Long life to Embark studio and its fabulous procedural artists dream team !
Architect Feileacan McCormick used Unreal Engine 4 to build an incredible thesis project that explores the possibilities of Epic Games’ tech for filmmaking and visualization of complex environments. In our exclusive interview he talks about finding the right assets, building the level and working with the virtual camera in UE4.
My name is Feileacan McCormick & I’m a bit of a jack-of-all-trades and mostly work in with architecture, photography & video. I very recently graduated as an architect from the Danish Royal Academy of Fine Arts school of Architecture where I studied at the department for Art & Architecture (KKA).
Cabin/woods is my thesis project that uses film to communicate a series of architectural interventions that explore the intersection of culture/nature as a productive friction. The modern cabin has been for a while now moving towards a purely aesthetic experience of nature. Nature may be experienced when outdoors, but within the safe confines of the cabin the space becomes at times hard to distinguish from a generic urban dwelling.
Simultaneously with intending to develop a new type of cabin I wished to explore film both for communicating and developing architecture. A had to find a solution to both architectural (designing a new type of cabin) and technical (how to find a good workflow for using film in an architectural process) problems this is where UE came into the picture.
Biggest Selling Points of Unreal Engine 4: Speed and Interactivity
I’d known for a fairly long while that I wished to work with video as a primary part of my thesis project, however it seemed fairly impossible object to do as traditional digital rendering is incredibly time-consuming and span of 100 days that we have to produce our thesis project within would have been far too short a period of time to create a satisfactory, not to mention that the the architecture itself would have had to be finished before starting the process of filming. Luckily for me I’d noticed a little while earlier some posts on various architectural blogs on Koola’s architectural videos & decided it’d take a closer look.
It all seemed rather alien & daunting at first, but I decided to try and produce a couple of minutes of video/some renders for my then semester project as a way of testing it out. I ended up only having a weekend within which to try out UE, so the results weren’t especially good, but in the process of trying out the engine I was struck by how powerful a tool UE in fact is.
Initially the sheer speed of UE itself was really quite something. Normally it takes quite a long time (when rendering) before one can see how a material/lighting change etc. effects a space as a whole. Now materials, lights and other parameters could be swiftly adjusted/replaced in real-time, all the while being able to see every change reflected in detail. This was naturally quite an a strong sales-point in itself, but the real advantage of the technology took me a little longer to grasp: UE speeds up the visualization process greatly, but as a visualization tool it really just replaces vray/corona/maxwell etc. at the natural “end-point” of a project. Where it really shines for me is that it can be an active part of a process: I was very early on able to set up a simple plane with sun, sky & wind. From there on I could import early sketches of the cabins and start to explore them in first person, seeing how various materials affected their spatial behavior.
Just the ability to see how a quick 3D sketch of a curtain would affect the space of the cabin when it moves with the wind made it possible to understand far more precisely how the consequences of a design choice enact themselves upon the architecture. Being then able to also “walk around” and discover details & inspiring moments really pushed the process forward.
UE4 for Education
I find Unreal Engine 4 to be an incredibly good tool for students of architecture. One of the core strengths with UE is its “roll your own” like quality, it’s an engine that comes with a starter kit showing off through examples how it can be used. Simultaneously these are only examples, unlike a renderer you choose how you configure & structure your project. It’s possible to blend visualization and interaction in countless ways: you can make a game that allows you to explore a/many buildings, or create a series of static “stills” that you can look around in with a VR headset and so on. In addition to visualization work it can be rather used as part of the form-giving process, allowing one to “test” how a design behaves with a far greater degree of spatial precision (lighting, environment, textures) than modeling software allows.
Another strength is the multitude of platforms that are open for exploration: Being able to create experiences/interactions on everything from mobile devices to VR headsets really expands the possible ways we can discuss & create architectural experiences/interactions. Until now these sorts of things have demanded technical knowledge that spans across several technical disciplines, this is now far more gathered within a single interface.
That said, for most students there is a pretty steep learning curve, especially for those without much technical experience. It is however really worth the time, sweat & effort. It’s an amazingly comprehensive & flexible tool that opens for many new & unexplored areas within architecture. Just in the three months I have been using UE I’ve been able to try out many new ways of thinking and working.
Production of the Assets
The forest environment is based upon an actual 3D scan of a typical scandinavian forest. I advance of the thesis project I spent a weekend recording an 20x80m rectangle with a Kinect setup I’d been experimenting with. However, the scan itself was far too heavy & unwieldy when imported into UE, even after drastic simplification. nor did it utilize the potential of the vegetation and landscape tools. I therefore decided to use the scan as reference point from which I could extrapolate the scale and density of vegetation and landscape, removing it after the forest had begun to become mature enough as an environment.
I used a variety of assets to create the forest environment. Due to a lack of experience with asset creation I bought a handful of asset packs that helped me build a library of materials and assets that I could then adjust & use to build the forest itself. I spent an awful lot of time figuring out what was required to make the forest look “right”, since the forest was supposed to match the sort of forest we typically associate with Scandinavia a lot went into what types of grass, bushes, plants & trees ought to be used. I’ve spent a lot of hours attempting to figure out the variables in the foliage tool (the variables for scale, density, vertical angle & rotation were key here).
