Chapter overview, jump to your favourite section
Hey Máté, welcome to Beyond Extent and thank you for putting together this article for us! Could you introduce yourself and let us know a little bit more about yourself and your work?
Hi! Thank you so much for the opportunity. It’s truly an honour to be invited to write for Beyond Extent.
I’m Máté Válent, currently a 2nd year environment art student at Breda University of applied sciences in the Netherlands.
As a student going into the environment art field, what got you started with video games and wanting to make art for them?
When my brother and I got a PlayStation 3 for Christmas, it came with Uncharted 2. This was my first exposure to proper AAA. I fell in love with the game. Every time I stepped into a new environment, it just amazed me (and it still does). I found myself replaying the game every once in a while, and as I grew older, a whole new level of intricacy was discovered with every playthrough, whether it was the story, the music or just the visual fidelity of the game.
At 15 I downloaded Blender, because I was bored, thought it was a free-to-play game. I soon realized that I was wrong and it turned out to be a software in which you can MAKE art like CG films! (I was more interested in ArchViz and VFX at the time) Then I started fiddling around in it and so my 3D journey began.
We’re here to talk about your latest piece of art, ‘As Favelas.’ What was the idea behind this environment? Did you have any main references or concepts to work from? How did you adapt them to your project if so?
Yes! The piece initially started out as a school project. The brief stated that we should create a building(s) from location X in the style of game Y. I found the favelas of Rio de Janeiro to be a perfect fit with the world of The Last of Us. My first 2 weeks were spent on research only and some quick scribbles on the side. The sketches were super rough, and I think if I showed it to other people, they wouldn’t understand them, but they were just good enough for me at that stage and helped me to iterate on environmental composition and come up with ideas quickly.
Research was one of my favourite parts in this project. Since Rio is quite an unreachable location, I discovered it through Google Maps. I spent days just ‘walking’ around on the bigger streets, taking screenshots. The hardest part, however, was to get into the slum complexes, especially the tiny little alleyways that they have between the stacked houses. For this, the best source material turned out to be vlogs on YouTube, showing tourists discovering the favelas. Here I learned about not just the visual references, but also the culture – the living conditions, what kind of people live there and how, what they eat, how the communities work, how religious they are. Later on all of this helps with set dressing the environment and provides a source of inspiration for storytelling.
I gathered the images into a huge PureRef board and later categorized them and cherry picked the main references. (This is my 2nd pass of the gathered reference, the screenshots that were worth keeping.)
After this I looked for the final reference. After I had some chosen ones, I knew what I need more references of.
Some of the final references. (For vegetation I gather substantially more reference, more on that later on)
1) In this image I really liked the contrast between the pink bricks and the pale turquoise paint (I adjusted the hue slightly so it blended better in the scene). This reference ended up being my tower. To make those flowers pop even more they are not only illuminated by the SSS but they are always kept contrasted, by either value or colour.
2-3) combination of concept art and found reference on the streets (Google maps).
7) The initial reference for the main building – it changed a lot during the project for mood and composition purposes.
The initial layout, composition and concept of the environment is mine, but I did adapt and took a ton of inspiration from concept art created for The Last of Us Part II.
Two of the many concept art pieces that were most influential to me:
The environmental composition of the diorama is made so the opposite sides contrast one an other. It’s made this way, because as much as a single camera composition is vital, the player would never just stand still and look at it from that exact angle; this in my opinion turns the environment into a more memorable experience.
On one side, a hole in the wall, dark and gloomy with dry grass and just a hint of orange to bring the shot together, and on the other side a bright saturated and colourful graffiti. This also adds to the story and the whole feel of it. The bright side gives hopefulness.
Your scene makes great use of modular kits- important in any artist portfolio, especially a student’s. Could you talk us through your thought process on splitting up your environment and getting the most variety possible with a relatively small kit?
After gathering the main reference, the feeling of the favelas, I started looking into the architecture, blueprints of houses, and patterns that often occur. However, the favelas are very chaotic: a lot of unique buildings always in different sizes, stacked on top of each other with some of them sticking out in corners. This turned the process of making the modular kits into a long iterative process but it was 100% worth it.
These kind of references are goldmines. They give context to how and why the buildings are built the way they are.
First I sketched out the potential geo that I saw reoccurring. It’s an overpaint of a few key references. I do this on the left for multiple houses/complexes. And choose the building blocks that I liked the most and give me the most opportunity to create variety.
I ended up making the modular pieces a bit smaller, since there is a lot of ‘whooops this ended up a bit longer’ type of walls happening in my references so I want to reuse those smaller pieces to skew the houses a little.
I highly recommend reading this pdf regarding modularity: Beyond Extent weekly advice on Modularity
Once that’s done I immediately started on the modular kit and tried to build together quickly something, to figure out what other pieces I needed, what I missed out on.
These are 2 very early blockouts, one (left) to check out how the buildings will look like in my scene and how the modularity works for that. For the second one (right) I created some of the pieces as I go, while looking at reference, and the overdraw. This blockout is just a quick verification process of the modular kit. This modular kit changed a bit during production.
