Knights at Chaumont

In this article we’re taking a trip to medieval France with Omar Souissi as we discover the process behind his scene, Knights at Chaumont, inspired by chivalric romance and historical management games. We delve into how Omar used modular kits to construct his castle, effectively blended external assets with his own, added variation when using tiling textures, created alternate lighting scenarios and much more!

Intro

Hello! I’m Omar Souissi, currently 3D Environment and Lighting Artist with around 3 years of professional experience working at Hutch Games on Top Drives, a mobile car racing game. I previously worked for Capsule Studio in Paris, a game trailer outsourcing studio. Before this, my first job was at Ubisoft on a little VR space simulation game called AGOS. I spend my work and personal project time building environments.

I lived the first half of my life in Morocco and the second half in Paris so far. My introduction to visual arts came rather late at 22 years old at a point where I couldn’t find any meaning or purpose in the biochemistry and business studies I was doing. The only thing I was waiting for everyday was free time in the evenings and late nights to spend exploring books, paintings and video games. 3D is a great way to see your dreams come true. It can make you see what you could only imagine while reading or make it finally possible to jump inside a painting or be the hero of a great adventure. I have always been fascinated by this medium. Right after graduating, I threw myself into it and I can’t imagine doing anything else with my life.

Composition

My main goal with this scene was to make it feel like a moment in a game. As a first-person RPG lover, I wanted to communicate the feeling of immersive exploration that comes from playing a character evolving in a wonderful universe, like in the elder scrolls series.

I divided the shot into four sections with the intent of inviting the player to the castle. A horse (number 1 in the underlying image) is placed on the foreground to invite the observer to take the spot of the player in this scene. The knight and his dog on the foreground (2) seem to stare at the player, pulling them into the action and interrogating their arrival to the castle. The path leads straight to the castle, indicating the ultimate objective of the player.

We can imagine the possibility of wandering around the tents area (2) to check what’s happening and talk to the knights as an optional interaction before reaching the main objective that is the castle. The player could be another knight, a noble coming to assist the tournament, a peasant carrying provisions for whoever is currently taking residence in the castle, or anyone else. It doesn’t matter. This bit is left for interpretation. What’s important is to feel immersed in the scene. At last, comes the background vista with the sky and mountains to make the world feel grand, free and vast.

Colour and Light

Regarding the use of colour and light, the idea was to make sure that the castle is the main subject of the scene. It had to stand out compared to the rest and appear grand, idealized and romanticized. Its central framing, unusually clean look, bright bricks and blue shiny roofs felt perfect to make it special and set it apart from the normality of its surroundings. The more peculiar something is compared to its surroundings in terms of colours, textures and shapes, the easier it is to make it stand out and attract the eye. The tents and sky do a good job at framing the castle and path towards it. The tents are quite bright but their positioning on the sides of the frame and the fact that they are more common elements than the crazy castle prevents them from stealing too much attention. The weather is nice, it’s a good day and people are gathering under a positive light. There are clouds in the sky, but nothing seems to announce any bad event. They are just preventing the sky from feeling too bland. The sun is the main light source and hits the castle diagonally from top left to bottom right. This allows for having some light vs shadow contrast on the towers, a cast shadow on the entrance and a lit vs shadowed side, giving the castle more volume and depth than the rest of the scene.

The scene is quite busy but that doesn’t steal the attention from the castle. There are lots of knights, tents, small props and vegetation on the foreground, but all their colours are close in terms of temperature and values compared to the castle with its unique bright walls. The foreground and midground details are a break from the hero castle. The player’s eyes can wander a bit and notice some things, like those two servants carrying equipment, and the knight looking at another castle in the background behind them. Of course, these are not things most viewers will notice, but they all play a part in the big picture, transmitting this feeling of a busy social moment.

Another detail: the knights are divided into two teams: red on one side and blue on the other (noticed the colours of the tents and flags?). Those same red and blue are found on the castle (roof tiles, window frames, flags). It makes the castle act as the element that gathers the two opposite sides together for this event that appears to be the preparation for a tournament.

Goals For The Scene

I had a few main objectives that defined my approach and workflows for this scene. Those were:

