Interview with Don Showalter, 3D Artist
November 02, 2006

Don Showalter recently completed work on an in-flight refueling trainer for the United States Air Force. Using LightWave 3D and low-polygon modeling techniques, he was able to produce geometry that took the level of graphic detail in the government program to a new level of realism. Now on other projects, Don took some time to tell us what went into this project to help train those serving the USA in uniform in one of their most high-flying and potentially dangerous operations.

Can you tell us a little about yourself and your history with 3D?
From the time I was a kid, I was very interested in traditional special effects: as he's been for so many others, Ray Harryhausen was a huge influence. I did a lot of drawing and model-building, also some amateur stop-motion work, but never tried to get a job in the field, as I never considered a move out west. I got into computers in the early '80s and did some goofing around on my Atari 800... wrote a VERY simple paint program in BASIC and some games. When TRON and The Last Starfighter came along, it was easy to see that CG had a big future... but, again, learning about these things was just a hobby and eventually relegated to the closet. Fast-forward to the mid-'90s when I got my first PC. This was 3 years after Jurassic Park came out and I decided to look into what 3D packages were available for the home computer. I tried things like POV-Ray and settled on TrueSpace. First thing I ever did was to try and re-create the titles from Superman: The Movie. At that point, I was hooked.

How were you first introduced to LightWave 3D?
I wanted to learn a more serious package; when I found out that Foundation Imaging had used LightWave on Babylon 5, I decided to look into it. I wasn't very familiar with LightWave at the time. I read about a lot of different software but the more I read, the more I came back to thinking about LightWave. With some packages, it seemed you had to not only buy the core, but a lot of expensive add-ons to get the best results. With LightWave, you didn't really need anything else. Sure there were great plug-ins that offered expanded capabilities, but you could do real work with just the core. As I was just a hobbyist at the time, the price was an important factor... but the clincher was the quality of work one could do. I knew LightWave was used in TV work, but it seemed that amateurs too, were producing great stuff. So I decided to try it out. About 18 months after that, I was hired by Foundation Imaging.

What feature of LightWave do you like the most?
I mostly build models, so it's not surprising that Modeler's my favorite part. I like the simple interface with text buttons, and I like the fact that it's separate from Layout... maybe the compartmentalized nature of my own personality speaking, but I prefer them separate. I like the fact that it's easy to make things in Modeler; most of the tools work the way I expect them to... not too much "why is it doing that?" Though I mostly build models and don't do serious renderings, I do love how LightWave's renders look. And, again, the ease and speed; I place my scene elements as I would in the physical world and they work they way I expect them to. I can do a setup and the way it renders will largely match how I've seen it beforehand in my head. Changes are easy to make and the outcome is, to a reasonable extent, predictable.

Are there any plug-ins you use on a regular basis?
Vertibevel was a Godsend for a modeler. The PLG UV plug-ins are another. Some other handy ones are Taut, Multiweld, MultiSetValue and PlanarSelection. Big time-savers, all.

What is your position at Southwest Research Institute, and what does it entail?
I've been working as a freelance artist for SwRI on a new training simulator for the U.S. Air Force. For this project, I've been a researcher and troubleshooter as well as a modeler, UV mapper and texture painter. My involvement with that project has since wrapped and I'd decided not to look for work for awhile. Wanted to just take it easy, work on my website, write some tutorials and do some models that I want to make for myself... but I'm getting itchy already. Just applied for another job. 8^)

Can you tell us about the flight simulator project you worked on?
It's a new U.S. Air Force training simulator for the KC-135 Stratotanker refueling boom operators. The display aspect of the simulator is an array of image-processing PCs, each feeding a projector responsible for a portion of a 24-foot screen. The trainee will be housed in a "boomer" station removed from a real tanker aircraft.

Traditionally, for a project like this, engineers using MultiGen Creator would've been assigned the duty of making the models... and, as I understand it, the military real-time display industry is still using a 1990's standard of realism. Eric Peterson, my contact at Southwest Research, explained that, with this project, they wanted to break away from established processes and expectations in many areas, modeling being among them. It seemed obvious that with the capabilities of the latest generation of PC computers, more detail could be put into the models. The spec called for models of “near-cinematic” quality. So instead of hiring engineers to add detail to existing, simple geometry, SwRI elected to hire artists to simplify high-poly models that SwRI would buy from online vendors, add the parts they needed (like the refueling equipment, various surfaces details, control surfaces and cockpit, etc.) and paint the models with very high resolution textures. For the most part, the approach worked as planned. There were 2 other artists besides myself who worked on the project and I think the others would agree when I say we needed to do more rebuilding work on the models than SwRI believed would be required. In some cases, nearly none of the original geometry was used and entirely new models were built. Some of the models, however, needed almost no work done to them, so things went according to plan in those cases.

