Course overview Course overview
Build the world you've dreamed of in UE4
In this course, students will create a game level from scratch in UE4. The course will cover UE4’s object placement and layout basics, foliage systems, and lighting systems. The course will help students better understand level and environment workflows, as well as how level designers, game designers, and environment artists fit into the game pipeline.
Environment design WHAT YOU’LL LEARN
The more you know, the better.
Real heroes don't wear capes they teach
Lectures by Anthony Vaccaro
Anthony Vaccaro is currently an environment artist working at Naughty Dog in Southern California. He started his career at Bungie working on Halo Reach after graduating and obtaining his bachelor degree in Video Game Art & Design. After leaving Bungie he joined the team at Naughty Dog where he has been for the last 7 years working on critically acclaimed titles such as: Uncharted 3: Drakes Deception, The Last of Us, Uncharted 4: A Thief's End and most recently, Uncharted: Lost Legacy. While at Naughty Dog Anthony has specialized in creating the largest and most open organic levels from the ground up done by the studio.
winter TERM Registration
Nov 5, 2019 - Feb 1, 2019
Companies that hire our students
Speak to an advisor
Need guidance or course recommendations? Let us help!
Show us your skills
Not sure if you have the skills, or are you proving you do? Show us.
Building an Organic Environment in UE4
Interview with Leonid Prokofiev
Hello, my name is Leonid Prokofiev, I started to learn 3D by myself at the university. First time I worked at Threedex studio where I created interiors, exteriors and sometimes assets for games. Currently, I am working as a 3D Artist at Gameloft studio in Kharkiv, Ukraine. In this company, I had a pleasure to work on several games such as Gangstar Vegas, Spiderman, Asphalt 8 and Asphalt 9.
Every 3D artist is a designer in some way should not only create good assets for the scene but also know how to harmoniously fit them into the scene in order to hook the player. The main goal which I decided to study CGMA course of Organic World Building for was to improve the skills of creating an environment: gain some knowledge in order to know the way to properly locate the objects in the world, learn how to make the environment more interesting. And of course, I wanted to study Unreal Engine.
How the Project Started
The main idea was to create a forest area somewhere far away, which would be captivating not only because of its natural beauty but will also include man-made objects. It’s also very good when you choose a place, time period and culture before you start creating your own scenes. I decided to make a place where the plane crashed (it determines the time period) and a survivor’s house in a location similar to South Dakota. I did not have time to do everything that I planned, but at least I did something.
I found a few references for my location and started creating a rough top-down map of my environment.
Then I started to block out my location in 3ds Max.
Also, I created a list of assets that I will need to create for the scene: this will help with planning the time. After that, I started learning UE. I exported some large block out meshes from 3ds Max and began creating my environment using the Landscape tool.
First, I started making rocks. I uploaded a block out mesh of cliffs into ZBrush and began to sculpt, guided by my references. Just for the sense of scale, it is very convenient to upload an additional human reference or something that will help you understand the scale. I needed to create large shapes and tried to get an interesting shape. It is not necessary to detail much since I will add detail on the normals already in the material in UE.
As for low-poly mesh for organic objects, I created it automatically using decimation. Usually, I do it in ZBrush. Then I go to 3DCoat and adjust my low-poly mesh and do UVs there. There are very convenient tools for editing and creating seams plus organic objects are unwrapped very well.
Normal maps, ambient and height maps are baked in xNormal.
Creating organic objects is not an easy task, it takes a lot of time not only to create the model itself but also to set up the material in UE.
It is necessary to constantly use the rule of PST (primary, secondary, tertiary) form, Anthony (the instructor) always spoke about this. In addition to the large rocks, it was also necessary to create smaller rocks that fell off the large ones for a good transition from rocks to the landscape. Boulders and stones were sculpted in ZBrush.
To make a rubble pile, I reduced the number of polygons of my stones, created a nanomesh brush and scattered them on the ground mesh.
This is very useful and can be used to create textures in ZBrush. For the convenience of texturing and creating material in UE, you should thesubtools of your stones and ground fill with color, so that later you do not have to paint everything manually. It’s easier to bake vertex color in xNormal.
