AI applications are only beginning to materialize. It's so exciting what neural networks will do to art workflows in the coming years.
W T F!!! Why, really don't understand, EA = just a fucking looser company!
Amazing... Congratulations for the new way to show information.. I hope they could use this to teaching.
Jared Chavez showed some of the things he learned about environment design during the classes given by Andres Rodriguez and Clinton Crumpler.
My name is Jared Chavez, and I’m 3D Character/Environment Artist from Albuquerque, New Mexico. I attended the University of New Mexico where I began my studies in 3D art.
During my time at UNM I gained a knowledge more focused on film, and basic video game development. My primary focus in college was modeling for film and games. In an attempt to further my knowledge I decided to join CGMA’s Character Arts Program. I have been taking classes with them for the past year and a half. During my time there I have taken a wide range of classes mostly focusing on character arts as that’s where my passions reside. I am currently working as a 3D Character/Environment Artist for Northrop Grumman. I joined the team at Northrop a few months out of college. At Northrop I have had a wide range of responsibilities during my time there, ranging from 2D training material, to 3D VR training courses. Over the past year we have been pushing forward with VR courseware and training materials where most of my efforts lay. In an effort to help myself become a more valuable asset in our VR pipeline, I decided to try my hand at Environment work. I started by getting my feet wet with Andres Rodriguez’s Intro to Environment class, and closely followed up with Modular Environment Art in UE4 with Clinton Crumpler.
Well, when starting any project I think a huge thing I like to make sure that I do is start by organizing everything. I setup a file structure so that I can keep all of my WIP files and my final files neat and organized. Secondly, I like to make sure that I gather as much reference material as possible. I really like to do my best and plan out as much as I possibly can. A resource that I use to do this is actually Pinterest. I usually start most projects by creating a board and adding anything that I can use on it. This allows me to have a place where I can easily scroll through and look at any of my reference material, and in the moment if I need more I can just look through related images to that picture.
When starting this exercise off I knew that I really wanted to create a library. Well, being a huge fan of the Harry Potter series I had always wanted to do a place in a similar vein. So in looking up libraries I stumbled across the Library of Oxford which many of the Harry Potter movies used as a set, so I figured it was a perfect match. I also looked to pieces by Clinton Crumpler, Aiko Shinohara, Andres Rodriguez, and Ognyan Zahariev for inspiration as well. Once settling upon a location I was then able to determine the mood, story, and feelings that I wanted the scene to evoke. I then searched for as much reference as I could to find things that would help me develop the assets I needed to construct and fill out the scene.
Once all that was finished up it was on to the planning stage. This is one of the most important aspects when creating an environment like this one that is based around a modular set. At this stage I really try my best to figure out how I am going to build each asset, what the scale and measurements are going to be, and how everything is going to assemble and fit together.
One of the big challenges that came with this project was figuring out the correct scale and measurements to use for things like the pillars, bookshelves, railings and things of that nature. To resolve this issue I did a lot of measuring based around real world things in my apartment, or approximations based around what I found in my reference material.
I would say the next key to making a successful scene that feels accurate and to scale is blocking things out. Spend the time to make sure that things are feeling to scale with basic shapes because this is the time to make adjustments before getting too far ahead of yourself. I know for me I have the constant itch to get to the fun parts of actually making the assets and texturing them, but take the time to plan and get things right before touching anything else. You’ll avoid a lot of headache down the road.
My modeling workflow is relatively straightforward. When it came to modeling many of the modular assets like the poles or bookshelves I like to take my box mesh which is usually just a cube and bring that to a new scene and use that as a guide for how much space that object can take up. From there I usually start with a basic primitive. Luckily most of the modular pieces were relatively simple to create because they could be broken down into simple shapes like cylinders or cubes. From there I would just refine the shape to closer match the silhouette that I wanted. When doing this step I tried to not go too crazy because I wanted this project to also be an exercise for optimization in VR. The final touch that I added to all of the modular pieces is beveling any hard edges. The main reason I do that is to avoid any sharp CG edges as well as give it some nice light play to the model once it’s in engine.
For the unique props I tend to follow a similar workflow of getting a base volume size that I can work in and just block out from there. When I was creating these assets I tried to get my blockout as close to a low poly asset as I could, so from there all I would really need to do is subdivide it up to get a nice high poly for baking. For my normal workflow I like to take most of my assets to Zbrush and work things in there, but I knew this project wasn’t going to have time to do that so I made the plan to do any of that detail work that I needed in the texturing process when I was in Painter. If it can be done in Painter than do it in Painter. Which I think helped me save a lot of time in the long run.
I want to say about ten of the assets I would classify as unique. When I was figuring out what unique assets I wanted to use I gravitated towards assets that didn’t really stick out too much. I wanted to be able to use anything and everything to fill out the scene and not have it stand out like a sore thumb. If I used it more than once, for example, the hanging lamp was one of the latter unique assets I did. I knew that it was going to have a texture that was a unique material from everything else in the scene but it had such a common use that people wouldn’t give it much thought to being reused multiple times.
