Thanks for sharing and detailed production breakdown
i thought there wouldnt be anything better than akeytsu for creating easy animations. im happy if i am proven wrong.
Keith, I just wanted to stop by and say: Thank you.
Peter Dimitrov provided a detailed breakdown of his Animal Shrine created in UE4, Substance, Maya, and ZBrush.
My name is Peter Dimitrov. I am a student of Games Design at University of Central Lancashire in Preston, UK. That is where my first introduction to 3D started, almost 3 years ago. I am graduating with bachelors in a few months. Prior to that, I have quite a few years of experience working in freelance, painting and illustrating for games. I feel like my grasp of 2D is what sort of kick-started my 3D work and allowed me to get into it without too much struggle. I will talk about that relationship between the two mediums later on.
Figure 1. The Animal Shrine, main view.
The Animal Shrine is a Game Environment that I created for Artstation’s Feudal Japan Challenge. Most likely a lot of you are familiar with the competition. Participants were given a deadline of nearly 2 months to create a game level from scratch. Going into it, I knew that lots of amazingly creative people will take part and that the competition will be fierce. Also, this was the first project of such a big scale I created in 3D, so I stepped into it a bit insecure but knew that the only way to fail was if I weren’t to try.
The level was assembled in Unreal Engine. Modeling and sculpting are done respectively in Maya and ZBrush. For texturing, I created my materials in Substance Designer and then imported them to Painter. A small part of the foliage is created in SpeedTree and the mountains in World Machine. I will try to give more insight into the use of software later through the article.
Theme and Idea
The topic of the project came from Artstation itself. It is Feudal Japan: The Shogunate. Feudalism in the country was from as early as 1185 up until as late as 1868 (according to Wikipedia). I liked that, as the broad time period gave lots of freedom and lots of ideas to pick from.
Figure 2. Few books on the topic and lots of images from the Internet.
A few days prior to the start of the Challenge, I went to my Uni library and borrowed a few books on the topic. I wanted to see some images in them but also read a bit about Japanese architecture and why they created their buildings the way they did. Throughout the challenge, I would gather pictures on whatever I was making, put them on an A4 reference sheet, and keep that open at all time on my second monitor. I also did lots of stalking on Flickr, of people that visited places like Izumo Taisha – the Shinto Shrine that came to be the biggest influence on the visual style of my environment.
Figure 3. A quick GIF with a preview of some of the images I used through the process.
Main block-out and greyboxing
At the very start of the project, I started immediately with Unreal. I wanted to shape out the main place, see how it would look and feel. I didn’t want to spend any time modeling or sculpting anything, given that I didn’t know whether it was all going to look nice as shapes and silhouettes when assembled. I used the geometry brushes (BSP) that the engine provides. Instead of extruding any geometry by dragging, I used the measurements in the Brush Settings panel of each placed brush. That proved very useful in the sense that once the greybox was ready, I just took every measurement that was already set and used it in Maya. While making that first block-out in the Engine, I had placed a mannequin. That’s the dummy that comes with Unreal (you can extract it as a *.FBX from a blank 3rd person project). It gives me a sense of the scale of the place. That way, every prop created later, is using real-life measurements. That solves a lot of initial questions one might have when going into modeling and makes the whole process times easier and faster. It eliminates any guesswork which might prove troublesome.
Figure 4. First look of the project with a GIF.
When I felt comfortable with the look of the place, I started to slowly replace each BSP with props made in Maya. I deployed a modular approach, in the sense that I break up every building to small, repeatable walls, wooden foundations, and scaffoldings, and then I repeat the same props over and over again, building up the whole place.
Figure 5. Replacing BSPs with props.
Here you can see my initial sketch of the place before starting the project. Also, some notes:
Figure 6. Notes and views of buildings.
In the drawing to the left, you can see my initial idea. From the start, I had the concept of making a few buildings and showcasing the interior of them. I wanted something interesting, so I decided to go with a Geisha house and the workshop of a master calligrapher. The idea of scattering different scrolls and working materials of a calligrapher seemed very alluring to me. I really wanted to have a go at making paper posters with beautiful calligraphy. It’s an idea that I got after seeing a few old photos in some of the books I borrowed.
I quickly realized that if I were to make the exterior of everything, and then had to tackle multiple different interiors as well, I wouldn’t be able to make it in time. Not in 2 months at least. As such, I scrapped the concept of a Geisha house and left just the Calligraphers. I had some thoughts about making a garden too but quickly dropped that as well. After all, I knew I will have a go at making a shrine as well.