When I create assets, I almost always work in a combination of ZBrush for sculpting and 3DCoat for retopology, unwrapping and texturing.
To created the foliage, I began with modeling high-poly grass to make a texture with good alpha and rendered everything that I modeled in 3ds Max.
In the same way, I created branches for the fir.
In order to add variety, I created several large, medium and small clumps of grass. The variation in color was made in the materials in UE.
Next, I started creating my trees. I made several branches and arranged them along the trunk. Just like with the grass, I made several versions of the tree to create a variety. In order to later adjust the swing of the branches in the material, you need to paint vertex color of the branches before the stage when you arrange the branches. The same applies to the grass.
I did not have enough free time to make a water shader from scratch, so I found a river water tool in the UE marketplace and downloaded it. In the material, I made the river a little calmer, reduced the displacement, made the foam not so intense, reduced the color saturation and added more opacity to the water.
I downloaded the textures of the ground and made them seamless in Photoshop. Then I created normal and height maps in Crazy Bump. For landscape, I used the material with LandscapeLayerBlend node. This allowed me to mix 11 textures for my landscape. As a blend texture, I used height maps, so that the materials are mixed more naturally.
Some textures such as the texture of the rocks were made in ZBrush.
In 3DCoat, I created basic textures that do not need to be tiled. For example, this is the basic texture of the airplane. I used vertex color for many materials in order to blend a few textures and make the material more interesting. Sometimes I added a slope texture in the material using the WorldAlignedBlend node, for example on fallen trees it was moss.
I also did the basic paint of vertex color if it was necessary. For example for an airplane, I made a base color for rust.
Before I started lighting, I found references and created a mood board to understand exactly what I needed.
As the sources of lighting, I had only DirectionalLight and SkyLight. Lightmass settings were not changed. I left the light source dynamic and didn’t bake lightmaps.
For skylight, I used an HDRI map, I made a sphere with inverted normals for the background and assigned the material with HDRI on it.
I also added AtmosphericFog and ExponentialHeighFog and slightly adjusted the settings. In addition to the fog, I used planes with the fog texture which I placed near the rocks and above the water. In order to add more sunlight I also used planes with sun rays. In post-processing I added bloom, lens flares, change contrast. I made the picture in warm colors, slightly changing the white balance.
After that, I added LUT in the camera settings. I found the pack with LUT presets, chose the most suitable for my references and changed its influence.
To adjust the night render, I reduced the intensity of DirectionalLight, made the HDRI map darker, removed the sun rays and changed the LUT in the camera. I made the night LUT using the standard Photoshop settings.
One of the most interesting things for me as a 3D artist was the UE Landscape system. I also found interesting the foliage tool: this tool really helps to plant your vegetation over the surface very quickly.
Of course, I do not consider UE4 as an easy set of buttons that allows doing everything right. The tool is very powerful and has impressive functionality, so one cannot learn everything at once. However, if you set a goal and get enough patience, you will succeed!
Lessons of Organic World Building
Interview with Christen Smith
Christen Smith showed how he created an amazing modular set, which allows building beautiful northern-like environments in UE4.
Looking back, I believe my last article with 80.lv was almost 2 years ago on the Dead Space Cockpit, which in hindsight was at a very high polycount, rendered in Mental Ray, and textured with Mari, being that at the time I wanted to work in the film industry. At the time of making that piece, I was a couple of months shy of graduating Gnomon, still trying to find my way in the industry. For a long time at Gnomon, I had the idea of being a VFX artist, then a generalist, to which I again transitioned from into environment art exclusively. After my contract with Naughty Dog working on Uncharted 4 with the environment art team, I think that job became the bar for me, to which I compared every subsequent job too. I worked at Lightstorm Entertainment as a Lab Artist and several other VR contract jobs, such as being lead environment artist at a really cool startup called 3D Live, working on a project called Flatline, a VR experience about well…flatlining and witnessing dying and then coming back to life (the game is available on the Oculus Store, it’s a good time), then a contract as an asset creator with The Third Floor, a previsualization studio in Los Angeles for 4 months, which was a ton of fun. Since then, I’ve found my happy place working on environment art for games.
