-painted textures.
Modeling
I didn't use a dedicated reference for modelling because I already had beautiful concept art. This helped me save time in the pre-production phase.
For modeling, I chose a slightly different workflow. I started with ZBrush instead of modeling in Maya because there are a lot of curves in this sword and I wanted it to be perfect; that's why I started with ZBrush directly.
Since I had decided to do retopology for the sword, there was no point in making a base model in Maya first, which saved me a lot of time. Here's a tip: for projects where you plan to do retopology, just start with the high-poly mesh. There's no point in making a base model first, though it's very subjective and depends on the person. This workflow saved me a lot of time.
First, I made the primary shape, as there is a rule of thumb that you should always start with the big shapes and after that, focus on small shapes; this will provide clarity to work more efficiently.
So there was a very important thing: as this sword defined pride, I needed to make sure that the silhouette should be heavy, all curves should be perfect, and the shape language should tell the story.
There is a very important thing that every 3D modeler should follow, and that is the flow or rhythm of the model.
I created most of these sharp and curved shapes with a Move brush in combination with a tool called AccuCurve. To make it sharp and flatten the surface, I used hPolish. To make it a precise shape, I used the Pinch brush and ClayBuildup to fill unwanted gaps. For the ornament design, I didn't use alphas; instead, I masked out the design on the mesh, inverted it, and used the "inflate" tool in the deformation section. Also, for extruded borders, I used the same process.
These are the only basic brushes and tools I used for sculpting this:
Retopology & UVs
Topology is the most important thing for any 3D asset; good topology will lead to good optimization and good UV seams. Before starting the retopology, I took a screenshot of the high-poly model and brought it into Photoshop to mark how I was going to approach retopology in 3D space.
It's a very good practice to do because it gives an artist a clear vision of where to add loops and seams. Also, it saves time later while doing retopology.
I decided that I would not go for high polycounts, so I limited myself by creating a target of 3K tris. I wanted to make a fully optimized and game-ready asset. Although I could go up to 4-5K tris, putting myself in a very low-poly target for these hero-type assets will improve my skills in terms of optimization while achieving visual quality perfectly.
In League of Legends, the camera is set up high from the battleground, so players will not be able to see those tiny details, no matter if it's a hero asset or not. That's the reason I think the 3K poly target is pretty good in terms of optimization and visuals.
Optimization doesn't mean converting circle into triangle.
While doing UVs, there is one important thing, and that is texel density, which controls the texture resolution in-game. To achieve good texel density, you need to pack the UV shells in a way that doesn't leave huge empty spaces in the UDIM.
Also, if there are too few shells, it may cause some stretching issues, and if there are too many shells, it will affect texel density. So you need to keep a proper balance. A good rule of thumb is that there should be 2-4 seams in one mesh; this range handles most situations.
I followed these basic rules and got my UVs ready, which are packed into a 4k texture map with 33px/unit texel density. All the processes of retopology and UV unpacking were done in Maya.
Texturing
Before starting texturing, I collected multiple references for gold, crystals, blade designs, and style variations. I used Substance 3D Painter software for texturing.
I started with the gold first; I didn't think about the crystal at first because it would make the work very slow. Instead, I focused on easy objects first. Texturing gold is very easy and simple; you just need to know the proper values between the colors.
I wanted to make a rich gold, so I went with a slightly darker base color to make the gold pop.
As the base color is already a little dark, the highlights pop automatically. After adding the base color, I started adding highlights and shadows back and forth, and it was completed pretty quickly – on average, it took me 3 days.
After completing the gold, I started playing with colors because I really wanted to give color variations like Arcane; I experimented a lot. It took me another 2 days, adding and deleting those color patches on the gold; sometimes it looked good, and sometimes it messed up the gold.
So I decided to drop this idea because it did not match the gold that I created.
After this, I jumped on the crystal part, which was very challenging at first. At this point, I decided that I should drop the crystal idea and make a normal stone blade with some design.
In this design, I wanted to keep only the glowing edge, but it did not feel unique compared to other swords. There are many swords with these designs; every second sword is like this. So to make it stand out, I finally decided to make a hand-painted crystal.
First, I made a 3-color gradient using a gradient filter from light to dark, starting from the top.
I got my base starting point for crystals, but now the most challenging part was to make those triangles on the surface. In Photoshop, we are able to do it very easily due to the lasso selection tool, but in Substance 3D Painter, sadly, it's not available.
