Last month I was contracted by Toronto-based Xenophile Media to create mockup screens for a proposed project involving a historically-based game-like environment. Xenophile Media is an innovator in the cross-media space, having produced two seasons of the award-winning ReGenesis Alternate Reality Game (I was involved as Game Designer), the convergent feature documentary Beethoven’s Hair, and numerous iTV gaming shows.

In developing their latest proposal, Xenophile wanted to simulate a 3D environment in mockups, but few tricks of design or illustration will achieve this effect. The only convincing way to simulate a game environment would be to use 3D graphics. I chose the virtual world of Second Life as a production tool, and was able to convincingly portray a game-like environment within a very reasonable time span. Following are my reflections on the process, which assumes the reader has a grasp of Second Life's basic landscape and features.

This was my first professional engagement with 3D tools. Normally I would never take on a pro job that required 3D work, since I lack skill in industry-standard software, but for the purposes of concept work, I already had two consumer-level software solutions in mind that I've worked with extensively. The first was BioWare's Aurora Toolset (used in the game Neverwinter Nights), and the second was Second Life. Because Aurora's available medieval-fantasy building blocks were too dated for the project, and because specific characters, items and settings were needed, I settled on Second Life.

For the purposes of production, I created an alternative avatar to minimize unwanted contact and distractions from other Second Life residents. I could have purchased a private plot of land for building, but instead opted to use a series of public sandbox areas to save costs (whatever I built would only need to stand for a day or two). The choice of an alternative avatar turned out to be a minor mistake, as I had to transfer many of my custom textures as well as Linden Dollars from my main avatar. It took a little bit of extra time to log in and out between the two avatars. The use of public sandbox areas meant that I was, on rare occasions, interrupted by curious avatars, weapons-testers, and troublemakers. If Linden Lab (makers of Second Life) offered an offline client for building, I would have gone this route.

Based on Xenophile's needs, I was tasked with creating two main environments--one outdoor area and one inner-city area. The scenes didn't have to be as polished as an actual game would be, and a moderate level of detail would suffice. Each scene would be comprised of basic sets and modest props, with the outdoor scene requiring a group of soldiers, and the inner-city scene requiring a panicked citizen to give it some life. While creating each element was relatively simple, the entire production process turned out to be slightly more time-consuming than I expected.

In building the sets and props, I first turned to Google Image Search in order to source textures based on the real-life locations to be depicted (locations I've been to in person, I might add). I managed to source an excellent photograph of a suitable historical house that included the entire home from pavement to roof. With substantial manipulation in Photoshop sliced it up into textures. I quickly re-created the house in Second Life using basic primitives and applying the appropriate textures. I isolated the door, window-shutters, and hanging flowers as separate objects so that the house wouldn't look so flat when seen at an angle. This single house formed the basis of all the houses on the inner-city street.I truncated the house lengthwise for some houses, and shortened it to two stories from three for other houses. I then tinted the door and shutters of the houses to further differentiate the dwellings. I added details such as adjoining awnings and a cobbled sidewalk to my row of houses, which was curved inwards to enhance the sense of perspective. Once the row was tweaked to my satisfaction, I simply copied the entire row, and rotated it 180 degrees to form the other side of the street. I added brick pavement and details such as crates. At the end of the street (which was supposed to be in a besieged town), I added a broken-down cart I'd built over a year ago for my own use, and some animated fire objects available freely in Second Life.



The outfit for the citizen panicking in the streets (seen in the above inner-city scene) was created using the basic "Prince" outfit available freely to every avatar in Second Life. I merely changed some details to make it look less royal, plumped and balded the avatar.

The outdoor scene was done in much the same way that one might make a miniature set in real life. I created a scaled-down set of hills, and a tiny wall, then added foliage at increasingly small scale as it approached the wall, as well as a similarly-scaled fence to exaggerate the perspective. I needed a group of defending soldiers between the foreground and wall, so I sourced some photos using Google Image Search, knocked out the backgrounds in Photoshop, and plopped these into Second Life, intersecting a ridge. At a distance, this gave the impression of an army cresting a hill.


The main character type required was a somewhat historically-accurate soldier. It proved more challenging than expected to find reference material for creating the outfit. I did a very rough job of costuming using templates available freely through Linden Lab. As any fashion designer in Second Life will tell you, it's seam-matching that is the most time-consuming. The costume consisted of a shirt, pants, socks, shoes, and dress (which formed the basis of a tailcoat). I also added minor costuming details with attachable objects such as a belt-pouch and saber. A hairpiece was created to work with the default hair-sliders; a pony-tail and bow were added. In order to achieve decent character poses in my scenes, I used the free third-party application Avimator. It was easy to learn, but a rather time-consuming method to arrive at still poses (no animation was needed). However, it's the quickest and dirtiest method available. I captured screenshots of the avatar poses in front of a "no set" (white background) at various angles, imported these into Photoshop and cropped out the figures to overlay onto my backgrounds.

During the production process, I would have preferred to buy pre-made objects directly from residents rather than to make them myself. This could have saved my (more expensive) labour. I found the built-in search tools of Second Life to be nearly useless in finding relevant items. By contrast, the services of third-party shopping sites (SLexchange.com and SLBoutique.com) were invaluable, allowing me to search for a few specific items by name and having them delivered to my avatar in-world. There appears to be a gap in the market for historically-themed outfits and items with the exception of medieval fantasy, although I did find suitable (but ultimately unused) props such as horses, lanterns, and merchant ships.

Once a scene had reached a presentable stage, I raised it above the horizon of the sandbox (so that no land was visible), set the sun to daybreak (for rosy lighting) and took a number of screenshots for review by the client. When shooting the scenes and characters, I manually placed my viewpoint (camera) every time I needed a picture, and would have preferred a consistent way to shoot each scene. I did buy a photography kit from a resident to this end, but it seemed complicated to implement the fixed cameras. For a bigger project involving static shots, I would definitely set up such a system.

From a client-management perspective, I found that the screens almost looked too good for the amount of time spent on each (about 3-5 hours per scene). It wasn't a surprise that Xenophile was keen on having an additional dockside scene built. Unfortunately this wasn't possible in the given time frame, particularly because the scene would have needed to be populated with several more characters. Second Life is a great tool for building simple 3D objects, but because every character has to be a logged-in avatar, and because there are no built-in posing tools, it's very time consuming to do group scenes--particularly if new character types need to be created, requiring skinning and accessorizing.

I couldn't have created the concept screens without using Second Life as a production tool. The work took longer than I expected, but considering the final product compared to the relatively short execution time, it seems Second Life can be a cost-effective way to get concept work done. Despite being a good overall tool, Second Life's building features were buggy and frustrating at times. I wouldn't recommend the software for highly-polished work, but for rapid 3D prototyping, it seems hard to beat.