Breaking Down Repetition in Epic 3D Spaces

Desmond Man shared some techniques, which helped him to build the stunning Atlantis environment for the Artstation’s Ancient Civilizations challenge.

Desmond Man shared some techniques, which helped him to build the stunning Atlantis environment, which was created for the Artstation’s Ancient Civilizations challenge. Nice overview of the way you can use Substance Designer and Substance Painter.

Introduction

Salutations 80.lv! My name is Desmond Man, I’m 27 and I was born and raised in London. I studied at the University of Hertfordshire and upon graduation, I was employed by Eurocom Developments where I worked on 007 Legends as a Junior Artist. Unfortunately, Eurocom was forced to close its doors and afterward I ventured out into Hong Kong for 3-4 years where I jumped around the creative industry working on anything that utilized my 3D skill set such as Interior, ArchViz, Product, Fashion, Exhibition setup, Advertisement, Film and Games.

After Hong Kong I returned back to London and joined Sumo Digital as an Environment Artist where I am currently still at today. Unfortunately, under several NDA contracts I can’t disclose what projects I worked on in HK & currently in the UK since they are all still in development or unannounced.

1 of 3
1 of 4

I first began taking environment art seriously back in 2015, not to say I hadn’t been doing 3D art but I wasn’t dedicated to becoming a great or better environment artist until then.

I picked up ‘The Last of Us’ and I fell deeply in love with the visual aesthetics of that world, it really helped me understand that I had been wasting a lot of time not building myself to be a better artist and too much time playing video games and never picking up the pen, as ironic as that is.

Around that time I was working for a company in Hong Kong that only used flat lit hand-painted models for their game and I was forced to adapt and learn.

However learning hand-painted texturing opened my eyes to the world of concept art and I began nurturing myself to learn digital painting. If I could pin point what single aspect defined my career turning point it would honestly be when I began trying to learn to be a concept artist. I can not stress enough how useful it has been to me as an environment artist to have taken that amount of time to learn to paint and to learn the fundamentals of what a concept artist has to think about when producing their work, once you apply the same mentality to environment art you open a whole new world of artistic theory & storytelling.

Since then and until now, I’m very much in love with what I do and this industry. I only plan on getting better, so to the artists around me I hope we climb far up this ladder together.

Now that I’ve introduced myself let me begin my breakdown of how I tackled this challenge.

Beginning the Project

This project started off with a very tactical breakdown on what existed in the concept by Leon Tukker. Essentially I start out with a visual analysis, you point out what in this scene needs to be built and you strip the concept materials down to it’s foundations. From that you end up with a spreadsheet that essentially tells you what parts are bespoke and what can be done in modular.

Afterwards all that’s left to think about is appropriate application, you define your methods and pipelines to obtain the result you want while making sure there’s a backup solution in case the initial workflow fails.

Once the method/pipeline has been decided, I divide up my budget on time.

This environment was huge, I knew that from the beginning and I really had to sit down and figure out how to tackle a structure as huge as the tower and all the elements that resided inside of it.

The biggest obstacle in this environment actually wasn’t building this tower in it’s prime elements but to figure out a way to populate/dress it and fast. Due to how early on in the project it was, I couldn’t possibly note down how long it would take me to dress this building properly because the asset list would have been enormous and unpredictable however I was able to estimate on large scale segments of the tower itself which was where I started.

For this project I roughly calculated that I had about 2-3 days of production time per major segment to the tower structure. I would spend the first day sculpting and the next 2 days retopologizing, generating UVs and texturing.

1 of 2

However, the texturing process in this iteration would be a very fast pass for visualization purposes and in most cases wouldn’t look pretty at all.

Once everything was planned out I began blocking everything out in 3dsmax, my initial approach was to get a block out into Unreal Engine 4 as soon as possible because the most important part of the block out phase for me is obtaining proper scale. Having the block out ready inside UE4 gives me a direct experience on how you feel inside this world as a player and it’s by far the most necessary element that I needed to nail before advancing on.

As I progressed through this project I was aiming mainly at getting all the biggest elements of the environment to a ready-to-polish stage as soon as possible so I could get a grasp of how much time I’d need and have left to further judge what I could get done for the rest of the project.

The Tower

I first blocked out the tower in 3DS Max, using Leon’s concept in my viewport background to help me camera match. People have wondered whether I built this to exact measurements or have I simply eyeballed the entire environment? Well, truth is it’s a little bit of both, I have exact snapping measurements for the majority of this environment but there are times you just simply break the rules for the sake of visual benefit.

I believe I began defining scale by trying to figure out how big that bridge was and building a block out for it which I camera matched with the concept. I then used a box (60x60x180) to justify player scale and I began dragging that box around my viewport to see whether it fit together. Essentially I worked my way around things by comparing one piece to the next piece such as a single story in a building can be 3-4 meters high (300-400 generic units) and I used that as a measuring guide for the majority of the block out stage however there were moments where exact measurements no longer played a part and it was really down to eye balling e.g. the concrete shell of the tower, fragmented parts of the tower, metal panel coating, etc.

