Jordan Smee explains how the Lost City scene was blocked out in Maya and textured in Substance Painter, shares some great tips on creating stylized foliage, and talks about the importance of tweaking lighting settings slowly and patiently as even a minor change can make a huge difference.
In case you missed it
You may find these articles interesting
My name is Jordan Smee. I’m a Technical Artist based in the UK. I studied Games Development at South Essex College of Further and Higher Education as a college course and then again as a university degree, I originally had no desire to do university at all, but during college, I found my passion for games and alongside the support of the head of the university degree I ended up not only going on to do a university degree but only applying to the course at the university where I had done my college course, which in hindsight was a very weird move but it seems to have worked out so far, so I say it was okay. I was very fortunate with my university course that I got the full support of the head of the degree to pursue a lot of the more ambitious ideas I had at the time which allowed me to push my skills past where I may have otherwise gotten to. Alongside this, being able to go to Bossa Studios alongside events like Vertex, I was able to meet a series of industry professionals who helped guide a lot of my earliest work and pushed me in the right direction throughout my university degree.
Now within my professional career, I’ve worked on four different projects, firstly with Beyond Skyrim as a 3D Environment Artist, then with Black Clover Games again as an Environment Artist. After that, I joined Pineapple Studios as their sole Technical Artist at the tail end of 2020, and now I'm with Silver Rain Games as one of their Technical Artists. I find myself very fortunate to be in a position I love with a company and colleagues who make me feel accepted and that the work I do is valuable and appreciated.
Working on the Lost City
I’d always wanted to do a stylized scene even during my university course; however, I’d never strongly considered it as I always had the idea that you needed to hand paint everything. However, during the time I was making my previous project, I’d seen work posted by amazing artists like Jasmin Habezai-Fekri and it reignited my want to make a stylized scene, and so shortly after I was done with Journey’s Respite I went on the hunt for something I wanted to make and I ended up finding the Lost City concept by Elodie Mondoloni. I was really inspired by it because of just how interesting it looked, the colors within the scene, and funnily enough the little birds that are scattered throughout the piece. Plus, it had a lot of elements that would allow me to put my technical skills to use and to play around with foliage which was something I’d always wanted to do but had struggled with in the past.
Once I’d got my initial inspiration and idea, I then went on to get some more references from other places mainly for surrounding the water, foliage, and lighting as those were the elements for me that added a lot to the scene and if I got any of them wrong I knew it could knock the scene off completely. For references, I broke them down into my key and lighting-based references, something I’d learned during my short mentorship with Kieran Goodson.
It was also around this time that I got into contact with apaleblueeye who went on to do the music for the project, and so I wanted to make some form of narrative that would help support the music and give some ideas for how the music should feel. The story that I ended up creating was surrounding an explorer who had gotten lost in a sheet of thick fog, ending up piercing through the other side to discover a city long lost to time, one that was filled with awe and trepidation, and that was what ended up being the inspiration behind the scene alongside the music.
When approaching this project, the first thing I needed to establish was a scale for the scene and an actor for reference, so that I could easily get my proportions done quickly and then create assets that would be the right scale. This stage didn’t take too long and was mainly a case of getting the base character from Unreal Engine and ensuring the perspective was right so that I could reference it quickly. I have a bad habit of eyeballing a lot of things and so the scale for this was probably not the best, but it ended up producing the results I wanted and I made sure that I later took a picture of the base character with the models again to ensure the perspective and scale was correct to my eye.
Here is the super simple scale reference:
Usually, when I start a project I make the basic assets in as simple a form as possible so that I can get into the engine and get a blockout setup as soon as possible as well as to get my main camera angle and work on the base lighting as well. These aspects are for me the most important because figuring out your chosen frame and how you want to guide the observer with light and the environment helps to influence every other part of a scene. For this particular project, because I knew I wanted to make a complete adaptation of the concept into 3D, I had all of the assets I would need already present and could easily reference straight from the concept itself, this made the process a lot easier to some degree as it was a straight adaptation but it did mean that the use of other references for certain assets wasn’t always as useful.
I used Maya for this project as I still had access to my student license as I had only graduated in June of 2020, so rapidly creating blockout assets was very easy. What was also very useful with this project was due to the stylization of it and the large beveled corners I was able to adapt my blockouts very quickly and so I could get a clear image of my piece from just the blockouts alone.
