Enguerrand Michelin discussed his last New3dge school project, the Beyond the Line game, where he worked on the environment, shared his approach to architectural details and talked about the biggest challenges in the production.
Since high school, I was interested in 3D, where I started to learn by myself on Blender. At first, it was just a side passion, so after high school, I began to work in a different environment: Cinema and Theatre as a light technician. These were really good years! But after a few times, I realize that I want to turn my passion into my job, and work on CG every day! From this point, I start a professional career and ran back into studies at New3dge, working beside to pay these.
Rough times for 5 years, but it's worth it !!!
New3dge is a Game Art & VFX school located in Paris and directed by a wonderful team. It grew a lot in 5 years due to its success, as the school now delivers a national degree, but it always kept a warm and family mood.
Studies for 5 years include:
- A first preparatory year to learn traditional art and technics such as drawing, painting, sculpting.
- Two general years to learn 3D basics.
- Two years where you choose to specialize in Video Games, 3D Animation or VFX.
During these studies, you learn all the artistic and technical skills needed to follow a production pipeline in the Game Industry. Studies follow a two weeks session schedule with teachers specialized in different skills.
The last year is dedicated to a group project. Here you have to create a game, thinking first about the artistic direction, which has to fit in gameplay and a story you imagine. Of course, you also have to deal with production and technical constraints such as deadlines, optimization. Also, and it's maybe one of the most important moments, you have to specialize yourself and take responsibility for a whole part of the project. Many peoples already know what they want to do, but for a student who just wants to do one thing - making video games, the thing is more difficult. I
remember the rough time I had when I took the environment artist position and let the character creation to my fellows. But of course, it was the best idea and I really enjoyed myself on this.
A Team Project
Beyond The Line is our degree's project, realized by Johann Ehwalt and Gabriel Sessekouoh. It's an Action/Infiltration/RPG game that takes place in an alternative reality, where the Cold War has continued to our present.
The city of Paris is split into two blocks, and the river "La Seine" draws the frontier between the East Riverside Communist and the West Riverside Capitalist. The player embodies a member of a rebel organization, who wants to free the city from these two dystopian systems.
The story comes from an old scenario I wrote as a web series. I first present it to some classmate, (as it was: rough and incomplete) but together, we rapidly saw how we could turn this mess into a game prototype.
And then we ran into it.
We had to create a piece of the game (almost a prototype, actually) with some characters playable in the Unreal Engine. The artistry was our main goal, and we decided to circle it focusing the story with the wish to create a game from a universe and not the opposite.
In order to avoid the whole process of creating an artistic direction (and because we are mainly 3D artists and not art directors), we were highly inspired by the art of Dishonored and then later, by the art of Wolfsentein, concerning the Hard Surface.
Gameplay was a huge problem because, at New3dge, we don't have any game design or programming lesson, so we thought about to reduce as much as possible but still make it present. We first plan two different missions, one for each riverside, but we soon realized that this was way too much work for only three people in a year. Then we rapidly reduce the side to one mission focused on the Communist riverside.
The main challenge was to include the player, threw a playable character in a dystopian world. Characters were my teammates' tasks, and I was in charge of the Environment.
We first started our research by a full-week brainstorming, trying to circle our main goal and wishes. The main objective was to re-create from Paris a dystopian world (or actually two different). This was kind of tricky because instead of one universe, we had two to build.
The two sides had to be clearly identified by the player in a wink. But we also had to create a common universe with a correlation between sides. In order to organize these ideas, we created a big list of oppositions and analogy of our world.
(order/chaos - symmetric/asymmetric - dystopian - technologic - ...)
And of course, to reference these words, many differents mood board!
After all these very initial researches, we tried some first rough concepts to fixed some ideas.
The environment of this game was a really enjoying challenge to think about.
First, I had to create a small piece of Paris city, and I had all the references I wanted as I was living there! Moreover, I had to realize actually two differents cities and had the liberty to play with different values to create what I wanted.
In order to fit with the production deadlines, we decided to set the playable level on the communist area, while the Capitalist dystopian world would be still visible across the River, transforming this playable environment to an Optimization friendly vista (another challenge).