I happened to work on a couple of Anthony’s levels doing LODs and shadow proxies during my stay at Naughty Dog and discovered CGMA after noticing that several Naughty Dog employees were teaching classes there.
The new project started during the first week of Anthony Vaccaro’s Organic World Building class with CGMA during week 1’s block in phase of the assignment, where we had to come up with reference and an asset list for what we needed to build after picking any culture we could think of to base the manmade portion of our environment on. Coincidentally enough, I had just watched a really interesting documentary on Ghengis Khan and needed a snowy terrain to add to my demo reel. I think my main goal for this particular piece was to get the most I could out of the class while wanting to revisit the scene later on after I completed it. I have this ongoing idea that I’ll make an RPG in UE4 at some point and that this snowy tundra will be one of the biomes for it.
The main goal for the modular assets was to make something that could be easily rotated, scaled, and translated in many different ways to get a lot of reuse out of that particular piece. The hero cliff can be rotated around to get a lot of variation from each of the sides, as well as any of the rocks. There’s another mid-sized boulder that can also be rotated or scaled to squeeze a lot of mileage out of it as well. Since the class was only 10 weeks, we pretty much made one or two variations of each but were highly encouraged by Anthony to make more variations after the class was over to vary it up even more. For each different asset, I drew inspiration very heavily from my reference sheets we made during the first week and tried to come up with shapes that the hero cliffs and rocks could be for the purposes of modularity, kind of like Legos. For the hero cliff, I actually used my block-in geo (also mostly modular) for the basic shape and scale that could be brought into Zbrush for the sculpt. Even after the sculpting phase, it’s a good idea to bring the piece back into Maya to duplicate it around and make sure it works as a modular piece, then I continue with decimating the sculpt back down once I’ve checked to make sure it works.
Grid and scale
The grid in Maya, when set up properly, is a must for locking in scale and proportion. As an added measure, I like to bring in UE4’s default mannequin to further check scale and proportions of assets, or other meshes to further compare scale. The key challenge is creating an asset that complements the other assets in that they work in harmony with each other while thinking about primary, secondary, and tertiary roads. For one, the asset has to add purpose to the scene. For another, an asset can look amazing on its own but doesn’t flow well with other assets, as was something I learned the hard way when sculpting multiple iterations of the hero cliff, an asset that takes quite a bit of time to create. The 2 cliffs looked great on their own, but had very different directionality, so when placed side by side, they looked terrible. Adding to the gameplay is the most crucial part of making a pretty environment as well. If the asset detracts from or hinders the gameplay in any way, then it isn’t working and should be quickly removed. Establishing a good block-in with the gameplay as a top priority is what the assets are based on.
I’m in the process of teaching myself Substance Designer, so most of my textures are made through trial and error and skimming through tutorials, or through sculpting them in Zbrush and basing the rest of the material from the height map created there. As for the rock material for the bigger hero cliffs and boulders, I really wanted to get those deep cracks found in sheer cliffs, and am not great yet at making convincing ones in Designer, so I ended up sculpting those in Zbrush. To save on texel density, it’s a good idea to make even your materials modular, or getting as much reuse from them as possible rather than having one to one textures for every single asset in your scene. The cliffs, boulders, and small rocks all share the same material, as a result. Obviously, this also saves a lot of time, as you aren’t sculpting out a unique material for everything.
In games, it’s all about doing things the best way possible, but also the cheapest and fastest way as well. To get the snow on top, I used the World Aligned Blend node in UE4 and made scalar parameters to be able to edit these values on the fly. I think the biggest challenge, for me at least, is making a tileable texture where the viewer can’t tell that it’s a tileable texture. You really have to watch out for unique bumps or details in your texture that the eye can easily spot and see it being repeated over a large area, so finding the right balance of catchy but not standing out is the key to making the repetition subtle.
The dirt material was made in Substance Designer, mainly to add some nice break-up to the white of the snow.