For this particular scene I added in a terrain plane in the very beginning as it made the blockout process a lot easier than attempting to visualize the scene without it and making those adjustments based on perception alone, as in the concept the ground plane and water really helps to define the grounding of the assets, so having that in and soon after the water plane helped to get all of the proportions and position much easier. From there adjusting the light enough to match the concept was simple, but it gave me a good impression of whether my assets were big enough and were the right shapes, and gave the right initial lines of interest to guide the viewer.
This project was the first time I had made foliage myself and it was one of my main goals to create foliage for this scene as I was unfamiliar with the process. So for this part of the project, the first thing I did was grab as much blog/website/video reference of how other people made their foliage and then use that to help guide my own line of thinking when it came to it.
Kids With Sticks' blog post on how they create Ghibli-esque stylized art for their game was a large source of information, not only in their own process of making their foliage but many of the links they provided led to other useful resources, alongside that, Romain Durand’s post on stylized leaves was very helpful in allowing me to figure out how to create bushes and other similar foliage quickly.
With this information in hand, I first went on to create a set of basic bushes and grass shapes that I could throw around the scene, these also happened to stick as after I refined the textures achieving different looks became a quick process. For now, the bushes used a white texture that I then coloured inside of Unreal Engine by lerping between a dark and light colour and using the white texture as an alpha, I also then used the texture as a mask so I didn’t end up with large black segments in the bushes. As can be seen, the first attempts were okay, but they weren’t great, and this took a series of iterations to get right. I also realized later in the process that the mesh distance fields I had added were also messing with the look of the plants and this took some finagling to get right.
The next step for me at this point was to get some motion into the foliage and again it was another trip into my references to see what kinds of motion they had created and how I could do that in Unreal Engine. In the end, I settled on using two different sets of Panning Noise to generate World Position Offset and used an alpha to determine where on the mesh that offset could influence, because the one thing I noticed on my first attempts was that all parts of the foliage moved to the point that large "gusts" would move the mesh itself, instead of just the upper sections. To do this, I had to change how I approached my grass texture, and all the other foliage textures as well.
I ended up doing channel packing, channel packing is the process by which you store information in specific color channels i.e. R, G, B, A (if you’re using an alpha), and then when in Unreal Engine, using the different channel to gain access to that information, in my case, I stored the grass mask in the green channel and the influence gradient in the red channel leaving the blue and alpha empty as I didn’t need them. This allowed me to use one texture to determine a couple of different things without having to create a bunch of textures for those same purposes. What was also useful was I could invert specific channels to give even more uses, something I did with the red channel specifically as I used it to gradient my light and dark colors for the grass and other foliage. I also ended up putting multiple grass textures in a single texture in a 2x2 grid so that I could use fewer textures but still get the variation, as different grass meshes were matched to different parts of the grid.
Texturing for this project was very a slightly different process than usual as I’d never done stylized work before, and so again like most of my steps I looked into other artists' approaches to texturing a hand-painted look in Substance Painter, and I quickly found that it was dominated by artists who would hand-paint almost everything. Now I’m admittedly not good at painting and so going for a true-to-form hand-painted style wouldn’t have worked for me, and so instead I attempted to create my own smart materials that could mimic the different keystrokes and brushes I identified as key to the texture of the object. One major reference was a video by Stylized Station where they investigate creating Ghibli-esque textures in Substance Painter and that heavily influenced how I added details and brush strokes.
In the image below, you’ll see some of my ideas for how to approach the objects, a lot of these points put forward a very basic way of going about them with some of the quick ideas I had at the time. However, I’d say for anyone who’s referencing from a concept, being able to identify the things that are key materially in an object is also a very important skill because if, like me, you can’t draw too well, being able to understand the shapes you want to replicate can be useful when trying to make a smart material or a mask that will accomplish the same thing.
I also got myself a color palette of each of the different elements that I needed so that I could ensure everything was as close as possible to the concept during texturing and when everything went into Unreal Engine as well.
Usually, when I’m in Substance Painter, I will have my reference on my other monitor alongside several material references so that I can do smaller details and get a more realistic look. However, with this project, stylized work is so vast in style and can change so much it was hard to find a lot of consistent references, so I changed tactics and just tried to get as close as possible with what I could see.
