Creating a Cartoonish Seaside Scene in UE4

Desmond Man talked about his UE4 Real-Time Games Environment Gone Fishin’ and stylized art in general.

Life Updates

After the ArtStation Ancient Civilizations interview with 80.lv, I had finished my time with Sumo Digital and their incredibly talented dev team working on Dead Island 2 and moved to Splash Damage in Bromley, United Kingdom, where I am currently working on Gears of War: Tactics. Other than that I’ve spent most of my time trying to learn new crafts outside of digital art such as music production and other small interests.

Gone Fishin’

Concept & Design

This was the second time I’ve tackled a concept by Elena Ilicheva. To be frank, I’m a fan of her art style and creative mind which I feel connects with my urge to make colorful and somewhat cute artwork. It all began with a concept dig I went on flying through ArtStation where I eventually stumbled onto Elena’s concept by coincidence.

I find that I often pick a concept based on first glance. I’m a picky artist when it comes to choosing what I want to build so I need to feel a strike of excitement from the initial staring at an image because I know that once I get that first strike of excitement, that single thrill will drive me to build the work.

In this case, Elena’s concept was a 2-time winner for me. It often starts with me trying to match the concept as much as possible but I feel that when you stare at a static image for days, weeks and months you kind of begin fantasizing about what it would look like from the other side, in different lighting states or in different scenarios and that’s where I experiment.

My first major change in the overall piece was the wooden boardwalks. I love Elena’s original setup and it works great in portrait but I’ve always been more of a landscape guy and sadly, I couldn’t stand how empty this piece looked when trying to fit it all into a landscape.

I trialed so many ideas to save the original design (small rock islands, floats, better placement of vegetation, etc.) but nothing seemed to fill this piece up without taking attention away from the main focus. This led me down the path of deviating from the original layout. I feel that the final layout added a nice little framing that cupped the house and actually made it feel cozier. The additional brown flat space took away some of the screen coverage from the ocean’s blueness, and it was a massive plus.

The next big thing that came from this was “what does this house look like from the other side” and that’s how I began envisioning the silhouette of the hut from another angle. I realized it looked a little dull. This brought me to have to think about adding something eye-catching to justify the wooden boards curving around the hut and eventually led me to add the telescope peaking out from the side window.

Most of my decisions to step away from the concept usually come from spontaneous moments where I feel something could be more vibrant, a little less dressed, smaller, cuter which I guess you would just call your artistic intuition eventually developed by every artist.

Although this project developed slowly it wasn’t all built and thought up in my head alone. For example, the duckling float came from a conversation I had with a fellow artist at work. I showed him a work-in-progress shot and we both agreed the float could have been cuter. Most of the best changes in my artwork came from the synergy of my effort and the great artists around me.

Asset Workflow

The process I generally go through when building my assets always start out in 3ds Max where I block out a mesh to correct metrics. I don’t bother going into fine modeling because I want to get this blockout into ZBrush as soon as possible.

After that, I aggressively sculpt into the blockout using DynaMesh to get an attractive silhouette. When working with stylized art I generally stay in DynaMesh through the entire sculpting phase while increasing resolution as the sculpt starts to take better shape. I feel I only ever drop out of DynaMesh and increase the subdivision levels when I want really fine surface details such as indifferences in smooth surfaces.

After the sculpt is done I bring my high-poly sculpt over to 3ds Max or Topogun (take your pick) and spend a little time retopologizing. After all the footwork on preparation up to texturing I bring my low-poly and high-poly meshes into Substance Painter for the initial bake.

Now, I generally use Substance Painter to set up some really quick generators driven by masks from my baked textures just to get a good starting point on the albedo.

Essentially, all the main PBR textures (Metallic, Roughness, Normal, etc.) that I want will be exported from here and are good to go for the engine. However, I then tend to put the albedo through an additional detail pass afterward.

I bring my low-poly model into 3D-Coat and import only the exported albedo & normal map. With these two maps now rendering on my model, I begin to do an additional hand-painting pass on the asset simply by color picking and repainting areas. All the albedo data rendered from Substance Painter is very linear and the gradients are too perfect for the style that I want so I generally break it all up by repainting gradients and also adding fine details as I progress.

You’re probably wondering why I  would bother bringing a texture from one painting program to another painting program. Why would you not just do it all within Substance Painter? The reason is that 3D-Coat has a much more intuitive brush flow system. If I were to create a hand-painted asset, the brush flow & opacity that you get in 3D-Coat feels almost identical to the brush layering you can achieve in Photoshop but across a 3D object whilst overcoming the issue of seams.

Afterward, I bring everything into UE4 and hook it all up, there’s not much special treatment here apart from the fact that I refeed my albedo map into the emissive slot by 40% boost to all of my models so my albedo stands strong and brings out that final vibrant cartoony touch.

As for the details on the Seagulls and Shark, I feel that the way an artist tackles stylized art is quite interesting. If you they do it without the restraint of a brief and it’s purely personal leisure then through the choice of silhouettes, colors, etc. you can really see how it resonates from the artist within.

I’ll be honest and admit that I’ve actually got quite an immature innocent but also dark humor when it comes to cartoons. I like things that look dopey and have a bend towards cuteness, I love cute things. The little “x” on the seagull butthole was just something I and another artist from work find hilariously funny and if it were up to me I’d put little x buttholes on everything.