The snow shader was made entirely in Substance Designer, which was a challenge since snow inherently doesn’t have a ton of detail to it, plus it’s white. The challenge, therefore, is in making a snow material that’s interesting, so I wanted to make the snow look as though it had been pushed along by the wind, making small dunes. A variation of the soft snow was created afterward to break things up. As a final detail, I added shinier snowflakes using a roughness map that sparkle when perpendicular light glances over it. I then created a landscape auto material with transitions and hiding tiling textures in mind. When the landscape material is a certain distance from the camera, a distance fade is applied to lerp into a “Far” texture with bigger tiling, which masks the smaller tiling pattern.
I think the main thing that sold the look of the landscape was in varying things up and having good soft transitions between the different materials that taper into the others, rather than harsh transitions that didn’t make sense or feel natural. I think the most important thing though is constantly referring back to your reference, which I used to be really bad at during my time at Gnomon. You really want to have that look and feel locked down early on and to stay true to it, and your reference has all that information in front of you. I think I mentioned this last time with my Dead Space article, but “an artist is only as good as his or her reference.”
I think the two biggest things that I took away from this class, on a conceptual level at least, were the value of your PST’s, or primary, secondary, and tertiary shapes, and the importance of good transitions. Your primary shapes are the biggest and first reads in any composition. The secondary shapes are the slightly smaller ones, where the tertiary shapes can be thought of as the little details. These also create nice blending and transitioning in your scene rather than harsh ones that can break immersion and really take you out of it. I also think this was Anthony’s overarching theme throughout the entire course, and he did a fantastic job of reinforcing the idea time and again. I learned a great deal in this class, as the teaching and delivery of it was outstanding, but if I had to take away anything from the class, it would be those two concepts.
Approaching Organic Environment Building
Interview with Casper Wermuth
My name is Casper Wermuth, and I currently work as a Lead Environment Artist at Ubisoft Blue Byte in Germany.
I was born in Denmark and took a Bachelor of Arts at The Animation Workshop, which is directed towards the film industry, animation, and VFX.
After working with VFX in commercials for a year, I got a small job in Copenhagen doing real-time dinosaurs for a museum which wanted an interactive experience for their visitors. That lead me to games, and I moved to Japan working for Shapefarm and Valhalla games on a title called Devil’s Third. I was sitting at a group of desks where everyone was a much better artist than me, so it was a great learning experience. It also opened my eyes for how interesting environment art actually is. After we shipped, I got hired at Ubisoft Blue Byte, where I helped wrap up Champions of Anteria and have now been here for 3½ years, currently working on an unannounced title.
The first time I did real-time environments was working as a freelancer for Shapefarm.
I was paid a flat rate per package, and I would get the money when the asset was approved.
My skill level was far below anyone else on the team, so I was working really hard, just to figure out that my approach had been incorrect and not optimized enough from a technical standpoint, and then having to redo everything. My hourly rate was therefore not exactly impressive on the first few deliveries, but I quickly caught up, and the overall experience just taught me a lot.
Making environments is interesting for several reasons. One of them is the variety of tasks involved in creating one. I rarely feel it’s a grind. Cliffs, boulders, grass, trees, terrain height, and terrain textures. It can of course also be indoor environments and sci-fi hallways, but I haven’t done those in a while.
It’s fun to work on so many things, although at times a bit frustrating, as I have a feeling I will never get good at any of it.
Another fun thing is bringing a certain mood or feeling to a scene, and trying to reflect that in each asset. Their shape language, colors etc. Something as simple as a grass asset will look different, depending on the general setting of the environment. Is it a happy place? A scary place? Does the player feel safe here, or is it a place where every living thing could potentially kill you? A wooden plank also looks very different depending on who made it. How new is it, how was it used? Even when the entire scene is made, some quick changes in lighting and post-processing can also dramatically change the feel of the entire scene.
All of this combined is why is find environments so interesting. There are so many small things that all come together to create a place, a mood, a setting, which may or may not exist in the real world.
This specific environment is very generic, and probably conveys minimal emotion, but it’s something I think could have pushed it to the next level, and something I am trying to become better at in my work in general.
I’m full-time employed as a Lead, which means I spent the majority of my time writing emails, instant messages, outsource managing, in meetings or talking to people. There can be several days in a row where I create zero content for the game, or where I only get to open up the engine those 45 mins in the morning before the others come in, and the day starts.