Once I was in Substance, I started working in the Base Colour mode almost exclusively which was something I’d learned with Skyrim Beyond as they had to limit their texture scope due to the engine limitations of Skyrim, but it allowed them to get almost all of their details and color information into one channel, for me it allowed me to get all the colors right and where I want them from the get-go. From here I go into the material view to sort out things like the metallic and roughness values, which in a lot of cases can drastically change many of the colours that I’ve worked to create.
Each asset was textured in a similar way, firstly, I’d start off with the base layer/main layer, whatever the main color of that object needed to be. In the case of the towers, it was always a grey color so I created a smart material for my stylized asset which I could then change and use again for each successive layer and material. This just simplified the process down so that I didn’t have to continue to make new versions of a material that I had already made. The base material itself was based on a stylized material created by 3DEX who’s work on stylized objects was a big inspiration, so it starts off with a base color which can be swapped out for metal color, it adds in a color to the edges, cavity, and a couple of sets of color variation and then I tint the ambient occlusion and add some baked stylized lighting and edge highlights.
I’d opted to not add any additional texture details and keep these simple as they’re mainly focused on getting the main colors correct. For things like the paint splotches, etc. or additions on top of the main color, I sorted them into a larger folder and usually added them on top of the main stylized material. This allowed me to sort out all the main shapes before slowly refining down the process to get smaller details in or add more shapes and designs into the objects themselves.
Assembling the Scene
Assembling the final scene was a bit of a weird one for me, as actually putting the scene together, finishing the lighting, post-processing and VFX took me all of about a week once I had the scene in a place that I liked it. You can probably see this quite clearly in the gif below where many large changes occur quite rapidly and then some other smaller less noticeable changes follow suit as I adjusted the scene.
Once the blockout was in place and I had swapped out those assets for their mid-poly equivalents, additional textures then went in alongside the textures that had already been going in already to test their balance with the lighting and ensuring that it remained consistent with what I was seeing to UE4. This meant that all the assets then sort of got textured in almost complete unison.
From here I refined the lighting down with some advice I got from a video by Carlos Perfume on the Stylized Station channel, who did an excellent video on how he made this scene in UE4. I followed some of what he said to add to my lighting and then pushed it to align with the concept more closely. At some point, I added a pinkish hue which ended up suiting the scene, and some point lights to brighten some of the foliage which was too dark without it.
It was also at this point I made some minor material changes as well, like adding a blue hue to some of the foliage and changing some of the plant textures to add some more variety to the overall look of the foliage itself. I also added some additional plants, the small white and blue flowers that can be seen were an almost complete last-minute decision that I thought would look nice to some additional shades of color to the mainly greener areas, as they ended up just blending together which can be fine if you’re looking at a massive field of grass but not isolated pockets.
When it came to composition, my thought process was still very much emulation, I wanted to ensure I was as close as possible to the original concept and so from the earliest stages of the project the camera position was something I focused on intently. It’s actually the one thing that probably changed the most throughout the project consistently as I would get an angle, like it for a day or two, and then change it because looking back at the concept something in the background or foreground looked wrong, or the shot was too wide, or too tall, etc. But the other thing I needed to do was ensure that there was a focal point that everything was driven towards, in the concept, the focal point is the central building, all the lines of interest point towards it, and going by the rule of thirds it hits the most points. So my intent was to ensure I got those lines right and add a few of my own to subtly drive the point across, the wind trails were a part of that effort whether they were a subtle push is subjective but I like to think they were subtle enough.
Lighting and Rendering
In terms of the lighting for the scene, it was simple, I usually go for a two-light system where you have two Directional Lights pointing in opposite directions. One acting as your main light, and the other as a fill that doesn’t cast shadows. But with this project, I decided to go with a single Directional Light and a Skylight as it filled the ambient light position much better and reacted to the skybox in a way that enabled more natural lighting. Additionally, you can set up a Skylight to cast shadows and these will darken only the parts of the world where those shadows would be expected and so it removes the chance of getting doubled up shadows.
For smaller lights, I used two small point lights which I’d mentioned previously to brighten the foliage in the darker areas of the scene, and then four rectangle lights which add a rim light effect to the metal piping that leads towards the main building. This was done in order to distinguish the pipes from the other background elements as they weren’t standing out as much as they seemed to in the original concept but it was also to provide more lines of focus, and a clearer direction as well.
