Comparing Apples to Really Expensive Apples
Gone Tropos?
There was an interesting post in the new FSX forum on AVSIM the other day expressing hope that the airport environments in FSX will look as good as those seen in a promotional video for commercial flight simulator builder CAE’s Tropos visualization system.
A link to the promo video can be found here.
The full thread can be found here.
Like a number of people on the Flight Sim team, I’m very familiar with CAE products. I’ve flown a number of their simulators, we’ve even had the chance to meet some of their team members. As can easily be seen in their promo video, or firsthand, if you’re lucky, they do beautiful work.
The first time we saw the Tropos promo, a lot of us on the team essentially echoed the sentiment on the forums – I wish we could (or were going to, had time to, could guarantee our customers had the hardware to) do that!
The video shows some fantastic features – layered fog, smog, great runway / taxiway textures, wonderful falling and blowing snow, gorgeous ice effects on runways, and landing light effects that, because they use a real light map and not just an overlaid night texture like we do, are seemingly perfect. And you can’t overlook that wonderfully fluid performance, even with the compression artifacts in the video clip.
Let's set aside the fact that CAE’s products sell from 20 to 50 units a year for somewhere around eleventeen trillion dollars, they have a lot more computing power and storage than we do and complete control over the hardware, while our products sell . . . more than that, need to run on some pretty ridiculously low-end hardware, and, after a year or two, can be had at Wal-Mart for about the same price as a case of State Fair Corn Dogs (the official Corn Dog Of NASCAR).
We'll also ignore the fact that CAE builds a number of airports for familiarization purposes in excruciating detail, while we build . . . all of the airports in the world and everything in between in varying degrees of excruciation. Not to mention the fact that CAE gets to walk around the tarmacs of said airports, taking pictures and even measurements, etc, while we . . . buy books, snap photos on business trips, and stare at pictures from places like Airliners.net and Windows Live Local.
With those things comfortably cloaked in denial, there is one additional disclaimer:
I think CAE does spectacular work. They deserve every dime they make, their products are fantastic. I am a fan. The paragraphs that follow reflect neither the stuff nor the things of the Microsoft Corporation, especially its lawyers. The subsequent ramblings are intended simply as an intellectual exercise, and must not be used against me in a court of law.
Bearing all of that in mind, I decided to take a short break from testing the animation of the float retraction system on the FSX Goose, and watch the CAE promo video again, this time, as a tester. My inner skeptic (who lives just across the hall from my inner pretentious b**tard) just couldn’t automatically accept the premise that their stuff is “better”.
So, I took a look with a different assumption – how would I improve on it? Did they make any mistakes? Are we doing anything “better” than they are? I gave myself half an hour, watched the video a number of times, and this is what I came up with:
- Their sun effect is static and it tends to look cold and small – our new “bloom” is much prettier, and I think even our FS9 sun was more credible.
- The sky doesn’t change color during the accelerated sunrise scene – the lighting changes so it gets brighter, but it starts and ends a maybe-oversaturated blue. We couldn’t get away with that, at least not without bundling a copy of ActiveSky in every box.
-
Their clouds are flat, 2D, FS2000 era sprites. Ours . . . aren’t.
-
Aircraft shadows are extremely heavy, and dark, almost black all the time and don’t lose intensity in fog (in other words, they do this just as badly as we’ve done it, but our shadows at least aren’t as heavy to begin with).
-
Not all of the aircraft and ground vehicles cast shadows – it looks like they’re not rendering shadows when the viewer’s angle to the vehicle gets too close to zero.
-
Some static ground objects don’t cast shadows either, but some do.
-
Shadows remain fixed underneath those aircraft and ground objects that cast them – they don’t move or change size based on the position of the light source (the sun, in this case).
-
Shadows don’t interact with other lights properly either – the taxiway lights get darker and harder to see when they are in an aircraft shadow, and the headlight lobes of the ground tugs and baggage carts actually draw underneath the aircraft shadows.
-
Speaking of shadows, there’s no self-shadowing of the aircraft – you can watch the sun shine “through” the vertical stabilizer when the Emirates A380 taxis on the icy runway. We haven't modeled self-shadowing in any released products either . . .
(Note: I hate to seem so obsessed with shadows, but we shipped FS2000 without aircraft shadows because not everybody who was in a position to make decisions agreed that it was a problem that needed to be fixed. We ended up having to release a patch, which has a dramatically higher cost (in time and resources) than people realize.)
-
The surrounding terrain is using some pretty low-resolution DEM.
-
Summer and Winter, but what about Spring and Fall? And Hard Winter?
-
Certain ground objects have no night textures at all.
-
Specular lighting, but no reflections on aircraft models. In FS9, we did reflections in chrome, for example, using an artificial environment map. <Edited to reflect Jason's comments below.>
-
Gorgeous bump-mapping and specular, but again, no reflections, on the icy runway.
-
No touchdown smoke (we could probably afford to give them some of ours since we use too much.)
-
No articulated bogies on the A380 landing gear. They’re supposed to do that weird A380 “hang forward” thing.
-
There’s something wrong with the way they’re animating the compression of the landing gear as well – watch the bit where the Emirates A380 lands very closely: at the moment of touchdown, the airplane jumps and seems to be forcibly repositioned. It looks to me as if the simulation engine is taking into account a compressable landing gear, but that’s not reflected in the animation.
-
They could use some more variety in their trees. I don’t remember the numbers, but in our building, in the hallway just down from the restrooms, we have pictures on the wall of all of the currently available Autogen trees. There are a lot of them.
-
Aircraft control surfaces don’t move – no flaps, no spoilers, no fun. If ours didn't, my work this week would have gone a lot faster.
-
Landing lights don’t cast a beam in the fog (watch the 747 land in the snow). It's arguable whether no effect is worse than an ugly one . . .
-
Great snow trails, but where’s the spray from the wheels?
-
Taxiway lines are inconsistent – some areas are really smooth, in other areas, if you look closely, they’re really faceted – just a few straight lines with hard angles making up a curve. Overall, I’d say subjectively that our best is pretty close theirs, and our worst is a good bit better.
-
So, what’s the verdict? Will FSX look better than Tropos? In some ways yes, in some ways maybe, and in some ways no. After looking at the promo video with a more critical eye, I can say that, in toto, it’s definitely not a slam dunk in favor of CAE, even discarding all of the disclaimers I laid out at the beginning. When all is said and done, however, I haven't proved anything here, other than the highest truism in software: there's no such thing as "zero bugs."
Most importantly, of course, I’ve based my observations and opinions entirely on some pretty limited information, but I’m not the first flight simulation fan that’s ever done that. And that's one area in this imaginary and slightly irrational competition between us and them where I'm happy to say we win, hands down: we have orders of magnitude more dedicated, enthusiastic, and passionate customers than they do.