The Shark was made following the concept but I did feel like it needed to be a little chubbier/rounder. The original Elena’s silhouette had a much sleeker feel to it or what I kept referring to in my head as the “Mustang bonnet type shape” where the upper half of it was flatter. It came across looking too “cool for school” to me which is why I decided to round it out. The same applied to the duckling float, in Elena’s version the duckling looks generally more elegant. Going back to what I said about the style resonating from the person within, I wanted a more derpy/dopey looking duckling so I started off with reference from the giant rubber duckling float by Florentijn Hofman. From there I started experimenting with ideas of turning its base into a doughnut type shape for comedic/cute purposes.

Wood Material

The wood material was built entirely within Substance Designer. I feel that approaching stylized substances are a real breath of fresh air because automatically you don’t need to go as in-depth into the graph as you would with a substance that you’re trying to recreate in realism. Most but not all stylized textures tend to strip a certain percentage of macro detail to tone down noise and to deliver a comfortable soft image to stare at.

The building structure of my woods graph isn’t complex, actually. It is entirely constructed of shape nodes & tile samplers utilizing Paraboloids as it’s main shape, scattered across and blended together at different heights vertically.

After that I added a lovely splosh of everything vibrant you could think of, this is my own personal preference when it comes to stylized art. Anything that can make you vomit rainbows and heart-shaped eye beams is overlaid on the colors that are actually ACCURATE to what wood looks like but blended at like 15% into it and that’s essentially how I like my Albedos. It adds this whimsical look, in my opinion.

If there was any single piece of advice I could give an artist on how to colorize your height data in Substance Designer for stylized art it would be that no matter what mask data you can extract out of your graph, as long as you layer them up without any insanely harsh blends, it’ll play a part in making your albedos look beautiful. This mentality comes directly from my skills with actually hand-painting textures, it’s all about the subtleties that tie it all together.

How I usually go about my texturing phase is quite simple. You start with a base color and you pipe your final height data through a gradient node and overlay it. This creates your very base and an easy mental leap away from the blank canvas obstacle.

After that I start to layer based on these masking principles:

1. Random Noise Nodes (Nothing with a lot of detail, plain blobby data)
2. Dirt (Crevices)
3. Highlights (Closest areas facing you)
4. Ambient Occlusion (Colorized with a hint of any color, never greyscale)
5. Direction extracted from Normal (R, G or B channels)
6. Hard edges (Extracted from Curvature Node)
7. Overlay Curvature Smooth on top at about 0.3 for an assisting “sharpened look”

As for everything else, I don’t tend to pay much attention to my roughness map unless metallic is being utilized. Reason being is because as earlier explained I like to run my Albedo map through the Emissive slot in UE4 at around 0.35-0.4 just so the Albedo colors are truer and less driven by the lighting in the scene. I personally find too much roughness on stylized things to break away from the cartoon look quite a bit. This again is just personal preference and there’re a lot of cases where not doing this method can have great results *cough* Overwatch *cough*.

As for the shader setup in UE4, there wasn’t anything special to it. I piped it into a height lerp blend which enables your vertex painting blends to be driven by a height texture. I ran painted wood/paint wooden board textures into it and used the height map data from my bare wood substance as a height mask which helped me fake the look of wooden chipping from the paint.

Ropes

Funnily enough, I actually custom placed the ropes all in 3ds Max before exporting them to UE4. Usually when working on a game that may require heavy re-use of a spline blueprint actor for ropes/chains/etc. it would make sense but I felt, in this case, I would only be using this blueprint actor for what originally was meant to be 4 times. It felt like the time spent building the spline actor vs. time spent placing them didn’t feel like a good trade-off so I decided to just DIY it this time around.

Lighting

There wasn’t really that much to lighting with this piece. I used the default UE4 lighting setup with a directional light, a skylight. The only additional work I did was activate Lighting Propagation Volumes, add a few Reflection Capture Spheres where it mattered most and tampered with Post Processing effects such as Lens effects, Exposures, Colour Correction & Global Ambient Occlusion values. There was a lot of faffing around with the Exponential Height Fog actor to get the atmosphere feeling not completely fake but also sold the idea of distance.

Feedback

I would say my biggest take away from this project is that I have begun to understand more about my artistic influences and decision making while developing an environment. It’s shown me a more personal theory on why it is that I like specific shapes, colors or themes and that’s always the most valuable lesson while trying to develop yourself as an artist: understanding what it is that you like and why it is that you like it.

I feel like my favorite part of this environment was probably the floating rubber ducky as I was quite shocked how accurate I got it to the vision I had in my head whilst most of it was off the grid where I stopped looking at reference and went with my imagination.

However, the biggest challenge to take away from this was probably trying to figure out how to fill out the scene without overdoing it with assets. I feel a house floating in the middle of the ocean is a lovely idea but when tackling it in 3D it really is a pain to generate an asset list to fill out an open deep ocean space other than with rocks, more islands, more floats or a half sunken ship at shallow water.

Desmond Man, Environment Artist at Splash Damage

Interview conducted by Kirill Tokarev

Sponsored