We had the wish to re-create an Iconic building of Paris but without the classics "clichés". Our choice set on "The Louvre Palace" that was both architecturally interesting but also present on the border of "La Seine".
Thinking of the Production
As a first approach for production, we decided to use a system of layers in the environment construction. The base was the actual architecture of the city of Paris, with Haussmanian architectural lines and some iconic props like the street lights or park benches.
After this, the objective was to apply a communist layer, composed of military equipment, red banner, and massive superstructure that broke the classic architectural lines. In order to improve the speed of production, we did some researches on how to break the monument into modular pieces.
I first worked in 2D to analyze the facade and then created each piece separately : (column/wall/windows/ ...). This method let us knew very early where we needed general textures/specifics textures/sculpting. Then we just made a strict production plan to fit into precise deadlines.
Modular assets were assembled through blueprint which let us create a whole facade building in only a few clicks.
Communists' assets were way more difficult because they had to grow above the classic architecture. Moreover, it was difficult to be clear about what artistic direction had to be chosen about these. Wolfenstein's references came too late in the production but were a real life-saving. We realized that point that we had closed doors in pre-production and it was time to re-open a few.
We build a blocking for the whole mission in order to have a playable prototype as soon as possible. During this time, characters were brought into the engine to verify scales and metrics in the level.
When all assets on the nav were referenced, it was time to start the real production with a strict pipeline.
For an important part of the architecture, we used Substance Designer to create a texture library that was used all over the level. These ones were integrated into a Master Shader created in the Engine that was designed to let us control many parameters.
Moreover, it let us turn the texture library into a bigger material library as many textures and masks could be interpreted in different ways using saturation/values/...
We also brought Vertex Paint to break texture tiling across the level and bring more life to the environment. This was really useful to bring the weather and set the mood in a snowy winter.
However, in the last software, we re-used a lot of the previous material created, thanks to Substance Designer (but also add some new ones), in order to keep coherence in all the environment (but also with characters in metals and wood).
Most of our environment was occupied by the facade of the Louvre. In reality, this one stretches all over a long street for almost one kilometer.
Modular Pieces are the key to how to build such a massive facade in small time.
We split the main wall of our buildings into parts, each part assembling a part of the wall, a window (and window jamb), and some architecture decoration (frise, column, etc.).
To avoid modular problems, each of these elements was blocked into 3ds Max together, but they were textured and sculpt separately, on both sides, in order to be used in recto and verso and to get some variation.
Imported separately, architecture assets were assembled in a blueprint, recreating a small part of the wall, to let us duplicate it along the street in few times.
At first, windows were just a specific shader applied to a basic plane insert into a window jamb. This was a good point to start, to focus on the translucent shader and tweak it for reflection and color. The problem is that when you act so, window and glass don’t integrate very much with the other part of the façade.
At this time, I tried to integrate with the jamb and texture it together. This has revealed very usefully.
Baking the map of this bundle (Window + Jamb) lets me access to Position/AO/Curvature map, which lets me texture way better.
Moreover, even if the window is integrated as a simple plane, a small trick consists of creating a normal map, where each windowpane has a different orientation, creating many variations intro reflection and bringing life to the facade when moving around.
At last, this lets us create a snowy effect on the window but also extract an emissive map to produce a lighted window.
Concerning the streets stuff, modular can be difficult when you’re going to pavements or street barriers. You can create some different meshes and assemble them or you can use a spline mesh component.
This way, using only one mesh for pavement (or for streets barrier, dock border, etc.) and spreading along spline threw construction script, you can resolve almost every specific case of your street and most importantly, you can retake it whenever you want.
Warning, depending on your scene, this technique can bring some heavy draw calls and bring down your FPS count.
To avoid this, you can merge your construction blueprint into a definitive static mesh.
To add street elements, instead of creating many specific meshes such as sewer drain, gas plate, and others, we created a single texture set where all these were packed.
Applied to simple planes with a small border, this lets you add many different details at a lower cost.
