Leanna Russell talked about her recent large-scale project Fantasy Scene: modeling & texturing approaches, background creation, lighting, and mood.
Hey there! My name is Leanna Russell and I am a Game Art and Animation student from South Central Pennsylvania. I am currently in my Junior year at Champlain College and I am just starting my specialization as a 3D Environment Artist.
I particularly enjoy hard-surface modeling, texturing, and environment scene creation. Before the Fantasy Scene project, I had only previously worked on two smaller scale environments and a handful of prop assignments.
My interest in 3D art started in high school when I was figuring out a practical way to use my artistic ability for my future. I took every art class that would fit in my schedule in high school and loved every second of it. In truth, I had always known that I wanted to follow an art career, but bringing this to practical form I found difficult. It was not until one of my older peers had mentioned that she was going to school to major Game Art that I even knew it was a possibility. Since a very young age, I had always found a passion for video games so Game Art seemed like an easy fit. At the same time, I had never touched digital 3D art until entering college.
Starting off with a traditional 2D art background and transitioning into a new digital 3D space can be really daunting. Artistic ideas are the same but the ways in which you go about creating are completely different. There are so many programs, terms, specializations, skills, and ways of working that can be really overwhelming at times. I find myself to be a quick learner and was lucky enough to have good professors surrounding me so I picked the skills quickly. The two most difficult topics that I’ve learned in the last year were breaking down modularity and understanding how to build environments in terms of scale. Once it clicked though and finally started to make sense my work really began to grow and improve.
My main reference was a concept piece done by Pierre Fabre called Malombre. I also referenced Raphael Lacoste’s concept art called Return of the Knight for the castle in the far background. The key silhouettes and buildings came from Fabre’s piece, but it was decided to take a little bit of a different mood and tone from the original reference art. My main goal was that I wanted the scene to feel lived in and have the possibility of seeming like a real game concept. Also, I wanted the way your eye travels naturally through the scene to be much in the way a player would process their options in a game.
The early blockout for this image was crucial given the camera angle and scale dynamics of the shapes as they traveled further back. I mostly used Unreal’s basic shapes and geometry in the early stages of the blockout and it helped to give me a sense of the modularity and scale within the scene. I spent the first few days of the project just laying down the blockout and adjusting the camera angle. I think this is one of the most important early steps to get right. If things are off in your blockout they will be a lot harder to fix later on.
When starting any environment scene like this it always helps me to break down the modular and unique assets in an image. I personally take my main reference to Photoshop and break it all down. Creating such lists makes a larger scoped scene like this much easier to handle and wrap my head around.
I first break down all modular assets needed and compile them into a list. After this, I find all the unique assets within the scene and list them out by the number of assumed importance. Measuring the sizes of the modular assets is based on the human scale seen within the image. The average male’s height is around 1.7 meters so from here I can gauge the approximate height of all the building pieces surrounding any given person. This gives me a good scale of the whole scene and what modular scaling I will be using for the whole project.
This particular project opened my eyes to modeling from a plane rather than a box. I had only previously worked within a box before so this was a new challenge for me to take on. Extruding out or in from a segmented plane allows for a lot more control. When modeling I try to keep my poly count as low as possible while still creating the same basic shapes needed. Removing unnecessary edges is the quickest and fastest way I used to cut down on a bloated polycount. Additionally, I find thinking in basic shapes can be the most efficient way to break down a complex asset and create things quickly.
I had a blast working on the materials for this project and one of the main reasons my professor, Vincent Joyal, suggested that I do this scene was because of the number of materials found in the original reference. Pushing my skills in Substance Designer and tileable texturing was one of the best parts of my time working on this over the past two months. Before this project, my confidence in Substance Designer was rather low and I thought I didn’t have very competent skills in making materials. However, over the past 2 months of working on this project, I have come to believe more in my texturing and material making abilities.
My way of thinking in Designer starts much the same way every time I create a new material. I begin working with only the height detail of the largest shapes found in the texture and work down from there. All height and normal information are finished before I touch the base color or final roughness information in the material. This makes tweaks and adjustments much easier to do later on in the project. I learned to add a lot more little detail within my maps and that the slope blur node is my best friend. All of the materials in the Unreal project were instanced off of their corresponding master material within the editor. There are 29 tileable textures, 7 plant Atlas scans, 5 decals, and 2 particle effects used in the scene.
I focused on the background heavily first, and it ultimately decided the direction for the change in the background later on in the project time. I worked on it from the front to the back, adjusting the parts when needed. The image arguably gets more complex as it goes back and that is why it was so important to get the foreground looking good before even moving on to other areas. The whole background is modeled and textured just like the rest of the scene. I found the depth to read much better this way rather than just importing a large image plane.
The lighting in the scene was one of my greatest challenges. There is one main direction light that has control over most of the scene and background lighting. As for the foreground, things got more complicated in order to replicate some of the shadows and reflections found in the reference. A handful of spotlights handle the moody, dark foreground. Some are for getting the cast shadows to look right, while others keep the foreground from getting too dark and hidden. Point lights were only used to create the yellow outdoor and house lights you see. My original reference has some complex light sources appearing within the image and replicating this was challenging.
Setting the mood for this image was another fun challenge to take on. In order to get the fading effect in my project, I used an exponential height fog with the volumetric fog turned on. This affected my lighting quite a bit at first but the final result is completely worth it. The color grading effects were a pretty late change but added a lot to the overall tone. The global saturation is turned down slightly while the white balancing temperature is set to the max warm setting to achieve the sepia-esk coloring found in the final image.
The biggest lesson I learned on this project was to trust in my abilities and to have confidence in my texturing. Keeping up the positivity while putting in the time helps to motivate yourself and those around you. Never be afraid to ask for help or google up a tutorial. Learning everything you can will keep you open-minded and help you move forward to your next project.
I used and referenced textures.com, the Unreal Engine 4 Documentation, and the Substance youtube channel, as well as the feedback from my peers and my professor Vincent Joyal during the span of the project. By the end of the project, I felt confident being able to find all of the necessary information quickly by looking through the sources mentioned above.