Creating a scene that feels like a moment in a video game, which forced a first-person perspective point of view from the beginning and thus influenced the composition choices.
I wanted three nice shots showcasing the castle in different lighting scenarios along with a video.
Artistic freedom of interpretation and romance are important to me. I took liberal decisions when it comes to historical and physical accuracy of the scene. I did not want to spend countless hours researching and making sure each piece of grass, cloth, or brick is from the exact right historical period or geographical location. A higher priority was given to artistic experimentation with shapes, colours and lights.
Another objective was producing the castle in a way that would allow me to sell it in online marketplaces. I never sold anything online before and saw it as good training. Knowing people might be using your assets makes you more cautious about making your stuff simple, clean and easy to use. It increased my quality standards and made me expect more from myself. All the castle’s textures and models had to be fully produced by my own means, without using online resources that would make my work illegal to sell, like Megascans and such. I didn’t apply this logic to the other elements of the scene as I was only intending on selling the castle.
As always, learning new things and overcoming my current skills by pushing up the quality bar was the most obvious objective. I think this is my best work so far and it took more effort, research, observation and attention to details than all my other works. It’s a good feeling to aim for when making art!
Regarding timing, I didn’t give myself a deadline, because I really wanted to push the work until feeling like I couldn’t do more and had to turn the page. It was a personal project after all, and I’m already used to deal with deadlines and plannings at work.

Inspirations for Knights at Chaumont

I have a passion for history and romanticism. Among my favourite books are Ernst Junger’s Storm of Steel, Joseph Conrad’s Heart of Darkness and Walter Scott’s Ivanhoe. When it comes to cinema, I love Gladiator, The Last Samurai, 1917, Dunkirk, Kingdom of Heaven, and Lawrence of Arabia to name a few. I also grew up with games like Age of Empires, Medieval II Total War, Crusader Kings, World of Warcraft, Assassin’s Creed and the Witcher III. My passion for chivalric romance and historical management games motivates me to construct idealized buildings and scenes from a fantasized past.

The idea of making this castle scene came from a trip along the Loire, a French river. Its banks are full of interesting historical castles, most of them displaying a very particular Renaissance architectural style that became part of the visual identity of the country. One of the castles specifically stood out to me: the castle of Chaumont (top left in underlying image). The impressively thick cylindrical towers which frame the main entrance of this medium sized castle make it look as if someone tried to mix and match elements that are not scaled in proportion to each other but managed to make it work perfectly well, better than if the proportions were more conventional. It transmits this image of strength and pride that comes from the imposing figure of a castle and its towers, with a twist of confident craziness that comes from the fact that the castle is designed in a fun way, and made primarily to host events, and probably the craziest parties as well. It says, “Why bother being huge, damaged and intimidating when you can be unique and elegant instead”, and that just sold it to me, along with its intriguing CGI look.

Lighting

The three scenes all have a very simple lighting, mostly relying on one main directional light and some supporting spot and point lights. Then it’s all about playing around with the different settings of the different lighting elements. Everything is set to moveable and the scene is fully dynamic.

I first start by placing a skybox, skylight, main directional light and exponential height fog. I’m using one of the skyboxes from the Velarion’s packs. The exponential height fog and skylight do a good job at lightening shadows and give a layer of depth to the scene.
Once I’m happy with the settings of the elements mentioned above, I add points and spotlights where the shadows are too dark. I sometimes also add a non-shadow casting directional light facing the main one with an extremely low intensity when the shadow side of the scene is too dark. The downfall of this alternative is that it can really flatten some areas and make them boring in terms of lighting. Playing with spots and point light, adding light sources, colour variation in lights and textures or layers of objects can sometimes help. I also sometimes turn on RTXGI and set it to brute force in the post process settings. It can give a nice global illumination, but it tends to hit the framerate very hard. This setting is a bit of a hit or miss for me, it sometimes makes a scene more interesting, and sometimes kills the light vs shadow contrasts. In the end, it’s important to be aware of the different options and use the ones that are right for each case.
At last, comes the post process phase, where I just tweak to my liking the post process values in the settings of the cinecamera element. I also always work by viewing my scene through the camera that will take the shots. This way I’m sure that I’m looking at the right aspect ratio. You can position it and lock its position to avoid moving it around by mistake.

Here are a few links that really helped with the lighting of this scene:

Modular Pieces

I needed something quick and easy to make and iterate on, so I could have time to focus on other aspects of the scenes such as composition, lighting, shaders, and materials without feeling completely overwhelmed by having to make hundreds of modular pieces. I didn’t want to tackle interiors for this project as my objective was to create outdoor landscapes showcasing the castle and work on my lighting and composition skills.

The castle set needed to include the lowest number of pieces, be simple and allow for the fastest kind of assembling. Pieces can just be merged into each other, there is no snapping mechanic. I split the castle into big chunks, mostly entire buildings and started assembling them and trying different combinations. I’d first only make three assets. A big, medium and small one. I would then see how these can be assembled and conclude on what the next piece should be, then try different combinations with those four assets and so on until I feel like I have enough assets to build a few interesting castle variations. This production routine allows for making the minimum amount of assets and not get drowned in too much planning and thinking to end up with a quantity/quality/timing riddle. None of the assets or their components are baked. Even the ornaments are simple geometry with a tiling texture and detail normal on them. The workflow doesn’t allow for the highest possible visual quality per individual piece, but it’s very versatile and I prioritized the big picture and iteration potential over the extreme polishing of each individual asset. With my full-time job, I could not afford having each prop taking weeks to make with their own individual texture. I love vast landscapes and wanted to finish the artwork in less than half a year.