I’ll be perfectly honest and tell you that texturing this structure wasn’t an easy task, it took a lot of time trying to get the material shader working the way I wanted.

There’s a common issue that all environment artists hate and it’s repetition, how do you avoid repetition? How is it even possible over a surface this big? Well, it wasn’t easy but I’m happy to share my approach on tackling it.

The trick to breaking repetition is often overlooked because a lot of artists tend to think it’s all done in just the texture, I personally feel that it’s the wrong way to think. Fact is it’s extremely hard to generate a single material that doesn’t show repetition when you’re tiling it over 10-20 times and the most efficient way I’ve learned to tackle this is to utilize this simple equation…

Multiple texture sets blended by a mask / Surface Decals / Prop dressing

This combination of elements are what break repetition in large areas and that is exactly what I used during this project, however there are still other tricks such as modeling into the geometry & flipping UVs when applicable.

I setup my shader to use an RGB mask as a second UV channel as a method for vertex colour blending which I would use to texture specific parts of the tower. In the final master material, it would tell my model that anything with black vertex colour should utilize Texture A, anything with red would be Texture B, anything green would be Texture C and anything blue would be a flat colour which I used specifically to paint additional grunge into for further surface break up.

Although this method is already very powerful it doesn’t always finish the job and it’s necessary to add decals and dressing on top to further hide any potential repeats in the model.

There are other ways of achieving the same results but this was definitely the most appropriate for me during this project because it’s fast and it requires far less components to build it.

The Terracing

The city I built within the tower was a very time-consuming segment of the project, I essentially built a 3-4 story modular building which followed the same texture method as the tower segments, this modular asset would be able to be rotated and snapped together on all sides.

Although the buildings snapped together every 100 units there were many cases where it was necessary to intersect the buildings into each other and I’d later cover up any abnormal cramped looking holes with props such as my cargo container. This was very necessary to obtain the look of over packed terracing like the Kowloon Walled City or Favelas.

To further make this asset versatile I set up each block of the complex to have it’s own material ID in order to be able to switch out material instances for colour variation across the environment, in order to make these surfaces blend and transition between each other I would place a hovering piece of geometry with an alpha cutout texture of plaster to fake the look of these buildings being welded together with cement, I also used the same technique to eliminate a majority of hard edges on the edges of my buildings as well.

To finish off this building I would have 8 windows & doors pre-built that I would snap into place on the building before using the entire segment to duplicate around the world.

After the building segments were finalized I generally used it as a guide to building the rest of my assets to dress them in further detail, for example, the wooden walk bridges/canopies/stairs/etc.

In general most of the time building an environment it just feels like I’m playing a video game quest, you complete one quest which opens another and sometimes it’s necessary to finish a specific quest in order to move onto the next. This asset specifically opened a whole new ball game of the environment so it was necessary to get this segment done immediately after the tower shell.

Substances

For the duration of this project, I had built around 26-30 substances, not all of which made it into the final version and was cut from the environment.

On an average when I need to build a highly detailed substance and fast I can get a decently detailed one done in 4-6 hours of absolute focus. This is a skill I’ve managed to acquire from very intense deadlines received from clients during my time as a freelance artist, I’ve managed to train a skill where you can generally just put your pride and ego aside and just call it decent enough to move on. Of course it’s easy to say you could keep improving something but you have to be responsible with your time in order to get the project done so in order to knock out the entire material list I would treat my substances much like my models and build them to a state that’s got all the main elements that I need in order to see how it works in my final composition. I’ll build it up to that point and then leave it until later in the project to polish.

Although I built a lot of tiling textures in Substance Designer it was still necessary to be creative with geometry to get the most out of them. In fact the metal plating that sits on top of the tower shell is made of one metal sheet substance, I applied this tiling texture to a plane which I then cut around and made several assets from to duplicate across the surface of the tower, I then built accompanying assets to place on top to give the illusion of panels falling out of place.

Although I built a lot of substances for the environment I cannot stress enough how important it is to approach your environment with the intention to blend these surfaces together with props and decals wherever necessary as it helps a lot to tie things together, it really is hard to build a substance that doesn’t show repetition at one point; not to say you shouldn’t try but again in a deadline like this you take what you can and work with it.

Considering the size of this environment whatever I added in terms of substances all had a huge involvement one way or another. For the majority of this environment, I’ve built my substances fully and completely within Substance Designer, the only few instances where I had used photographs were for assets such as the floating fish decals & the ocean foam. It’s important that you can make a conscious decision to avoid wasting valuable time on building something from scratch when you know you’re already short on time, so in this instance I felt using photographs were the smartest move for small details such as these.

