Setting Up a Realistic 3D Toymaker's Workshop Environment

Introduction

My name is Timothy Lee, and I am a Junior Environment Artist in the UK, currently looking for an art role in the games industry. I recently graduated from the University of Staffordshire on a Games Art course 5 months ago and have been working on multiple different projects, pursuing my career as an Environment Artist. Some of these have been group projects working on some new games, one called Forgive Me Not, working on the environments, which really helped my environment art skills. 

I've always been a big fan of video games and have loved the different art styles and stories that video games can convey. I wanted to become a 3D Artist mainly due to how I can create an object or environment that has meaning/a story behind it, making it feel alive. I gained the skills for environment art from learning at University, but mainly from tutorials, as well as continually working on my 3D and technical art skills by creating projects like these.

Toy Makers Workshop

I started the idea of the Toy Makers Workshop after my friend told me a story about how his grandad made basic wooden toys for him when he was a child. This made me remember how I had the same moment where my own grandad made wooden planes and soldiers as a kid. This being said, I wanted to create a project that encapsulated the idea of a Toy Maker who created things for a hobby. 

I started by collecting multiple references for the overall design of the workshop, and I found ones that fit quite well for a shed that could have multiple types of props. This was an important step, as finding many types of references for different things like props, location, lighting, and even types of materials would help me down the line and pinpoint the exact type of feel I wanted to present in this project.

I wasn't using a baseline 2D concept art for reference for how I could create the environment, so I had to splice multiple together to get an idea of what the scene could look like. This helped me to create the blockout and what the level design could look like.

I have seen many artists try and rush into creating props and environments straight away without proper planning, myself having done this before, realising the mistake as you need to find correct proportions for areas in the scene.

With the blockout created, I wanted to create a benchmark asset that would be the requirement for the other assets in the scene. As I wanted to focus on complete realism, I had to practice creating a hand-crank drill prop with the full pipeline, low-poly, high-poly, UVs, texturing, and material setup.

This really allows me to grasp the exact sort of optimization I want to achieve for other assets in the scene, making sure I have a consistent style.

Creating tiling materials is always something necessary with environment art, for optimization and for texel density, so that it fits the same density as other props. I did not need to create many tiling materials in this scene, however, the ones I made would take up quite a bit of the screen, so they had to look good.

I used online references for these materials, but I do have a shed that I can easily access and get real-world references from. Being able to physically see and feel some of the materials really gives a big improvement in what different values I could put into the materials to look more realistic.

With a slightly more technical talk on texture packing, I had to decide between different ways of packing these textures for Unreal Engine, and whether it would be easier to choose an easier or more optimised method. I could have packed the textures into a basic ORM Map, Ambient Occlusion, Roughness, Metallic, or a more optimised method, or created an Albedo, Roughness Map, and Normal, Metallic, and AO Map.

I could unpack the Normal data using only two color channels, and have 2 left for other maps. This would be really good for optimisation, however, I did want to make this scene into an asset pack, and having textures that were more user-friendly made more sense than over-optimisation, so I just used basic ORM packing.

Modeling Workflow

Finally, onto the biggest part of the project! I wanted to make every asset myself for the environment, other than any exterior assets. As stated before, I created the hand crank drill, which was my base model, to follow the same quality.

I could now create the other assets in the scene, which I fully modeled first, keeping them in bundled up areas that would be atlas packed on the same UV Map. Bigger assets would need to be on their own UV Map, but many assets, like the saws and hand files, could all be placed on the same atlas to save texture space.

As part of my development, I made all the assets first before texturing, so I could get a count for the amount of work that I needed to do. The assets themselves involved a few hard surface models, like the large Lathe and many box-modelled assets.

Keeping the polycount low enough for the assets was a good start, however, I had to make sure to keep the silhouettes for props, especially when it came to curved surfaces not to show faceting. This did become extremely difficult for areas like the cogs on the lathe, as I needed a lot of topology to show the teeth of the cogs.

Now that I had the models made and unwrapped, I could finally get onto my favourite part, bringing color to the scene by texturing the assets. My usual first step when creating any type of styled texture inside of Substance 3D Painter for an asset is to block out the different areas on the texture.

This gives me a base to work from, adding grunge and areas of detail like rust to the object. I like to add storytelling, like edgewear, after this step to give the prop some more life and purpose for the scene. Making sure to watch my Roughness and Albedo values by adding the most detail possible is extremely important, as these maps will make the prop stand out inside the engine.

Wooden Floor

A challenging area in my environment, which had to go through quite a bit of iteration, was the wooden floor. I started off with the basic designer texture using parallax occlusion on the master material to get a better 3D effect than just using a Normal Map.

This, alongside a mesh that had some geometry from wooden planks, helped add a more realistic level to the flooring. I also added some storytelling to the mix by adding a hole in the ground in front of the metal table, which also went through quite a few changes.

For the open gap in the wooden floor, I created a tiling material for chipped wood, and created an Alpha Map to place some fake splintered wood between the broken areas. This spiced up the damage and made the harder edges less pronounced. You can notice it if you zoom in on the wooden split section, but you wouldn't if played in a game with the player height being at an average human scale.

To blend between the damage and the wooden flooring, I used RGB Masking of this same material, as well as a dirt material to blend the cracks to add some more unique detail.

The wooden floor was looking so much better than when I started here, but what really made it stand out was the Normal Map Trimsheet that I made using ZBrush, which was blended with the Wood Floor Map in its shader. This added some more splintering effects and damage to the ground, which it did not have in the previous floor material.