The goal of this course was to brush up on some skills and make sure I stay sharp, at least in terms of knowledge. I knew the 10-week course wouldn’t be enough time for me to deliver a stunning art piece, but I reached the main goal of updating my knowledge, and refreshing some things I used to know, but had forgotten.
The approach to this course was, therefore, to follow Anthony’s course completely, being a newb for 10 weeks, and asking as many stupid questions as possible, and failing whenever possible. The great thing about these courses and personal work, in general, is the lack of consequences. I can be stupid, trying out things I have no clue about, and generally fail all I want, without endangering a deadline or a project.
There was initially a setup of 6 cameras, also looking in the other directions, but it turned out I didn’t have time to do enough level art to make them work. Towards the end of the course I therefore only worked on these 3 very identical shots, focusing on the same focal.
In hindsight, I think this was a mistake. I tunneled myself into something more of a single shot, instead of an actual environment that would be interesting to talk around in.
During blocking out of the scene, I tried making a lot of leading lines towards the focal.
I overdid it, and the scene became extremely cluttered and impossible to read. I, therefore, had to tone back, deleting a lot of assets, to free up some space both for the eye to rest, but also for the scene to even read in the first place.
First cliffs in the engine. As this was the first completed asset, I ended up spamming it everywhere, completely overcrowding the scene:
A great example of how NOT to spam rocks:
Trying to expand a bit on the scene, to give some more breathing room. However, I didn’t go far enough in this direction, and the scene was still too tight:
The final composition where there is more room to breathe, but the lines are still directing the viewer to the focal point:
When lighting the scene, I used Light Functions to create the illusion of light passing through clouds, thus lighting the environment unevenly. It’s an easy and fast way to tone down all the irrelevant areas of the environment and highlight the areas I want people to be looking at. As an afterthought, I tossed this screenshot into Photoshop, and by using a Threshold adjustment layer, I can check how the values hold up.
As you can see, the path is standing out between the darker grass and cliffs, and there is a hint of readability on the focal. It’s not exactly a perfectly readable composition, so I could definitely have improved in this area. Ideally, the focal point and the path leading to it would be readable like this, and the background mountain would be less of a mess. The focal could also have had more of a dark ring (trees?) around it, to better contrast to the light background.
The last thing I did for composition was the colors. During some early iterations, I had much more trees in the scene, but they were too dominant with their bright red colors. I then removed them all, except the ones just around the focal, but that felt too artificial.
Eventually, I put in a few trees throughout the scene here and there but gave them another material instance with a Desaturation parameter. This enabled me to tone down all the background trees to a more muted color, keeping the saturation for the trees around the focal
The terrain is simply Unreal’s basic terrain, with no bells or whistles. Using the basic sculpt tools with high strength and big brush radius, you can quickly lay in the base topology, and then just smooth it out to get something less “hand-sculpted”. All interesting shapes come from the 1 main rock I have, that are simply poking out from the terrain.
If the look was more realistic, I would probably have looked into a procedural tool like World Creator or the like.
The scene has the main rock, a boulder set, and a rubble set.
The main rock is sculpted in Zbrush. I wanted to keep it relatively flat on top, so I could use it for directional leading lines in the scene. It’s made using standard sculpting techniques and alphas, and then decimated down to a bit below 10k triangles and brought into Unreal with just the normal map.
I then set up a shader that would lerp between 2 tileable textures using an RGB mask. Obviously, it could have used more than 2 with such a setup, but I didn’t have time for creating them. I created one with structure, and one which is smooth, to try and bring some more interest to the surface.
I then exported a random mask from substance, which was actually just to serve as a quick test to see if the shader worked, but it held up fine enough, so I never got back and made a better one.
Now the rock would have a unique normal map, and the 2 tileable texture overlaid on top of it.
The last thing was then to add a WorldAlignedBlend to have some dirt or grass on top of the asset, independent of the rotation. This controls the alpha of the last lerp, where I simply input the same grass texture as I used for the terrain. The vertex color is for masking out this blend, in case I don’t want grass everywhere.
Problems with the scale
This is completely my own fault, and perhaps one of those funny fails I am learning something from.