Did you have any traditional game experince before this project?
Never worked on games before but, from what I know of the discipline, this sim work was pretty much the same: I got a polygon budget, texture and surfacing specs and a deadline. One difference may have been the size of the textures I could work with; I could use up to 4k, which allowed for a lot of detail. Another difference may have been the precision involved... the Air Force reviewers were looking for these models to be very representative of the real thing, within the geometry/texturing limits. Even in this aspect, I'm sure it's the same in most sim productions where real-world objects are involved.

What was the schedule for the project?
The schedule was elastic, particularly in the beginning, as the specification was evolving. As things went on, my final deadline kept being pushed further away as I took on more work. At one point, I was to have delivered 10 models by a certain date. I missed that date by 8 days, but, during that period, I turned in an extra model that was not part of my original assignment. Still, I was asked to produce even more work beyond that deadline. Lots of nights and weekends were involved, but that was my choice; I knew they liked my work and the more I delivered, the more they would ask me to do.

How was LightWave used in the flight simulator?
Nearly all the work I needed to do was done in Modeler. Often, I'd get a mesh in .3DS format so it'd need to be imported. Then the model would need to be compared to the real aircraft, and I'd have to decide what could be simply altered or what required rebuilding to make it accurate; the usual determining factors here were the density and accuracy of the geometry: if either (but not both) needed work, then I'd work with the existing polygons. If both need work, I'd rebuild. After the model had been made accurate, I built any missing bits. I always had to make the refueling bays, receptacles and doors and usually had to do the cockpits. As the rebuilding/accurizing was progressing, I'd be optimizing the model by hand as I went...lots and lots of merging polygons and welding points. I very rarely needed to use any reduction plug-ins as I got proficient in knowing how many polys I'd used and how many I had left to make what was needed. Once the geometry was done and approved by the customer, I'd UV map it and use Surface Baker to pump out the base textures. I preferred this method as opposed to using something like UV Imaginator to spit out wireframes as the renders would give me precise boundaries when it came time to paint. Once the geometry, surfaces and basic textures were done, I had complete models, save for detailing the textures... so I saved two version of the model, one to create the lower level-of-detail versions of the aircraft and one that was broken up into subcomponents (fuselage, wings, engines and such). For the lower-poly LOD versions, I started out by using qemLOSS since it retains the UV mapping. Later, I used the PLG "Simplify Mesh" plug-in as it does the best job I've yet seen in squeezing down a model: I could go from 25,000 faces to 995 and the model was still recognizable...and, as with qemLOSS, the UV-mapping was retained.

Lastly, all these objects were taken into Layout and assembled into a hierarchy as specified by SwRI; various lights were also added to the airplane at this point. SwRI was using the OpenFlight format and LW null objects served as .flt nodes in the hierarchy. I believe SwRI used PolyTrans to convert the LightWave scene file to the .ftl format.

Were there any particular challenges to the flight sim project?
Most of the work was pretty straightforward, but LightWave helped to avert a crisis early on.

There was nothing in the specification regarding the number of textures or the type of mapping to be used on the models so I went ahead and did the first one as I'd always done things in the past, namely, simple planar or cylindrical mapping and individual textures for these mapped areas. Delivered the model to the customer and they loved it. So I started on the next one. Just as I was finishing it, the customer contacted me and reported the high number of textures (about 70) in the first model was causing slowdowns in the system; a maximum of 3 or 4 would be preferred. Also, they'd like to see the number of individual surfaces reduced from about 70 to a maximum of 5. Obviously, these specs meant UV-mapping (which I should've done from the start, really) and this was easy enough. The problem was to do with my original textures; was I gonna need to repaint the model entirely or would I have to cut and paste the individual textures into new, UV-mapped ones? Sure, UV Imaginator would give me a template but it'd be hell trying to match up all the discontinuously-mapped areas where things like panel lines and smudges had to match up. So I thought about it and wondered if I could use Surface Baker to render the existing textures into these new, UV-mapped textures I needed to make. So I did a test and it worked... and not only did it work, but it worked perfectly, the first time. Huge, and I mean HUGE, relief. So, the next day, I turned in the second model which used 4 textures and 5 surfaces. The client was ecstatic and wondered, inevitably, if I could do the same to the first model I'd turned in. They freaked out when, just a few hours later, I delivered that one, similarly converted. Made me look like a hero.

Now that I think of it, there was another instance where a variation of the same trick helped us out. After the early models were done I settled into a routine of using 3 textures for each plane: one that comprised most of the aircraft; another for things like the cockpit, refueling equipment and external stores, and a last one for the tail. The tail always needed its own texture as SwRI wanted different serial numbered versions available... they would only need to swap out this one texture. And the tail was always the lowest-resolution texture, 1k, compared to say, the 4k fuselage. But down the road, a wrinkle arose where some planes had visible numbers not only on the tail but elsewhere. For example, the F-18F had numbers on tail, forward fuselage and flaps. When I said I'd be happy to paint numbered versions of the other textures, Eric Peterson, the engineer who was my contact, explained that they wanted to keep things simple and have only one swappable texture. So I wondered if I could UV map squares of geometry where these numbers needed to go into the tail's UV map, then use Baker to bake the texturing for these areas straight into the tail texture... then I'd just need to add the numbers and change the surface for these squares to the surface used by the tail. And there'd be no seam as the original texturing would be rendered directly into these squares on the tail texture. As before, it worked perfectly. And I heard the magic words from the customer: "That's exactly what we want."