I included those notes in the picture, wanting to mention modularity again. I am going to share with you a mistake I nearly committed to. There are lots of other mistakes I did commit to, but let’s explore this one for now. It will probably sound basic but might prove helpful to someone out there. Look at the wooden scaffold in the picture above. As I said I had blocked that out with basic brushes at first. Those brushes were multiple, all varying in size. I went and measured each one. It came out to 12 individual pieces. I was actually very, very close, to modeling 12 individual pieces… Then I would have to texture each individually too. I quickly blocked most of them in Maya. I then decided to take 2-3 different in size into Zbrush, sculpt them and texture them after that. Just to make sure I can make one look the way I want, before committing to multiple. I was fortunate to do that, because after I did the initial two, and tested them out, I immediately realized that I can use those, rotate them, scale them around, have bits go underground, and replicate all of the scaffolding, with just those two pieces, instead of 12… Not only does this save time, but it also is amazingly more optimized for the engine, in case that this was to be an actual game. I knew to work modularly, but still nearly missed on that. As the old proverb goes – measure twice before you cut.
Figure 7. All props the scaffolding is made from.
I even went further with it. I just sculpted and textured the second, very tall piece. I then went in Maya and inserted an edge loop in the very middle. I separated the two pieces and then exported them as 2 other props. Wherever I needed very tiny support, and couldn’t hide part in the ground, I just used one of those.
Boring is boring. A change in layout and composition.
My research and the fact that I was looking at Izumo Taisha, led me to the idea of making everything enclosed in wooden fencing. It is in fact based on reality, and all the reading I did, made me believe that indeed, in Japan they would make a bunch of flat lands and then enclose everything. Each building from the Shrine would then be in there. Sure, it works and looks just amazing in photos. But I would fly around, try to capture interesting and dynamic angles and time and time again struggle with it. It was then pointed out to me, by friends at Uni, that it’s all great that I have multiple levels of elevation, but it still looks a bit flat and boring.
I had already invested lots of time, so I could have ignored that and continued. But instead, I wanted to take advantage of the fact that everything is modular and have a go at breaking it and making it even more dynamic.
Figure 8. Change in the layout. The calligraphers and shrine go on a lower level.
That’s how the following shot was born, and I knew I am once again in the right direction:
Figure 9. Main camera after the change.
Painting has taught me, no matter how big and scary a mistake might look, you should never simply ignore it. Go into it, try to fix it, and even if you struggle a bit at first, at the end you will realize that it wasn’t that scary after all, and fixing it actually didn’t take nearly as much time as you thought it would. Fear is what stops most people, and even when they feel the need for change, they ignore it. Making this change in layout took me only about 2 hours.
Figure 10. Different bits.
Change in layout was relatively quick and easy, once again, partially thanks to modularity. Green stone walls are 2 pieces in order to have more variety. Their sizing is the same. Stones are geometry, not just texture because that way it looks better and is actually quicker to make than investing lots of hours into a realistic texture in Substance Designer.
I came up with the idea of building the Shrine to be one where people leave gifts and maybe sacrifices to different animal spirits. Later on, in the calligrapher’s hut, I made it so that a few scrolls refer to the shrine as an “Animal Shrine”. Then some other scrolls explain that “…this Shrine is a place for those who want to command the animals as their soldiers”. Among others are “Spirits inhabit this place”, “The statues shall come to life” and “Point at your enemies and the animals shall follow”. I was fortunate enough to have a good friend from Japan help me with translating my ideas from English to Japanese. Thank you, Rika! This whole idea of the animal statues, I tried to express in the end bit of the showcase video. If you haven’t seen it already, make sure to check it out.
Figure 11. Some of the statues.
Materials and textures
As I said before, I made all of my materials in Substance Designer. The time to work on this was very limited, and as such, I had to make some stylistic decisions. An example of that can be seen in the bricks and stones. Instead of having detailed moss texture, or even some geometry or a complicated shader to give more dimension through tessellation or bump offset, I decided to convey it only through color and noise in roughness.
Figure 12. Some of the materials I created.
An example of what I described can be seen in the picture above. I created the “stone” material in Designer and then exported it as *. SBSAR with exposed variables for HSL and Roughness. Then in Painter, whenever I was to texture an object made from stone, I would use the stone, and then duplicate it and change its color to green, roughness to a bit wetter too. That changed material, I would then mask using the Curvature and Thickness maps I had baked from my high-poly version of the object onto the low-poly.