Texturing

All the assets in the scene are textured using tiling textures made in Substance Designer with the help of decals and vertex painting to introduce variation. UV seams are hidden through simply having a material that tiles subtly where the seam is, or by having the seam in a non-visible part of the model. It’s also possible to hide seams with decals, triplanar mapping and many techniques. This blog article from Chris Radsby goes through some good ways to hide UV seams.

The decal masks are made in Substance Designer in a very simple way, mostly just by using grunge maps, mixing and masking them around, and tweaking their levels. I also grabbed a few alphas online (Megascans, textures.com etc.) and reworked them in Photoshop to fit my purpose (changing the levels, erasing and adding parts here and there). Damages, stains, leaks, erosion are the keywords I used to find anything useful for this case. A basic decal shader is easy to make, you can just check this quick video.

Those details really help making feel the castle feel more grounded and integrated into the scene

The other texture variation technique used is height blended vertex painting. You can find a simple tutorial about it here. It’s a great way to blend different textures together and get some variation going on in your tiling surfaces.

Materials and Parameters

Having the maximum amount of control within the engine is what makes me feel the most comfortable. I have this logic in mind whenever I start thinking of making an environment. Parameters such as texture tiling, brightness, hue and saturation are very important to me. They allow instant iteration with immediate results in the game engine without the friction of having to modify textures on other software and reimport them to test every single tweak. I believe that editing real time scenes should be as ergonomic as editing pictures with Lightroom. It saves a huge amount of time when it comes to choosing which colours and values are best for each element and adjust things as you add elements to a scene. Having control over the roughness, metallic and normal map is also essential. I mechanically added a detail normal map to my shader, as it adds a lot of believability and an interesting layer of detail to textures when viewing them from up close. I also use parallax occlusion mapping (POM) for certain cases like brick walls and roof tiles to add some depth to the textures. An AO intensity parameter is also useful in case the scene uses baked lighting, but you can also overlay the AO map on top of your albedo one if you feel like you need more occlusion in a dynamic lighting scenario. It’s not physically accurate and not always advised but it sometimes does the job for me. I did not use any baked lighting on my final renders, but I experimented with it and made lightmaps for the UV2 channel of the castle’s models.

The above image shows the most basic parameters I like to start with. After this, depending on the need for each asset, I duplicate this master shader and plug in new features such as vertex paint, parallax occlusion mapping or anything I’d like to add to this basic setup. I usually work with this basic setup applied to most of my elements and use other master shaders for elements with extra features. The bricks for example use another master shader that incorporates a POM feature and two layers of vertex painting for texture blending. I avoid making a gigantic master shader with every possible option and tick boxes. I find that it quickly goes out of control in terms of size and readability. I always try to minimize technical friction and prioritize focusing on creativity and ergonomics. All elements are using material instances. I never plug master materials directly into anything. This tutorial is a good introduction to master materials.

Above, the effect of POM on the roof tiles texture. You can find out more about it here.

External Assets

Every asset besides the castle isn’t fully made by me. The sky, vegetation, foreground props and characters were all acquired in different online resources and repurposed to fit in the scene.

Working with external assets can make a project very messy when importing everything directly from online marketplaces into your project. I used another UE4 project with the same engine version (4.25) to download assets from the UE marketplace or import them from other online sources. Once I isolate the parts of a pack that I need, I apply my own master shader to the meshes, change the texture maps to fit my own workflow (for example, if a pack uses an MRAO map instead of my ORMH one, I repack the texture maps, so they fit my format) and then migrate what I need to my main project. This way, new assets integrate perfectly into my project’s architecture and organization, with my own shaders and parameters. It takes a little bit of time, but the clarity and organization are worth it and end up saving time and resources. It’s also much more optimized than throwing stuff from everywhere into your scene and see your draw calls skyrocket unnecessarily. Keeping control and knowing what’s in your project is great for working with a peace of mind and not let creativity being hindered by messy folders, different naming conventions and a terrifying FPS count… and eventually having to fix it all in the end!

Video Sequences

The workflow I used for recording the short video sequences is a simple cinecamera attached to a sequence. I had three different levels, one for each lighting situation, and a cinecamera and sequence as well for each. The first shot’s sequence has a camera shake function attached to it. I would end up with three short sequences that I would put together and edit on DaVinci Resolve. It’s a great video editing software that you can get for free on their website.

Above, a screenshot of what the sequencer looks like for the first scene. The camera just slightly moves forward and has a camera shake effect attached. On the other scenes, it’s the same but without the camera shake as it felt better to me without it. You can find more on Camera Shake here as well as the UE4 to DaVinci workflow here.