Lighting played a very important part of creating the forest. It became quickly prohibitively expensive & inflexible to build static lighting for a forest consisting over around a million instances of foliage assets (I say inflexible, I mean downright unthinkable). At the same time I wanted to have a dynamic lighting setup so as to be able to explore this environment at a variety of different times.
After a fair bit of reading and experimenting I settled upon using DFGI as a lighting solution. In retrospect this would seem to have been the perfect choice for this project as it works incredibly well in outdoor situations (if this project had been an interior project I think I would have gone for a VXGI setup/LPV instead). Using DFGI required then, as now, building UE from source, but that turned out to be virtually painless as long as one follows instructions.
I recommend spending as much time as possible with lighting in the beginning of creating an environment, it really has so much to say for the end result & also gives you some important resistance when you begin to populate the level with a landscape, foliage, structures etc. Also really working with the foliage tool, seriously, it’s getting so powerful now!
The modelling of the cabin was pretty straight forward (mostly using 3Ds Max/a little Blender), however in the beginning it was far too thin (in real life the metal plates are roughly a centimeter thick), which caused some odd lighting issues in-engine. For the sake of getting the correct light the structure was doubled in thickness to two centimeters which made it far easier to get a more consistent light/shadow look.
Texture-wise getting the metal right took a lot of time & I burned through every metal texture I could get my hands on until I found the perfect one in the form of a dusty chrome texture from GameTextures. I was looking for the right level of reflective ability, so that the cabin’s surfaces would reflect enough of both its context (and phenomena like sunlight) so as to be constantly fluctuating.
The (translucent) plastic curtains were, and are, a larger headache for me. I still haven’t quite managed to create a translucent, thick plastic. From what I’ve understood it’s incredibly expensive (relatively) for a real-time rendering setup, so it’s therefore not directly possible, but I hope with more time (and knowledge) about PBR materials to figure out a way to make it look far better than it does now. On the other hand, the metal curtains were quite simple, a simple (similarly chromed) texture applied to the Apex cloth mesh.
Rain & Fog
Tthese are mostly assets bought and then adjusted in order to save time. For example, I spent a fair bit of time modifying the rain particle system to get the right feel for the rain-scenes (finding the right size & speed of each rain-drop particle took a lot of trial & error so as to get the feeling of a heavy rain-shower that could reflect the recording of rain I had chosen). These are fairly simple blueprints that it didn’t take too long to figure out how to adjust when needed.
The placement of these various elements within the woods was done based upon how they could contribute to the narrative of the film,to underlie the eerie, uncanny quality if the cabins. How they were they reflected in the surface of the cabins, how they could hide/underline aspects of the cabins or their surroundings etc. The fires in the wood are a good example of how the qualities of the cabins, the curtains and their presence were both underlined & warped by the presence of of fire.
UE4 for Film
The camera work comes from experimenting with video projects of the years. I’ve always held a fondness for slow, steady pans and static framings that challenge the viewer, making them spend time within each scene, layering details subtly so as to push the viewer to discover them through time spent, rather than over-explaining visually. Thankfully the fact that one can in UE, as in the real world, “walk around” discovering compositions and moments allowed me to work as would in real life: I spent a lot of time walking around, hunting for moments & compositions that caught the eye, then delving into them through filming tests, experimenting with different speeds, durations, times-of-day etc.
In that sense working with cameras really is quite a breeze, although I’m going to be diplomatic here and just say that Matinee is bloody clunky by comparison. Thankfully there is a new version of Matinee in the works somewhere, so it’s only a matter of time before it ought to be drastically improved. The cameras themselves are quite alright to work with, although I hope there will be options for adjusting zoom in a more “physical manner”, that is to say being able to define what length (mm) lens the camera uses, rather than adjust the FOV. I also hope that we can in the future set up camera “dolly tracks” with a greater precision as that would further enable filming inside UE to come into its own.
UE proved to be an excellent tool for the creation of realistic animations & cinematic film. I really hope to see some fantastic new films & hybrid film-interactive creations in the nearer future. For the architecture profession it opens the doors for some so far unexplored avenues when it comes to working with speculative, experimenting & interactive architecture.
Plans for the Future
Right now I’m planning several projects that I hope can continue along the trajectory that the cabin/woods project points along (the borderlands between vulture/nature). Simultaneously I feel a great need to dive far deeper into UE & learn properly a lot of what I only half-way understand now. Some time this summer I hope to begin another (collaborative), more complex UE project that could hopefully become both a film, but also an interactive experience (grasping the potential of the “game” part of “game engine”). Apart from that I have a few other concepts bouncing around in notebooks that I hope to try out that could become analog/digital hybrids in installation-form, not to mention I’m dying to see what 3Dscanning could do in terms of creating a world of glitchy, non-photorealistic assets. Oh, and I probably need to start applying for jobs soon.