The pieces were made so that edge loops followed on to the next piece (important so vertex painting won’t have jagged lines). I also moved some of the vertices out ever so slightly on the walls, so they wouldn’t be just flat walls. This works well for the favelas and other ‘sloppily’ built buildings, but maybe not for more classical ones.
The materials I’m using are seamless and no matter where the individual pieces are they will always tile perfectly without seams, since I am not using UVs but world space projection. The only liability for seams now is the vertex painting. For this reason it’s important to keep the edge flow on the modular pieces.
Not all of the buildings have the floors sticking out, but I decided it would be unwise to skip them/make pieces that don’t have them. This little in-between concrete gives some more depth to the single walls. These houses lack depth in a lot of the areas and this would help giving some self shadowing to the building. Generally, very flat and blank monotonous walls are uninteresting. This also leaves some area of error so the upper floors can be moved out slightly out of the grid to achieve that chaotic/disorderly effect.
The next issue was that the walls between the modular pieces had slight seems, but this could be solved very conveniently by adding the concrete pillars they have, also adding extra variety and self shadowing.
To add even more variety the extra brick blocks helped tremendously. I used the same shader instance on them as the one I applied to the wall behind it. These brick blocks also use the same tri-planar shader.
You have some interesting shaders being used for the buildings. Would you mind demonstrating the features of your material blending and procedural decals and how you went about implementing them?
I don’t like to be limited by technicalities. There is a lot of little things that are not in Unreal by default, so certain material functions are very useful to create. The goal for me here wasn’t to make the most optimized code, but to create it, understand it and be able to achieve art that isn’t too limited by technicality.
This document shows several shader tricks that are in ND’s engine, that I found to be worth recreating. Most games have GDC talks and other kind of documents, and they are always worth checking out when doing research.
The most useful feature of the material is that they are world aligned textures, so the modular pieces never have seams on them. The second feature is the automatic blend-between layer in the wall material.
The blend-between layer uses the same texture as the mortar texture, but with hue change and heavily intensified normal. At the very edge of the paint there is a faked AO effect applied to emphasize depth, this is multiplied with and fed into the base colour. There is no parallax occlusion mapping being used. I made the blend-between layer possible with a lot of brainstorming and help from Tom van der Veeken.
To get the transition’s shape so the vertex paint’s stretchy gradients aren’t noticeable I am using a height map of the mortar and a detail noise to break it up even further.
These are the nodes to get the blend-between layer and adjust it’s properties. They are fed back into the plaster layer as a mask.
All the decals are done within the material, using the tri-planar projection, they are blended in to the colour only plus a roughness adjustment.
The leaks are very simple, since the buildings are the same height. There is just a grunge/leak texture Lerp-ed in, and a colour parameter exposed later to change its colours where necessary. There are tiling offset parameters exposed so the leaks can be moved where necessary. The out paintable with vertex colour.
How did you approach the foliage in the environment and make sure that it was suitable for all the many different spaces and angles on which it was needed?
Early in the beginning, during research, I looked at favela complexes that are close to the rain forests and also found ones that were abandoned to get inspiration.
I knew I wanted a climber, the flame vine (trumpet creeper) native to Brazil was a great one to choose, it grows everywhere, spreads like fire and blooms all year around. These aspects of the plant were of great importance to me.
First thing of course was reference gathering. Here it’s smart to start from big to small; look at the weather, the seasons… what is native to Brazil, what plants are common in the specific part of the country (Brazil’s vegetation is quite diverse). After this choosing the right plants that are going to be made is crucial. I looked into different kinds of climbers for example that suited my needs (flowering, dense and invasive, ivy-like).
Once this was done, I found a couple of potential species and just quickly overpainted the different colours at that point to check what will work the best. It’s just basic colour blockout to see what works, it’s not intended to be concept art quality.
These steps were very important. It’s not fun to jump into production, be almost finished and find out later on that the what you’ve made simply isn’t working or doesn’t exist that way in real life and was just based on an assumption.
I’m a huge fan of close-up and detail shots, so I knew simple planes for the blooming clumps of flowers wouldn’t cut it. So for LOD 0 I am using smoothed out boxes for the individual flowers.
All vegetation models are modelled with the subd workflow, then put into the sculpting software of choice to add the more unique intricate folds, edge silhouettes, bumps and veins. The small details and bumps on the texture are put there in Painter/Designer.
The sculpts are baked down, and as many maps as possible are extracted. I recommend Marmoset for this. It’s very fast and intuitive. Another alternative is Substance Designer instead of using Painter for baking.
Once the base of the texturing is done, it’s good to test out how it will look like in engine. Making and iteration on the plant’s meshes early on is very useful.
The flowers are done in a very similar approach to the leaves. First a high poly model in Maya, then detail sculpts.
The scattering of the flowers, leaves and vines are done in engine with foliage paint.
For the grass, I followed Daniel Peres’ ArtStation breakdown post on how to approach grass.
You mention that you took inspiration from the world of the Last of Us for this project. What are your thoughts on maintaining a well-known and recognizable style such as this and adapting it to make it your own?