Anthony kept reminding us from day 1 that the scene should be kept small and manageable, as 10 weeks of spare-time work is very little to create a big environment.
The scene in blockout stage was therefore relatively contained. However, I wanted to have more depth in the scene and started expanding it, but because all the original measurements were quite small, it never felt big. It did and still does, feel more like a smaller whimsical world than a big valley somewhere. Anthony was talking about some scenarios where he would scale down background trees to fake a sense of scale, so I thought I would use that, and cut down the trees to 60% of their original size, except for the 1 tree in the foreground.
My logic was that this would establish trees as being big near the camera, and therefore seeing the smaller trees in the scene would create a sense of scale. It didn’t quite work out, and the scene, therefore, has a slightly awkward feel to it. I always intended to make the scene stylized, so I decided it wasn’t such a big deal, and I didn’t have time to rework anything anyway.
I, therefore, kept it in and hoped it would all be fine and go unnoticed.
..It didn’t, lesson learned
First block in Max:
The vegetation is actually very simple, so I’m glad to hear it worked for you. It’s some basic leaves, stems and flowers set up in 3ds Max and Zbrush, and then baked down on a single sheet. I then cut them from the sheet and arranged the assets from that.
The texture setup is also very basic, as these details are so small on screen, that it’s more about how it works in a mass, rather than how each asset looks up close. Each of these groups is therefore just a few different flat colors with a random mask, to create some color variation. To get a bit more large-scale color variation, the shader has an Absolute World Position driving the UVs of a mask, which I used to tint everything to a slightly different hue, just to get some soft large-scale variation.
The grass and tree canopy is done in exactly the same way, except they each have their own sheet.
Because of the time constraint, all of the graphs are very simple and unsexy. I did get time to come back and iterate a bit on some of them, but most of them are just bashed together in the most simple way possible and left like that.
As an example, the pebble texture I use around the rocks, to visually blend them into the terrain a bit:
A shape which is warped a bit, then getting some planes from a gradient
I duplicated this part 5 times, and just changed the random seed on the perlin, to have them warp differently, and then plugged them all into 3 Tile Sampler nodes.
1 for large shapes.
1 for medium.
1 for small.
The large Tile Sampler node got a few Slope Blurs to break up the shape a bit, and then I blended those 3 together using height blend.
This is literally the entire diffuse. A random grunge into a gradient map to get some overall color variation, and then a few Uniform colors to get some more deliberate colors in specific areas. Using a histogram select to get the lower crevices, and some of the default dirt masks for further variation. That’s it!
The other textures, for example, the base dirt\gravel texture is done in exactly the same way. The base shape just has less of a bevel, and therefore becomes flatter and trampled looking.
Again, the same for the grass put just putting in a splatter circular to get more of a grass clump look.
Once I had the pebble, the grass, and the dirt, the other textures are just height blend between those. So when I needed to transition from grass to dirt, I wanted a texture with dirt and just a little grass.
I then simply blended between these 2 graphs, using a mask to cut away from grass, and an HSL node to tint the grass in a slightly different color.
I’m aware these are not exactly mindblowing graphs, but perhaps it can help someone out there to understand that even basic simple graphs can still go relatively far when the focus is a scene in its entirety, and not the quality of each texture. I’m also a big fan of iteration and generally work like this anyway. Getting something quick and dirty into the engine, and then iterating on it until it looks good.
I’m very new to substance, but I still chose this tool over Zbrush, as that would enable me to go back and iterate on the textures later if time allowed. I didn’t get more time closer to the end of the project and ended up just closing the project as it was.
Advice from Anthony Vaccaro
Anthony was very thorough and covered everything anyone would need to know to make an environment like this. During the feedback sessions each week, he offered advice on many things and gave special attention to the topics each of us preferred to dive more into.
Something we talked a lot about is how the primary, secondary and tertiary reads are, and transitioning nicely between assets and between different areas of the level, using that principle. If I have to be very specific, then the importance of PST and transitions is what I would say is the biggest takeaway I have from the course.
