Deadly Sweet: Complex Scene Creation

3d artist Eli Louis Creeley from Ringling College of Art and Design talked about the production of his amazing Deadly Sweet project. It’s a great animated environment, created in Unreal Engine 4. In this post Eli described the production process, the choice of tools and gave some advice as to virtual scene creation.

Introduction

My name is Eli Louis Creeley, I’m currently a senior at Ringling College of Art and Design, majoring in Game Art. The Game Art BFA program at Ringling College brings our feature film aesthetic to games and is focused on providing students with the professional artistic skills necessary to create compelling and believable interactive experiences. To this date I have not had the chance to work on any indie games or produce art for any companies. I have worked on a variety of projects mainly focused on environment art. For example the Fishing Village is a project that can be viewed in my portfolio. It was the first enviroment art piece I ever put together in an Engine. It was my first time modeling in Maya, lighting, doing 3D digital art in general.

Another project is my Wine Cellar, which was a Modular Kit I made.

I’ve also made a few characters, which can be seen in my portfolio.

I’ve also done a few projects focused on basic blueprint scripting which are currently not documented in a public format.

Deadly Sweet

Deadly Sweet is my senior thesis at Ringling College. It’s a trailer built in Unreal Engine 4 and put to the tune of “In the Hall of the Mountain King”. It focuses on a conflict between the player character, a trendy Final Fantasy version of Black Licorice, and mobs of mint candies that are coming in and changing the kingdom from sour to sweet.

The environments are, at a glance, normal. However, when you take a closer look, everything is made of some kind of candy. The arches on the castle are bent licorice, the tree tops are made of cotton candy, the whole environment is on top of a chocolate cake! My goal was to make all of this seem as normal as possible. Its only apparent to people looking closely at the details.

As I mentioned, the forest scene is a smaller part of a bigger project. The scene in question is the first shot of my antagonists, the little mint candies. Its purpose is to show that they have the ability to change the environment they interact with. I also really wanted to make a forest made of candy… that was the driving factor there.

Production

The first problem I had to solve was figuring out how I wanted to transform the environment. I realised I could fake it all in Matine, but I really wanted it to work in gameplay too. Also, switching all those assets manually in matinee would take decades. I also realised the limitations of matinee, and how when I changed a material instance on one object, it changed every object that had that instance. I needed each object to transform one by one, as the mints touched them. I decided that each object that transformed would have its own blueprint. The blueprint is really simple. When the mints touch the sour object, the sweet version spawns and the sour object destroys itself. And vice versa for the hero character. Each asset also got its own particle effect which is essentially a burst of colorful sprites to hide any popping when I switch the meshes.

In the forest scene, I begin by bringing in a mint, who touches a tree and triggers it to change. That is done with the blueprint for that specific object.

Each objects blueprint is the same save it’s on a different object and spawns that objects double. Using Matine I change a parameter on the Material Instance. ‘Radius’ a parameter that controls the radius of the mask that I have on the Lerp node between the Sweet and Sour versions of the ground material. Along the edge of the growing ‘sweet’ spot, I have a trigger box that trigger the rest of the assets to change.

This meant that every object I wanted to change, needed to have a double. This is a fact I knew right from the start. I’m not a concept artist, but I did all the pre production for this project too, and had to design twice the amount of assets. I planned to keep each related object a similar size and shape.

Preproduction:

Lighting

Another challenge was anticipating how the scene would look when the new assets spawned, and then trying to light them. In order to see the composition of the shot, I had to export it, check it, make adjustments, and export it again to see the results. This process took quite awhile. I then had to light it. The majority of lights in that scene are dynamic. Using Matine, I adjust their colors, radius, and intensity over time. Again, to see the results I had to play/export the scene.

PreLighting Shots

Environment Production Process

Block Out

I approached my block out by making simple shaped object that were the correct scale. In block out I don’t care about how it looks, I first focus on scale, which was immensely important for this project. I wanted my big cotton candy trees to be massive, and then on top of that the castle had to tower over the whole environment.. I then set dressed my scene with those objects. I iterated on the environment by simply reimporting those objects as high poly models later.

Meshes

For the assets in the forest shot, each object had a high poly sculpt I did in Z­brush. I then used quad draw in Maya, or Z­brush’s zremesher tool to create a low poly version which was then UVed in Maya.

Most of the assets are textured with materials, and tiling textures. However, the trunk of my cotton candy tree was textured in substance painter.

The whole project is rendered in realtime with Unreal Engine 4. I made sure to use the PBR capabilities of the engine to my advantage while creating materials for my candy pieces.

VFX

Effects was one component of the project I had no idea how to do while going in. I made them all in Unreal’s Cascade. All of them are basically the same with simple tweaks to the size and amount of particle spawned.

Time Span

I can’t say how long this one scene took, since I was working on the whole level together. The level was started in late August, 2015, and I’m closing the project on March, 25th.

What I learned

The entire project taught me a lot about myself as an artist, and ways to produce art. The biggest lesson was dealing with scope. As one person, with three environments (not counting the transformation), a boat load of animation, and transforming assets and environments… I realised that smaller scopes can sometimes work out better because it allows more time for polish, and reiteration… two things I didn’t have time for.

At the same time, it gave me a good understanding of how to handle stress, deadlines, and crazy amounts of work. I learned how much I could handle, and what level of work I could produce under those type of conditions. It makes me feel good about going to work for a company because I know I can work under that kind of pressure.

I’ve also gotten a good understanding for the importance of frame rate, performance issues, as well as other problems one can face when working with large environments.

Eli Louis Creeley, 3d artist