What other projects have you worked on?
I worked on the Max Steel TV series for Foundation Imaging and the pilot episodes of the new Captain Scarlet series for Lee Stringer. Working on Captain Scarlet was like being in Nirvana as it was a Gerry Anderson show and his work was an early inspiration to me. Not only that, but Ron Thornton was the CG Producer and a lot of FI people were involved, along with Lee.

Which kind of work do you prefer?
I prefer VFX work as it's nice to see your stuff on TV. 8^) That said, I got a kick out of working on this simulator project; it forced me to learn to work in new ways, learn new things and to really think about what I'm doing in order to gain the utmost efficiency. Efficiency is important in TV, of course, but not to the same degree as real-time. I like the discipline involved in having to get a lot out of a little. And it's satisfying to hear folks express surprise when they learn that these airplane models use only 20k faces.

Have the recent changes in LightWave impacted the way you work?
Not substantially, as I haven't needed to do anything that requires the new capabilities, but I want to take the Node Editor for spin. A lot of possibilities there...

Are there any new features of LightWave v9 you've found particularly useful?
The GLSL shaders in Layout's OpenGL were helpful as it wasn't really necessary to do full renders to see how a model was looking; with GLSL, the lighting in OpenGL looked close to what you'd see in a render. Very nice. The orthographic camera is pretty sweet, also... no more need to pull the camera back 3000 feet!

Do you have a "LightWave tip or trick" others might find helpful?
I've picked up some nice UV tricks along the way, but using Baker to render existing textures into new UV maps was a substantial time-saver for me. If you have a pre-textured model that doesn't use UV mapping or if the model uses multiple UV maps that you want to combine into one, start out by not altering the original surfacing but go ahead and make the new UV map(s) that suits your needs. What you'll now have is the fully-textured original object, which will also have this new UV mapping you've just done added to it... this new UV mapping will not yet be used by any surface. Next, go into Layout and setup Baker as you normally would, but select the new UV map you just made. It may seem strange that anything would render out to this map as no surfaces are using it, but what will happen is the original surfacing will be transferred via rendering to a new texture image in the shape of the new UV map...and it'll be an exact fit to that map as it's been rendered as opposed to hand-painted in the normal manner. Now you can go back and change the materials in the Texture Panels of the original surfaces, using the new UV map you made earlier and this texture you just baked out. Or, you could combine surfaces, going from say, 50 of the pre-existing surfaces down to a single surface. Very efficient.

What's next for Don Showalter?
Ultimately, I'd like the kind of career my friend Jose Perez has, that of a freelance artist working on a variety of great projects right from his home office. I've been lucky that the projects I've worked on just sort of fell into my lap... though I've worked hard on the projects themselves, I've never really worked hard at getting work. So I want to focus on that a bit. Upstate New York isn't exactly the center of the CG business but this is where I live... so I'll need to work online to raise my visibility. So far so good; I just put my new website up 2 weeks before being contacted to do this profile. 8^)

Do you have any advice for someone new to the 3D industry?
First, know what you want to do. Modeler, texture artist, animator...? Once you know that, break it down further; if a modeler, do you want to do organic or hard-surface? When you ask yourself these questions, the answers should come fairly quickly; do what you like and make that your professional focus. Doing what you genuinely enjoy will motivate you to want to become proficient. Besides that consideration, I think this field is going to become increasingly specialized as time goes on. Sure there'll be room for the generalist, but I believe the bulk of those in the CG field will be specialists (which is probably already the case).

Once you know what you want to do, decide the media in which you want to practice (do you want to work in movies, TV, games....) then study the industry to learn some specifics of the work. After you've decided these things, however, don't be completely inflexible. My work on this simulator is a good example; it's the type of work I like, but if I said 'no' because it's not the medium I prefer, I would've missed out on a great experience.

Compared to the average person, a fair amount of self-confidence will be very useful to you; there are a lot of great artists out there... if you want to be one of them, then you have to really believe that you either are, or eventually will be, able to sit at the same table. But don't be an egomaniac; if someone takes their time to deliver some valid (stress that: valid!) criticism, listen. Likely, they're trying to help you.

Make online connections. This can be an enormous help, particularly if you don't live (or see yourself living) in area with a lot of CG work available. There are a lot of forums out there, places to post your work and meet people. You never know if someone you meet online is gonna be in a position to give you work someday... really, you never know. With that in mind, it'll help you if your online presence is a positive one.

Lastly, be ready for the hours; in many cases, this is not a 9-5 job and you'll need to be prepared to make the commitment. But for God's sake, have a life. As far as we know, we only get one chance around, so don't spend it all in front of the computer; live a little!

Thanks for taking the time to talk to us, Don!

To learn more about Don Showalter, visit his website: www.Eclectric-FX.com