However, if we are talking about something less specific, I think the most important lesson I’ve (re-)learned from the course was, that there is no secret technique, no hidden magic button, to get a good end result. It’s all about composition, color, design, leading the player. It’s all about art.
I was yet another student back in school, who was neglecting this, and I focused more on the button-pressing than the actual art. During my career, I’ve learned several times that the technique you use is not the deciding factor for creating an emotional response in the audience. It was good to see that even the workflows behind Uncharted 4, which is one of my favorite games in terms of visuals, is not doing anything special or anything that I am not aware of. They are simply great artists and spend time iterating until they get it right.
It, of course, depends a lot on the game it would have to be used for, and how much resources would go to characters, weapons, effects, sound, UI etc etc.
If this scene had to be production ready from a technical standpoint in a more generic way, I would need to create some more custom LODs for the vegetation, and have them cull much quicker. I also used some fairly large lightmap resolutions on trees, and would probably have to reduce the lightmap size to half, and actually do some proper lightmap UVs, as they are now all automatically generated.
But again, it would very much depend on what game it would be for, what the focus would be on.
Setting Up a Tiger's Nest in UE4
Interview with Yulu Xue
Check out a little writeup from Yulu Xue, where he shared some of the techniques he learned during the recent CGMA environment production course.
My name is Yulu Xue. I’m a 3D artist living in Toronto, Canada. I’ve been worked on various types of games on multiple platforms in the last couple of years. Titles I worked on including Fragmented, Snowboard Party 1 & 2 and BBC Top Gear: Drift Legends.
Before I took the course, I mostly worked on hard surface models and city scenes. I really want to step out of my comfort zone and expand my skill set.
Special thanks to my wife who paid for the course as a birthday gift to me. Best gift ever!
The project started from gathering a bunch of references and broke them up with the PST method Anthony Vaccaro introduced during the first week of the course.
I made a blockout scene in Maya with some very simple geometry, based on a very simple (nasty) 2D map I drew. With the help of the blockout scene, I had a rough idea of how big my scene would be and how many assets might need to fill it up. An asset list was made to help to track the time used on each asset and the progress of the project.
Modularity is king. It applies to the organic world as well. Given the time we had for the course, I made 1 hero rock, couple smaller boulders and set up rubble piles. The models were done in ZBrush and brought to Maya and Substance Painter for UVing and texturing.
I worked on the “land” space after placing the blockout meshes into the UE4 scene. The landscape system in UE4 is powerful and easy to use. Anyone who has experiences with sculpting software solutions will get used to the tools within UE4 very quickly.
I tried to keep the scale of the scene relatively small. To get the scale right, I used a human model as my scale reference when I make my blockout scene. Therefore, I didn’t need to worry about the scale in UE4 too much later on.
I used a plugin for the water called River Water Tool with Flow Buoyancy by James Stone, since crafting a running water material could take me days (months).
I created all my terrain textures in Substance Designer. With the power of Designer, I can adjust my textures and create variations very quickly. Within UE4, the powerful landscape material system helped me blending all the materials I made in Designer. I used heightmap to control the blending with the Landscape Layer Blend node. To make a cleaner graph, I created material functions for each layer.
I also used vertex paint to blend materials on static meshes, so that I can have 3 different materials on one mesh. The alpha channel of the vertex paint was used to adjust darkness and roughness level of the surface. In this way, I can paint the wetness area of the surface base on the placement of the mesh.
I used only a skylight and a directional light in the scene. When I was constructing the scene, I set my main directional light at a neutral value of 3.1415 and kept everything else by default, based on this article by Alireza khajehali.
After I finished populating the environment, I started playing with the lights. I ended up with a sunset light condition that created contrast on the rocks and a nice reflection on the water surface. I also adjusted skylight intensity to brighten up the shaded areas a bit.
I love the feeling of pushing myself out of my comfort zone and learn stuff. This course contains stuff that I wanted to learn all the way from the conceptual to the technical level. Luckily, Anthony gave us a lot of useful pieces of advice that helped to overcome these challenges. By dividing references and concepts into PST, primary, secondary and tertiary, I got a better idea of the scene that I was making. I should also point out that the weekly Q&A sessions were extremely useful to iron out the questions students had during the course.