I think styles of games more often than not come hand in hand with the story. So it was one of my main objectives to come up with a story early on (that might not even be very obvious at first glance) around the buildings.
My first stop was the TLOU Part II Art Blast. I gathered a ton of various images but mainly concept art. I found concept art very valuable, because I got to work with similar source materials as the environment artists at Naughty dog were given and I took a lot of inspiration from them.
I noticed a lot of different elements that I ended up implementing in my own environment, such as how the vegetation grows was very much art directed. It’s like a gardener comes in every year in the midst of an apocalypse to adjust and put plants into clumps. In that sense, The Last of Us isn’t realistic. But it works. It works very well. One-to-one realism is (in most cases) boring. One other thing I noticed was how they add directionality to the grass.
The main takeaway from the vegetation set dressing of The Last of Us was that it’s never random. There is always a reason where the vines/flowers are put and where the grass grows. Here I applied the directionality and the clumping. I also tried to add contrast everywhere I could: in colour (e.g. warm grass – cold asphalt), scattering shape (red arrows) and negative space (black lines).
I also had a main in-game environment shot that I used extensively as my benchmark, especially in the beginning to have something to compare the mood, set dressing and materials too.
One of the greatest features of The Last of Us is its environmental storytelling. I tried to apply this as much as I could. With a shot of a ball, the barbed wires on the outpost and barbed obstacles on the ground that there was a military group here. Also, some extra graffiti on the side and the face on the corner. The inside of the hole of the demolished building: the kicked out door planked up in the background, the mess in the room and the old mattress.
In terms of presentation, what was your process for lighting and composing your final shots? Additionally, why did you decide to skirt the line somewhere between a smaller diorama and larger open space with the distant buildings in the background?
Early in the beginning, when I did the research, I settled on a moody, overcast scene taking place in the afternoon. Usually, I gather images from films as well for lighting reference.
Once I had a solid blockout I started fiddling around with the lighting. Almost every day I tweaked the lighting a bit, so the next day I could judge it with a fresh eye. After I had a decent result heading in the right direction, I didn’t touch it for a while. At the end, when all the props were done, camera shots set up, I did a final lighting pass, that is very intricate, and I worked on that in this case 2 weeks at least. I add main rect lights to enhance some of the angles I wanted the light to be stronger and after that some extra bounce lights in the shadow areas that I want to extenuate. I am also using raytracing because ambient occlusion is a huge player in this scene and RTX AO in Unreal is just so much superior in every way from the standard screen-space based AO.
The scene has 2 layers of lighting. One is the first layer is optimized and used to get the general feel. The second one is to make slight extra adjustments for the final shots and cinematic, based on the different camera angles. Adding those little extra lights in does make it a bit more stylized, but it really can push the shots and the scene so much further in quality and presentation.
In terms of composing my final shots, I started out with 2 main angles, that were kind of already decided early on. One of those I just ditched, but the main one stayed more or less the same. During the development I have created around 100 different CineCams in the scene. The first few are usually not the best. But eventually I tend to find better and better angles, readjust previous ones. I also start adjusting the set dressing and lighting when the potentially final angle shots are found. Towards the end I just spent days flying around the scene and looking for good angles. I tried to keep my aspect ratios the same, but if a different ratio works better, I just go for it. It was also helpful to experiment with the focal length a lot.
The video I made is made with DaVinci Resolve.
As for composing the images. I put all the camera shots I want to render out into a Sequencer and render them out as stills in the Movie Render Queue. I also put them in to DaVinci for some minor curve adjustments and later put them into Photoshop to add some grain (I add the grain in Photoshop, because I can paint out the grain in areas where the grain should be less noisy).
For the film grain I am recommending this website. They are free and very high quality with plenty of different options.
As a little extra, I also did B&W shots, with a very rudimentary graphic design and a bit coarser image grain and the with blacks pushed a bit further I think it helps a lot the overall presentation on the portfolio.
The right presentation is crucial, it not only shows care for your own artworks, but sells your environment. Lighting makes or breaks a scene and I hear it from discord servers from professionals over and over. The second very important part is the actual presentation on your portfolio. A video for environments with the right music helps so much, even if there is little to no animations going on, some nice camera movements, with slow zoom-ins helps selling the environment tremendously.
I got very valuable feedback along the way that the background feels empty. So I used the already created assets, very quickly made 4-5 building complexes, applied a dark uniform material on them and put scattered them in the fog (while looking at reference). This way that there are buildings in the background as well, it creates a sense of scale, context for the environment as well as interesting silhouettes and noise in the background while still keeping the uniformity, so the image and shots read well.
What excites you about the future of the medium and where do you see yourself in it? Tell us a little bit about your current goals and mindset finishing up your schooling and into the future?
Right now my goal is to become an environment artist and get my portfolio ready to land an internship. I am focusing on studying foliage creation for the rest of the year.
But I have a huge admiration for cinema and I am very excited that the line between film and game is starting to get blurred and that there are more and more artists trying to achieve film-like quality in real-time.