Areas Looking To Grow As An Artist

Among my future objectives, I would like to improve in individual prop creation. I am very drawn and obsessed by the big picture that is being served by composition and lighting and give it more time and importance than the polishing of individual elements. I will keep making whole scenes but will give more love to smaller details in my next projects and test myself more with close shots. I’d also like to dive into vegetation and Moorish architecture. A recent trip to Andalusia (southern Spain) reminded me how fond I am of this type of buildings. Here are some examples:

Inspiration

My first go to inspiration resource is art history documentation, books, and some YouTube channels that are fascinating and never fail to give me new ideas. This one is a good example. I also keep a list of paintings and favourite artists I often go back to. I have a Miro board of my favourite pictures and spend some time updating it every week. I try to keep the number of images under 300 so it really forces me to extract what I like the most. If cruising through the images gives me goose bumps or strong feelings, I know I’m looking at what I’m made for! Comics can also be a good source of inspiration, especially for the story driven aspect of their images. You can stumble upon a lot of interesting series of images that tell a story without using words. This can be inspiring when imagining a sequence of dioramas that would follow each other.

I like to stay away from Artstation when reference hunting, because I too often see the same compositions, styles and subjects on the trending or community pages. I tend to find more original stuff away from screens, like in books, museums or simply walking outside and traveling. Digital art feels to me like a watered-down copy of traditional art, the real world, or photography. I’m not a fan of taking it as main inspiration. There’s something raw, unique and mystical about art made without the support of a GPU’s computing power and I never had this feeling from digital mediums. Having said that however, I still have a few lists of Artstation posts that really reached me and sometimes add them to my ref boards. No need to be too close-minded.

One of my favourite websites is Pinterest. Its algorithm proposes similar visual picks to the ones you chose. You can also input an image you like and discover related propositions. It’s great when you already know which path you want to go and need more references, but you can also wander around, from subject to subject and discover amazing new things.

Feedback

I usually try to seek feedback on every aspect of my art, because I think that everything can always be improved. Currently, the most needed area would be the making of individual props. Attention to small details and storytelling in single objects is what I’ve been looking at lately compared to more high-level and big picture things. I like asking my colleagues and ex colleagues for feedback as they’re the easiest and closest people to reach first. Discord groups such as EXP are also fantastic places to ask for new eyes and opinions. A very important one is to also ask non artists. When I ask my partner about what she thinks of my 3D brick and her immediate reaction is “what is this?” or “it looks like some kind of weird packaging”, this is where I most strongly feel like I did something wrong. After all, we create art for an audience that is not supposed to know art rules or be experienced in the discipline. The feedback from people outside our profession and who are not trying to be nice is the most indicative to me. They might not know what makes a piece of art feel off, but their instincts and feelings are your greatest indicators. My partner can tell me my rocks look like plastic, but it’s my job to know that it’s because of the roughness value or the levels of detail, or both, or something else.

Additional Advice

Making visual art is like learning to play a music instrument. You get better by working an hour a day than 8 hours once a week. Quantity matters, but I find frequency much more important. Sleeping on your ideas really improves them from day to day and probably develops your artistic brain in the best way.

Diving in the world of your current projects by associating your free time to their theme is also great conditioning. Watching movies set in the medieval or renaissance era, going to museums and visiting monuments related to chivalry and the Middle Ages, reading about it, watching documentaries and other things really helped with creativity and art production flow with my castle scene. Linking your leisure time to the theme of your current work can indirectly improve it and bring interesting knowledge and ideas.

I also like to draw things related to the subject I’m working on. 3D is a rather stiff medium compared to pulling a notebook and pencil and just doodling some shapes, silhouettes, ornaments and elements related to your artwork. It often makes me notice, invent and exaggerate interesting shapes and details that do not come naturally when working with a 3D modelling toolkit. I then try to apply what I find out in my notebook into the 3D models and texture I’m making. When it’s possible, it can give nice surprises!

When doing something for the first time and learning new techniques and workflows, there will be a lot of times when something unexpectedly turns out to be more time consuming and difficult. In those situations, I prefer facing difficulty instead of trying to find easy alternatives that result in a negative impact on work quality.

Tackling vast and detail rich landscapes can feel intimidating, but it’s not as hard as it can seem. Choosing to make only a part of the scene and just import and recondition the rest allows for focusing on details and the big picture at the same time.

Future Work

I just recently came back from an amazing trip across southern Spain in the region called Andalusia. I found its history and architecture very inspiring, took a bunch of pictures and will try to make a modular set inspired by Moorish architecture, this time with interiors, and palm trees!

Outro

Thanks a lot for reading my article! I’m always happy to exchange about art on Artstation, Discord and Twitter so feel free to drop me a DM if you feel like it!