I tried to do it manually, but it wasn't working well; every experiment and iteration was wasted.
Finally, I found a Cube brush and started to paint with it. With the help of this brush, I created the basic shapes of crystals in the gradient.
After that, I added some color variations of blue, yellow, orange, and purple, then roughly added some highlights and shadows to make the form even heavier.
Quick tip: If you ever want to check whether your form has a good silhouette or not, you can check it by adding shadows and highlights roughly to your model. This technique works in 2D art also.
After that, I made dark areas darker in order to create depth in the crystal. Above it, I added more triangle shapes and used a linear dodge layer to make it shine.
The texture is almost done. I need to add only some overlay and multiply layers to create shine and depth. I also added some edge highlights and particles with a linear dodge blending mode so that the crystal feels sharp.
At first, the crystal was very challenging as I was painting a crystal for the first time, but as soon as I completed it, I felt it was really easy. It's just a matter of how you see and use the correct technique.
Lighting & Rendering
For lighting and rendering the sword, I used Unreal Engine 5.7. I chose Unreal Engine because I'm used to it; I know Unreal Engine very well, so I was very comfortable using it for rendering. Also, I use Unreal for all my projects to render them.
Before setting the lighting, I created a sphere with inverted normals, as I wanted my background to be clean. Inside that sphere, I set up my lighting.
I made a very simple lighting setup for the sword; it's a four-way light setup. First, in front of the sword for the key light; second, behind it for rim lighting; third, on the left side of the sword with a warm tone used as a fill light; and fourth, on the right with a cool temperature used as a fill light.
I like this kind of lighting setup for stylized work; it really looks good.
The lighting was very basic, and it was not popping; to fix this issue, I added a post-process volume. It's a massive tool for managing light temperature, color grading, reflections, and many more. At first, I turned on "lumen" and "lumen reflection" and tweaked the settings.
In very simple words, Lumen is a type of lighting that calculates every single bounce of light in real-time, which looks like realistic lighting. This is a very small definition of Lumen because Lumen is a massive topic to cover, so you can watch videos about Lumen on YouTube.
Note: The setting which I used for Lumen will not work in real time, as this setting needs really high GPU usage. It's good for offline rendering only.
After tweaking the Lumen settings, I adjusted the exposure and local exposure to balance the values in the scene. By controlling these values, you can get really good results.
In the end, I did some color correction, adjusted the temperature, added a vignette, and a little bit of bloom. All these settings are very subjective depending on your scene and assets. These settings may not work for your projects; this is just to show what tools I used for this project.
And finally, I took a render shot from a Movie Sequencer with a CineCamera actor. I didn't move the camera; instead, I animated the sword itself for a 360-degree view. The Movie Sequencer is used for rendering shots in a much more controlled way.
Now I just exported it with the Movie Render Queue; here, I added some settings to render properly without noise and pixelation.
- H.264 MP4
- Console variable
- Game overrides
- High resolution
And output settings are default; I changed them to 4K resolution only for taking JPGs.
Conclusion
So this was the full production pipeline and workflow; I have used it for almost every project that I created in the past.
To complete this project, it took me almost 10-12 days, working 5-6 hours daily, sometimes 8 hours but not more than that. The rendering and presentation itself took 2 days: the 1st day for setting camera angles for rendering images and videos, and rendering itself. The 2nd day was for the presentation process. The presentation was done in Photoshop.
The most challenging part of the production phase was the crystal; at first, it was frustrating, but later on, when I started making it, it really got very easy. You just need to be patient to make a good project.
It was a really good experience for me while working on this project. I learned a lot of things like time management, techniques to save time, optimizing assets while keeping visual quality, and many more. It was really fun; the most fun part was making the crystals. I really enjoyed that process; even though it's hard, it doesn't matter if you enjoy the process.
For learning hand-painted stuff, I would suggest following the ZugZug tutorials on YouTube. I watched all their videos and interviews with other artists.
It was fun to do this interview and share my process of modelling, texturing, and rendering with you all. I hope it helps you somewhere in your workflows.
I regularly share my hand-painted modelling and sculpting tips, as well as WIPs, on my social platforms; you can check them there too. And if you want to see a time-lapse for the texturing process of this sword, you can check it on YouTube.
One last thing I would love to say: “Observe, Experiment, and Apply.”
Find me on Instagram and ArtStation or send me an email.