Post-processing is in my opinion the best part, it’s the time where for me you can dramatically change the atmosphere and mood of a scene, adding or taking away values to add warmth or cold to a scene, or even making it into a place of a thousand bloom effects. For the scene, I’d added a lot of warmer light with orange or yellow hues, and so the first thing I did was balance that in the temp section of the White Balance bringing that back down so that the warm and cold values were fairly even, I think this is a fairly important thing for people to understand as adding too much of a certain temperature can really affect a scene and so learning how to balance that and still allow for both warmth and cold to shine is very important.
From here were the global values, I increased the contrast by 0.1, that might not sound like a lot but it added a lot to the scene and it might just be me but without that small change everything just looks different. The offset was another one of the values I changed with that being brought down to -0.019, this for me made a lot of difference as it added a lot of depth to the scene, and helped to define a lot of the shadows. Below is the comparison between the scene with and without that small change, and it really was small as I tried to go lower and it did not work and so it was a lot of very minor tweaking before I got to where I wanted it. This is another change that is incredibly minute.
After that, I adjusted the shadows, midtones, and highlights and a lot of the changes to them were mainly surrounding their contrast, gamma, and saturation levels. Shadows were made a lot more contrasting and had their gamma increased to reduce the blue shade they had taken on in the beginning, midtones were slightly more saturated and had their contrast values very minutely altered up. Highlights were probably the most edited, as they had their saturation increased, their contrast reduced, and their gamma reduced.
It was at this point that I realized that everything was a bit too contrasted it was almost exactly where I wanted it, I just needed to reign it in slightly, and like with the lighting I needed everything to balance out, I did this with the toe setting in the Film section as I brought this down to 0.42 and it helped to round out the quite intense contrast values that were happening. Below is a comparison of the scene with and without and post-processing.
Motivation was probably the hardest part about this project, and I think it’s something that every artist or game developer experiences at one point or another. At the time that I was working on this project, I’d graduated less than a year prior and hadn’t found a job, and so this project was a means of broadcasting my skills. I’d also just come off of my previous portfolio work and so the initial stages of this project went by quite quickly because I was running off of the motivation and energy of that project and the positive reception it was getting. But after the initial momentum fell away I found it quite hard to keep going, and so I actually decided to just take a long break away and remove myself from the project completely and do something unrelated until I wanted to come back. So if anyone is facing something similar, it’s definitely worth taking time away and completely removing yourself from your project for as long as you need, it’s definitely been harder to do that in the current world situation but for me, it definitely helped as after a while I came back to it and felt that urge to create again and within the week the project was done.
For me, stylized work was very new and so I think a lot of excitement and urge to create it was due to it being such a new medium for me, but it was challenging. I’d suggest for anyone new who wants to try and make a stylized piece, I would recommend you do your research first. Into what other people have done and how they’ve done it and how they could influence your work and style, as the one thing I’ve noticed most of all is that almost every stylized artist has a different and defined style. One of my main influences and someone who was very helpful when I asked for their advice was Courtney Fay, her stylized work is amazing but very different from what I was doing as she does everything hand-painted, but her understanding of colour theory and the way you can create different values even within the same hue of colour was very impactful on my piece and I tried to apply that knowledge where I could. So alongside researching other work, also just ask people for advice and feedback, sometimes the best way to learn is to ask others for their ideas as well because in many cases they’ll come to you with something you haven’t considered yet or may not have noticed.
And if you want some other places to help guide your stylized work then there are some great places to look:
- Stylized Station whom I’ve mentioned before are a great YouTube channel that focuses on stylized work, they always have different artists from within the industry, who do essentially either mini breakdowns of their work or cover important topics.
- Prismatic Dev is another person who has some amazing content on UE4 stylization, plus he’s just very entertaining which is always helpful as well.
- ArtStation can also be very useful although it is a lot harder to find the right kinds of content and tutorials there.
- Also 80 Level, several of my blog references came from here and it’s also a great platform to use as a basis and then use the links in the blogs that fit closely with your project and go from there.
Plus, they were also kind enough to let me write this for everyone to enjoy, so I am honored and humbled that I have been given this opportunity. So, thank you to 80 Level and to yourself, the reader, and I hope anyone who’s read this comes away from this with something productive and or has learned something.