The Great Hall of Hogwarts: Quick Set-Dressing Technique

Introduction

My name is Jacob Claussen and I work as a 3D Artist at Ubisoft Stockholm. Soon, I will celebrate one year anniversary at the studio. Before that, I started my career with creating a studio with some friends called Tarhead Studio where we made an arena brawler called Blast Out and then was an Environment Artist at Starbreeze working on Overkill's Walking Dead. This year, I have also been teaching at FutureGames in Stockholm which was a really good experience. If you ever have a chance to teach, do it! It really makes you reflect upon the way you're thinking and creating art.

The Great Hall of Hogwarts: Origin

This was a project that started a long time ago when I was studying at university. I took an online course with Jason Stokes at Futurepoly. During my last weeks in the course, I started a project where I wanted to explore the wonderful world of Harry Potter. I picked The Great Hall of Hogwarts but the project became too hard to finish at the quality level I wanted to achieve, so it was put to the side.

A year ago, I "blew the dust off" the project and thought I could quickly fix the project but after looking at the scene I saw that all the metrics were inconsistent and the easiest thing was to completely restart it. This time, I set the goals or "pillars" for the project which is something I usually do nowadays when starting a new piece. Substance Designer, a cinematic, and a 360 environment were my pillars and the topics for research.

It's worth mentioning that I tend to have around three big side-projects at a time, that's why this scene has been in development for so long (the scene itself was made in three months that were spread out over one year and a half). This habit gives me the creative energy to work on things as I jump between artworks and do whatever I feel like doing, both in terms of creativity and studying. 

Start of the Project

Luckily, I started the project from the pre-production stage where I already had necessary reference images in a folder. This allowed me to start planning out the scene right away.

Like I already said, with all the metrics being very inconsistent, I had to redo the blockout but it gave me a much cleaner start. I also used some references from the Harry Potter Museum in London, there are tons of pictures of it online. Using the movies was not a good option because the hall size changes throughout different films and lenses also make it hard to get a clear picture sometimes. Finding a picture with people standing next to the wall or something else was perfect for getting all the metrics right. I even traveled to London on a weekend to see it in person but realized that you had to book tickets a couple of months in advance. Still, I hope to see it in person one day!

When I work on a project, I usually make “proxy” models that I use for set-dressing. The proxy models have the right sizes, basic textures, and material setup. They allow me to do the first iteration of set-dressing quite early on and I can quickly update them when I do modeling and texturing. Also, they give me a chance to do some early lighting tests. When I create a big scene like this one, I always try to work on the bigger picture and not dive too deep into separate props or textures before the overall feeling of the scene feels right.

I first focused on creating architecture and finding the right sizes for different pieces. Assembling a kit for the walls made the iteration quicker. The same goes for the ceiling - I broke it up into several repeated pieces and focused on smaller parts. The scene itself was first built in Maya to have an overview of the whole picture, and after that, I start taking it apart to get some modular pieces for Unreal Engine. A place where the repetition occurs is usually where you can cut the object, for example, the middle beam of the roof.

Props

It may seem that there was a ton of assets to create but most things could be repeated. My workflow was to place everything in the form of proxy models that I could later update when they looked right. During the first iteration, I usually make proxy models 75% ready and when the whole scene is done I can add 25% where it's needed. Focus on bigger brushstrokes as the painters do.

Something else worth thinking about is the number of props needed. Speaking of the assets of the same "family", it's good to remember that they are not only working on their own but also together, especially when it comes to the silhouette they produce. By combining them in a scene at different angles and scales, you can create various different silhouettes. The magical number of the assets combined is usually no more than three, and even that is sometimes unnecessary.

What took the most time was the ornaments. Previously, I hadn't much experience in their production, and now I had to do a ton of them. I approached the ornaments by mixing modeling and painting heightmaps that you can use in ZBrush for Stencil. Michael Vicente aka Orb has a great tutorial for that on his ArtStation. This was mainly done for the animals on the door, the other stuff was simple modeling in Maya with the front view on top of a reference to follow the silhouette. 

Texturing

With this scene, I really wanted to get a grasp of the workflow in Painter, so all the assets are made without smart materials to get a deeper understanding of different ways of blending stuff. This took much longer but for me, one of the important things about personal projects is that you can take time to study something in detail and redo it to find new better ways of working. At the same time, I tried to spend a couple of days maximum per asset in order to work smart and stay efficient.

All the assets were detailed in Substance, - I even skipped doing the edgewear in ZBrush and used the height information in Substance to create some wear and tear instead. Because of the size of the room, I felt like I wanted to focus on the texturing part more than sculpting and it provided me with new ways to use Substance Painters.

I mostly focused on diffuse and roughness - starting with diffuse, I tried to build up the layers of "history" of the prop upon it. If we take a wooden chair, for example, it starts with the wood surface, then comes the coating that protects the wood, then damage and surface discoloration that occurs during the years of usage. After that, you add other things that tell more story - dust, grease, and other touches that show how much and where this prop is being used. I think it's easy to forget that environmental storytelling not only happens in the scene you are building but also in the props that are in the scene.

Let's take this chicken as an example to showcase the process: it was started in Maya from a simple box. After that, I went into ZBrush and used some noise of different sizes on top of the chicken to create some surface detail. After that, I used Decimation Master to get a low poly and added details on top of the flat color in Substance Painter - this was mostly done with the help of different smart masks and base colors that were added in layers as described before.

The statues were made in a similar manner but I only needed to do the texture once and then I could reuse it on other ones with minor tweaks. Many props were used for studying and as a result, I didn't settle on the first results and reworked them instead. As I mentioned before, you can use your personal pieces to really learn something, not just produce an image for the portfolio, especially since there are no deadlines.

Lighting

As said before, lightning started early on, and I continued working on it as the scene progressed. At the end of the project, I reached out to Carmen Schneidereit and Harley Wilson, the lighting artists I really look up to, and asked for some feedback. I did it mostly to see if my lighting worked at all and I got a very nice confirmation along with some tweaks suggested. Finding an expert in the field you only start exploring is always good. This could be done inside the studio you work at or on the internet, people are usually very helpful! Try to get feedback from different people and filter it.

The Christmas Tree was a simple setup - an emissive map on the lights, plus a few lights pointing at the tree to really push its values.

When it comes to the floating candles, it was a battle I fought looking for a way to solve the task. I wanted to get an optimized mesh with some VFX. At first, with the help of Elia Anagrius Stampes, a VFX artist at Ubisoft, I made an animated material using World Position with noise to create a wavy flow. The motion looked nice but I was missing the ability to add VFX and normal maps. The next thing was a blueprint which Lewis Thomas helped me set up - as a result, I was able to add different VFX and make a better candle mesh.

Jacob Claussen, 3D Artist at Ubisoft

Interview conducted by Arti Sergeev