I wasn’t too certain how I wanted my wood textures too. That’s why I created a few different variations (seen in the first row). I had every single aspect of them exposed as parameters as well. I would then import to Painter, and again overlaying different materials with different Smart Masks, changing color, roughness and so on, I created the 3 wooden materials you see on row 2.
Lights and Atmosphere
When it comes to lighting up the place, everything is very basic. I would love to have some secret to share with you, but there sadly isn’t one. Something I would recommend is to not overexpose your scene with light. As such, your Directional light shouldn’t be too strong. Leave it a bit weaker, and then you will have room to work with more spotlights and point lights. Think about it this way – if everything is super bright, no matter where you place lights, they won’t really be visible or pop out. Which leads me to my next point; when making a model, most people think about the geometry and the texture. No matter how good you make those two though, the object will still not be very impressive unless its lit in a good way. As such, try to “sculpt” your objects using light. Think about what conditions of light you will place your object into from the very first moment you start making the object. Thinking about light can never start too early.
I’ve made my statues, for example, pop out with simple Spotlights low in intensity and varying in color. Spotlights are cheap. A few times cheaper than point lights. You shouldn’t be scared to use them. Even better – in a lot of cases, you can insert one, and turn off “Cast Shadows”. Insert as many of those as you want, and the Engine won’t feel a thing, they don’t need shadows baked after all.
Figure 15. Hanging lanterns.
Think about the source of your lights too. A scene will always feel much more natural when the player/viewer can actually see the source light is coming from. It doesn’t need to be super realistic. You can have one lamp, and then cast 2-3 spotlight underneath it or going to its sides to achieve your effect. Having that lamp prop there though, that’s what makes the difference.
Figure 16. Lamps that use Light Functions.
A final tip would be the use of Light Functions. In the picture above, I have completely faked the light coming from the square lamp on the right. Instead of having a single Point light inside the object and struggling to make it somehow cast nicely looking squarish shadows, I created a Light Function to emulate the shadows. I used that on a spotlight and then deployed 4 more facing in different directions.
A Light Function basically tells the light in what region to be rendered. That way we can “fake” any shape we want. In this case, I went into Designer and quickly created a gradient with trimmed sides and a cross going through the middle. You can easily do that in Photoshop too. You export it as 1k texture, single channel b&w mask. In Unreal, right click in the Content Folder and create a Material. Open that, and from the Material Settings change “Material Domain” to Light Function. Insert the bitmap of your texture in there and connect it to the Emissive Color pin. That’s it. Whenever you insert a spotlight now, in the Settings Panel of it, scroll down to Light Function and assign your newly created material.
Light in the 2D mediums
One final tip about light, which is more of a personal opinion, so you can take it with a grain of salt; if you want to practice light, if you ask me, you should look elsewhere too, not just in 3D. This is where mediums kind of meeting, as I said at the start of this article. Read about light conditions and use of color in Photography, and even better in Painting. You can learn immense amounts just by observing and reading about the right work of art.
In painting, we use and exploit light before anything else. If you can look at a painting, imagine it in 3D space, and then imagine from what direction, in theory, the lights come, how many they are, what color and what intensity they are, you won’t have trouble lighting up a scene in 3D next time you make one. One of my favorite artists, from which you can learn lots, is James Gurney. He has a book on the topic of Light, which I honestly treat like the Bible. Even better, he has a Youtube Channel. And I know, I am asking you, who wants to do 3D, to sit down and watch someone paint, but honestly, his videos are very short and in each, he exclusively talks about Light and Color. He does studies from real life and he explains everything he sees broken into exactly that – light. Worth a watch.
In conclusion, I would say, don’t be afraid to make changes. No matter how daunting it might look. And think about the light conditions of the whole environment, with every object you make, don’t wait until the very end!
I hope you found this peek into my work insightful. There is much more I would love to talk about, but this article is getting too long already. Feel free to check out my Artstation blog about the creation of the project. Everything was written whilst in the process of making it, so there should be more insight over there.
While making the project, I took close to 450 screenshots… I assembled lots of them into a time-lapse where you can see in action a lot of the changes I talked about in this article. Make sure to watch it:
If you would like to follow my work, you can do so over at those channels:
Thanks for reading, and thanks to 80.lv for giving me the opportunity to write this!