The rules of modularity
The first rule I made sure to use was to keep everything inside of the bounding box volume I used as the initial blockout. I knew that everything had been measured out prior to so as long as I kept the models inside that space everything should fit together to form any of the areas I needed.
I would say there isn’t necessarily a rule to follow that helps something not feel repetitive but the way that I try to break that up as much as possible was by creating three different variations of things like the bookshelves or the railings. The reason I choose three and not two was because of how much every item was repeating. I knew that you wouldn’t see the same piece repeating next to each other and I was able to create some space between each variant.
The books on the other hand had about six variations that could have been used. I created kind of a brush palette out of shelved books. In the story telling of the scene I knew that I wanted it to feel a little chaotic and I wanted the books to portray that so I didn’t make versions that had a shelf completely filled. Once I got into engine I spent a lot of time using that palette and hand placed them on top of the shelves in no order with different books in hopes to create enough variation so that when someone looks at it they don’t think it is just one asset throughout.
The first step on my pipeline really started with breaking down what materials I was going to need to texture my assets. After determining that I had a list that wasn’t very long so I went the route of using Designer to make some of my own base materials. These could be used as tileables, trim sheets, or I could bring these materials over to Painter and use to texture my unique assets.
With all of the modular assets, I created a tileable wood texture, as well as a trim sheet that had some spots of edge wear to it that I used to map any of my hard edges to. That was another way that I was able to create some variation to an asset. The same process was applied to the metals, and any of the other assets that were modular.
When it came to the unique assets I brought them all into Painter where I baked down my smoothed high poly to get all of the maps that I needed for painting. Being that a lot of my objects were cylindrical there tended to be some baking errors. With a quick jump over to Photoshop I was able to just clone them out and fix them manually. Once back in painter I like to put together a pristine version of the asset, kind of a straight off the factory floor feel. From there I layer on the aging process and begin to think of how it’s going to wear out over time. Most of this time is spent working in the roughness map. When working my textures I put most of my effort into what the roughness map is going to look like because ultimately a final object will look flat if you don’t have those nice material responses with light. After giving that nice level of wear and tear to the object I like to run through a dirt, dust, and grease pass. I feel like this pass really grounds the asset.
Once inside UE4, I tried to keep things simple. I built some master materials that gave me some extra control in engine. Some of the additional parameters were to adjust color, roughness, or add some different tiling options. I tried my best to do everything right the first time so that I didn’t have to do much inside the shader, but I put them there as a backup in case I need to tweak anything.
One thing I was really surprised about when getting into UE4, is just how nice the cameras are. I used the basic camera actor and would pilot it around the scene looking for any possible composition that I thought could help tell the story I wanted to tell. Having some background experience in photography I have a pretty good understanding of how to operate a camera, and when shooting I have a tendency to use a larger depth of field to draw the viewer to where I want them to look. The UE4 camera has a couple of different options on what kind of DOF to use but I ended up going with Circle DOF because it is closer to using a real world camera. Another great feature in UE4 is their DOF visualizer. It was easy to use and read to get me the right distances I needed to setup the camera properly.
The first thing I did prior to lighting was looking at some of the reference I had gathered in my preproduction, as well as start to think about the mood I wanted to evoke.
When I light things, I enjoy creating darker lighting setups. I like to try and use shadows, and really push my blacks when possible. To accomplish that I really tried to use light only where necessary. I played a lot with their radius so that I wasn’t having light reaching areas that I didn’t want it to. I tried to keep it somewhat contained.
I really only used two different types of lights for the scene. The first was a sky light, which I set up closer to a moon coloring. The other light type I used were point lights. I kept these lights at a warmer value.
When checking my scene in VR the one thing I noticed that seemed to have an impact on performance was the lighting. If not kept in check things can get out of control really quickly. I tested a couple of different light setup ups and the one that ensured the best performance results seemed to be using static light. Monitoring how many stationary or moveable lights you use in your scene is vital because these can really cause a performance hit. For this piece I limited it down to a single stationary light being the moon light, and I left everything else as a static light. I adjusted some of the lightmass settings on the final light bake, and felt things held up just fine for what I wanted. Bake your lights! That is the key, or at least in my work.
I would say I truly learned a lot of things from this project. I think the biggest thing I learned was how to look at different objects and break them up in a way that you can create modular and reusable assets. Next I think it would be how important planning is for environments. You may think you have it all figured out, but even at that point plan it out a little more.
In total I think this project took about 12 weeks to finish. I had 10 weeks in the class, but needed an extra two or so to finish up some of the props that I wanted to add. I found myself continuing to find things that I wanted to add and just had to cut it off after a certain point. I think heading into the next project I can be fast in the planning aspect because I now have an idea of the things that I need to look for and figure out. Being that this was my first modular project answering some of those questions had me tied up a little longer than I would have liked. Luckily, I think I was able to make up some of that time with my speed in modeling and texturing some of the assets.
Lastly, I just wanted to say thank you to Kirill and 80lvl for the chance to do this interview.