Adding decals also made areas like the Roughness stand out, as I had wet mud on the ground as a decal that made distinctions between surfaces. I always suggest using decals as they really help with taking away the tiling look of surfaces, but you should never really try to rely on this, as using too many could be bad for optimization.

Assembling the Scene

With the scene assembly, I always start with the largest objects first for my own sanity, as it's also just best practice to get the biggest objects done to see how much room you are left with. This also goes for texturing, as I said previously about tiling and trim sheets on surfaces that are shown more than others. With models made, I would slowly start to place them in the scene.

This is completely different from me using a premade asset pack to make an environment, as I already have the assets, so less planning needs to be made, with more focus on the level design. As I made everything, I knew where props would need to be seen and what props would be modular and what would stand out.

Using modular props was incredibly important in this project, as I did not want to make unique props for absolutely everything, this would be extremely slow and would kill my work ethic. This is why modularity comes in handy to save time on prop creation, but you have to place them in specific areas that people wouldn't notice too much.

An example is the book models, I only made five books, so having them in different angles and placements could at least give me more possibilities to make them look different.

Now that I was done with modeling, texturing assets, and importing them into the scene, I had to start creating shaders and more technical art required to improve smaller details in the scene. As I had quite a few textures, I made sure to create a master material that was used for like 90% of the materials in the scene by just using a material instance.

I had parameters that could change the textures for the instance and scaler values for the Roughness and Base Color. Having this one master material reduces draw calls and is way more optimised than having a master material for each texture.

Following the basic master material, I duplicated it and created a surface dust material. I had areas in the scene that could do with more surface dust for better realism. This material allows for dust to be placed on the surface of materials that were controlled by the Z axis.

I had other settings that changed how much and where the dust should be covering, which you can see from the video. I only wanted to place this onto a few materials, so I created a separate dedicated material for this instead of adding this to the main master material for all assets.

Decals are a big part of any environment art project, as they allow for grunge and unique modular details to be placed around the environment at a low cost. For my project, I wanted to also develop a shader from my own knowledge that allows for easily interchangeable textures and movements for decals.

This is similar to a material that I made for my Bioshock Infinite fan art project, but with some more important steps. The shader allows me to change the texture, taking specific channels into the outcome using booleans. I am only using black and white values for the opacity of the decal textures, so I could pack this into the R, G, B, and A channels with parameterized color and normals for extra detail control.

The reason why I think it was best to create this shader with material instances is that I had multiple decals on the texture that I wanted to scatter around the environment, so making a shader that I could control for each instance was important.

Moreover, I used very similar techniques to create a trim sheet shader, as I had many different wooden planks in the scene, so I needed to be able to control different values in the instances. I used the normal trim sheet that I used for the floor for the planks to also give a different normal variation.

Composition and Lighting

This is the main establishing shot of the project, focusing on the back of the shed with the majority of the assets and scene in view. I wanted to use this as it gives a wide-angle view of the environment whilst also adding a specific focus in the scene, using the angle of the props to direct the eyesight of the viewer to the centre of the image.

I tried to balance light values, as most people will look at the brighter parts of the image first, however, this was quite difficult to accomplish, considering the shed has a darkish wood tone with most of the other props following this color.

Through the project, I had many lighting changes, from having the environment be extremely orange to a more tamed and realistic scenario. I wanted to express a warm and autumnal feel with some of the renders, as I felt this would help make a cozy imagery of the story behind the environment. In some instances, I chose colder tones, like when rendering the metal drill to support the cold texture.

My main lighting setup followed using an HDRI backdrop that pushed a lot of warm lighting tones into the shed. This, in turn, with a directional light that came through the window, allowed for realistic lighting that relied on this rather than using fake lights. I did, however, add spotlights to certain areas of the scene for increased Roughness variation for certain renders, but still tried to follow through with where the main light sources were located.

A must with any sort of lighting in environments is that if you add a light into the scene, it should have a real item, like a bulb or candle, that the light is emitting from. Having a random light that is emitting from nothing will look incredibly strange.

I used DaVinci Resolve and Photoshop for my post-process settings for the images and for the video sequence that I exported from Unreal using the movie render queue. I did not want to completely overhaul the images inside of DaVinci as I wanted the environment renders to be real-time from the actual viewport, with only some added changes in post process.

Conclusion

All in all, this project took me a little over 5 months to create everything in the environment. The only props and textures not created in the scene were those of the megascan trees and scattered leaves. I learnt quite a few new techniques in this project, from polishing up my Substance 3D Painter skills with the realism style, to prop creation and technical art skills.

I want to say thank you to all my friends and peers for giving me feedback and helping me when I got stuck in areas of the project. My final advice for other 3D artists is to keep improving on your craft, get feedback on your work, and try new things. There are still things I could improve in this artwork, but it is necessary to find a stopping point.

I had planned to render the outside of the shed and create a good megascan landscape for the outside, but I was just overworking myself for something that was not the clear focus of my skills and the project. I'll look back on this project in a couple of months' time after my next one and see areas that I totally could have changed, but this is because I have grown as an artist.

I want to thank everyone for reading this article and hope it was of some help to you. Thank you to 80 Level for allowing me to show my work to a broader audience. If you want any help with your 3D artwork, you can always connect with me through LinkedIn, Email, or my Artstation, and I'd be happy to help.

Timothy Lee, 3D Environment Artist

Interview conducted by Emma Collins