Essentially how I approach my substances are the exact same way I approach my models/environments. I tackle the largest forms and then work into smaller details later, it’s critical you do not buckle down into fine detail early because the more you dive into micro-detail the harder it becomes to self judge your materials.

As an open example of how I build my substances, I’ll talk about this concrete blocks substance…

This substance is used for the floor of the bridge walkways and also for a lot of the tower surface itself.

I’ve taken the liberty of cutting together a breakdown of the graph for this substance to explain my approach when building my graph…

1 of 2

Every substance is different and in this case I didn’t spend much time refining the roughness because it was a fairly simple substance to work with. In most cases when working with bricks, etc you’d want to work as much into your roughness as possible since it’s the most important map in PBR.

Lighting

Lighting in this level was generally very basic and far from complex. I chose to go with dynamic lighting to save time having to avoid creating lightmap UV’s for every asset and I chose to activate Light Propagation Volumes as well for much more visually interesting results. The use of cubemaps were also involved in my skylight and one or two capture spheres for cinematic effect but in the entirety of this scene I have probably around 6-7 lights, 8-12 capture spheres and 1 reflection capture plane for the ocean.

Although the lighting was not complex it was far from easy to light this scene in a way where everything felt balanced. What I mean by that is that if I adjusted my lights to look good on my ocean shader it would sometimes make the tower look flat and boring, if I adjusted the lights to work with both the tower and the ocean shader then my foliage would start looking flat. Essentially I spent a decent amount of time just fine tuning the position of lights and shader tweaks just to get a result I liked.

Most of the time while lighting my scene I tried to make sure my highest reflections were coming off the main tower for obvious reasons, I didn’t want to jeopardize my composition by having too many reflections coming off all the towers in case it developed conflicting focal points in the final shot.

It’s generally details like these that I focus on when dealing with lighting and prop dressing.

Adding Motion

I don’t know about you but I generally like to see my environment or artwork come to life, it’s a whole new feeling when you do. I chose to add little essences like birds, wind effects, etc not just because some lovely people suggested ideas to me but also because I didn’t want to just post some still images and call it a day. I wanted my viewers to feel like this was a real location with sounds, life and elements from the real world.

It’s weird because I always like to dive deep into environments when I find myself at a polishing stage, I will add sounds, particle effects, animated sliding doors, dynamic lighting functions, etc… all for the sake of convincing myself that this is a real location. Funny enough, I was actually considering making a night-time lighting setup of this environment too, that was partially why I decided to build the lamps on the bridges which I had planned to use in the inner city as well but I simply didn’t have enough time left to attempt it.

You probably think I’m crazy for even thinking about attempting a second lighting setup… Well, I was dead serious and I had emissive maps ready.

Generally, I feel that building something to feel realistic requires hint of life and motion.

Efficiency

To round this up I would really like to take a moment to re-emphasize what made this entire project possible and it was understanding what my work speed was and understanding what is necessary to cut out of the pipeline if it isn’t needed.

Substance Painter and Substance Designer are an extremely helpful addition to my workflow for high-quality results and time efficiency.

Having Designer in my pipeline really helps me not get attached to my textures because I don’t end up spending obsessive amounts of time refining small areas by painting things out like how we used to in Photoshop and because of the nature of Designer it’s made it easy to make decisions like discarding a material that isn’t working and it allows me to either start a new material very quickly or recycle parts from the failed material that actually works decently for the new one.

I would say the biggest time saver for me is centralized around Substance Painter. It has cut out a very time consuming part of asset creation which is the high-poly stage. Although a high-poly is still very necessary to do when you’re dealing with a comfortable time frame, it can almost always be skipped when you’re dealing with tight deadlines. It has essentially allowed me to create high quality assets with the use of customized normals on low-poly unwrapped geometry and manually drawing/stamping height details that free up a huge amount of time in the pipeline.

Having Substance in my workflow has definitely been the single reason I was able to knock out such a daring environment in such a small amount of time.

Other than that, just try to approach things in a modular mentality, don’t try to build everything uniquely because most of the time you can get a lot of variation using modular pieces smartly.

This environment has an incredibly amount of geometry in it and it’s all built using as much modular use as possible.

This world space normal view can give a brief idea of how much geometry sits within this world.

END

I would like to thank you guys here at 80.lv for taking the time to read through my break down, it has been a pleasure and I have enjoyed being able to share my thoughts and workflow for this challenge with you.

I hope this has proven to be informative and helped people in further creating their artwork!

Desmond Man, Environment Artist at Sumo Digital Sheffield

Interview conducted by Kirill Tokarev

Follow 80.lv on Facebook, Twitter and Instagram

Join discussion

Comments 0

    You might also like

    We need your consent

    We use cookies on this website to make your browsing experience better. By using the site you agree to our use of cookies.Learn more