Building a Vista
The particularity of this Environment was that it was split into two parts across the riverside. Hopefully, we decided to create only one playable. Still, we had to create a huge vista in order to represent the other side.
As we were learning 3D and not matt painting, we really wanted to create this vista with real 3D assets, avoiding camera mapping or else.
This was pretty tricky, but we had a lot of fun working on it.
To manage this, and because there was a complete skyline to represent, we decided to use some procedural creation.
We created a blueprint in order to create a Haussmanian facade, where you can choose the number of floors and how long the building is. To represent Haussmanian architecture lines, the blueprint adds some offset in the center or side of the facade.
You can also change windows or balconies mesh to create variation.
This process was time-saving for creation because it let you manage your skyline very easily, but at least, it grows the number of draw calls very fast, meaning, here also, you have to merge some buildings at the end.
Although, this kind of blueprint was also used to create some « junk elements » in the Capitalist’s Vista.
We used, replacing public variable into random variables, the same kind of blueprint to spread all over the dock some «Junk» houses and roads. Randomize all the structures let us reach the level of «chaos» we wanted there without spending too much time on the level art.
Finally, we used also these kinds of blueprints to add some life bringing details, with Tags and advertising posters all over the vista.
Speed Level with different blueprints was used to create streets of both sides :
The interior environment was another challenge. When visiting and searching for references in Paris in pre-production time, we all fall in love with the Apollon's Gallery at the "Musée du Louvre".
A few times after, it was decided that this place will be the office of our big evil guy.
The base of the room followed the classic process of blocking and modularity that we initiated, using Substance Designer's textures and master shader to get proper materials.
Yet, all the difficulty of this environment was in the high quantity of details in the gildings you can find on the ceiling of this room.
To avoid long hours of sculpting in ZBrush, we take some time to prepare a TrimSheet sculpted in ZBrush and painted in Designer. The UVs of the assets were tweaked to fit with the texture and it was applied above all the ceiling.
This technique saved us many hours of work and gave us a very high quality in the final render of these gildings.
For the rest, we just had to extract from the web the paintings you can find in this room and also tweak mesh's UVs (everybody is not "LeBrun" :)
In environment creation, lighting is a big part of the final render.
As we set our scene in a winter night, most of the lighting was emit by handcrafted lights. To manage narrative lighting, we first decided to set the communist area in cold lighting and do the opposite on the capitalist’s side: warm ambiance.
When we decided to set our game on a single side, this choice revealed to be less important, and we decided to bring this difference between exterior and interior. Many of the lights used in the environment are set through blueprints. This lets us use, almost all the time, a combination of three different light components for a single environment light.
Windows, for example, depending on the outside a warm area was created using emissive through the shader but also supported by area/rect light to have nice indirect bounds (static). Finally, a stationary spotlight gives the proper feeling when characters walk in the radius and cast dynamic shadows in a smaller area at a lower cost.
We had a lot of fun playing with both, classic and military lighting.
Although, our first objective was to re-create realistic urban lighting instead of focusing on gameplay lighting. (I think retrospectively that maybe it was not the best choice we made.)
Creating a Trailer
In a few words, we had to create a small trailer to introduce our project in our presentation.
Inspired by EndlessSpace 2, United Empire Introduction, we decided to create a propaganda clip to introduce our universe through the eye of the communist side.
As we were neither 2D artists or video artists, we decided to create this in the software we knew: Unreal Engine 4.
I will not spread myself on this subject but know that with few shader tricks and some «sequencer» moves you can achieve some useful presentation without spending too much time!
This project helped us learned a huge pack of different things.
Our main problem was that we wanted to put way too many things into our game.
Two different worlds, Interior + Exterior (I mean concerning the environment only), trying to fit both in gameplay and narration.
All these matters deserve equal attention, and that’s sometimes too much for a school project.
But the most important thing is to be straightforward for your goal.
At this point, we were pretty good because what we wanted, first of all, was to spend this last school learning all the things we needed for the next step!
It’s an evidence that this project has some lacks almost everywhere, but by keeping constant teamwork and a learning mindset threw all the year, we had the chance to conclude our studies in the best way!