Interview with Richard Harvey

Friday, January 12, 2001

We had the very good fortune to interview TRI's Richard Harvey despite his recuperation from a recent operation. Timed to coincide with today's announcements, we think everyone anticipating the new Fly II release will find both insights and answers in Richard's most gracious responses:

AVSIM: First, a general question: Please share with our readers some general background about the development team's objectives for Fly II. For example, our perception is that two types of simmers must be accommodated. One type focuses on scenery and aircraft visuals. The other focuses on realistic systems and procedures. How will the Fly II enhancements accommodate both types?

Richard: When we started Fly! years ago, one of our biggest dilemmas was how a first generation flight simulator could stand out in a crowd of very entrenched products (during our original development, Flight Simulator 95, Pro Pilot 98, and Flight Unlimited 2 were all already on the market). We knew that we could not surpass each of these products in every category on our first try, so we looked at what was missing in each of them. The common thread was system detail and cockpit fidelity, so that is where we focused much of our attention. Fly! delivered a pretty good combination of flight dynamics, satellite imagery, cockpits & systems, and expandability to compete well against the other products. However, this meant our feature set also catered almost exclusively to the existing flight sim fan, someone who was already familiar with aircraft and piloting. We failed pretty bad for the beginner pilot, particularly with our documentation and lack of tutorials. Our cockpits and system realism made it so that even experts had to consult HOW-TO documents just to figure out how to start some of our aircraft (the Hawker earned a quick reputation for needing some procedures training before even attempting to use it.)

With Fly! II we knew we already had a good solid foundation for the serious flight simulation user, and we knew that any improvements in our strong areas would still leave us well ahead of our competitors. Where we really needed to focus our initial attention were areas that even the beginning pilot would see as a weakness in our product. Our initial concern was the quality of our scenery and ground resolution when compared to MSFS, including aircraft and building model quality and ground scenery imagery. We also knew that users of all levels judge a flight sim on it's ability to mimic reality, and there were many visual areas of the product that we could enhance that would heighten the realism.

For the advanced user, we actually had less fears. Although we knew they would be much more critical of the product, we also feel we understand their expectations better because they are so similar to our own. The good news for the advanced user is that Fly! II will not be a "dumbed down" product; we do not intend to take steps backwards in areas where we set new standards. Although the primary focus of the product will be in bringing Fly! to a broader audience, there will certainly be enough new gadgets to keep the advanced user happy.

AVSIM: Considering the last year since the original FLY! came out, what do you wish had been done differently? How has that affected the development of Fly II? In other words, what were the biggest lessons learned from Fly!?

Richard: Easily the biggest lesson is that the community wants a stable product above all else. Although in hind sight I think we built a stronger relationship with the community much faster because we had to interact with them daily, it certainly would have been better for everyone had our 1.0 release been postponed until we could address stability and compatibility issues. I think we also underestimated the amount of effort needed to fully test a product of this scale, particularly one that our company and test team did not have a history with. Another big lesson we learned, and one I hope other companies in the game industry might learn through us, is how positive an experience it can be if you are honest and open with your users and have good two way communication with them. For example, we allowed the AVSIM forum to vote whether to wait for a single unified patch resolving the 1.0 issues, or to receive multiple patches as fast as we fixed major problems. The users chose the latter, and even though it pushed everyone's patience to the limit in the end we let the users decide what and how they received support from us.

AVSIM: Following up on that, which of the new features proved to be the most challenging to implement? Please point out the more innovative solutions the team found for these.

Richard: Revamping the ground terrain system was the single largest task for Fly! II, and we probably went through four or five iterations before we finally found a solution that gave us what we wanted. Once our patch cycle was finished, Scott Johnson (who wrote the 1.0 ground system) began a complete rewrite of the existing terrain code. The caveat for Scott was the new system had to solve all of our problems, but also had to be backwards compatible with all the existing scenery released for Fly! (I'm sure the AVSIM Library admins are familiar with these add-ons!). The original goals were:

  • Increase our ground texture resolution to be competitive.
  • Increase our visibility range for better high altitude flight.
  • Support a higher density elevation mesh.
  • Add true coastlines for the world.
  • Improve our satellite imagery
  • Make the terrain look less static and "stale".

The terrain system is really an advanced real-time GIS application; we have to take real world data from different sources and piece them together to make the entire world! The terrain is also the backbone for the entire product—there is not much between point A and B except terrain and weather, so both better be pretty outstanding. We also had to find ways to increase our data (resolution, sample points, etc.) without increasing disk space, stutter, or hindering performance.

Shortly after starting this huge task, Scott accepted another job offer and we were left without our best terrain expert which was disturbing to say the least. Through a job posting on the AVSIM Fly! Forum, we were contacted by Neal Hall who happened to have the perfect combination of programming background and GIS experience. Neal has come in and done an outstanding job with the new terrain, and I have no doubts users will be very happy with our new system. Neal is as demanding on flight sims as any user we have found on AVSIM, so his expectations for himself and the product were extremely high.

We also put our money where our mouth is and signed very key relationships to improve the quality of our imagery. The first relationship we signed was with How in the World? to create our generic ground imagery and new custom Terrascene city areas so that the terrain would look top notch regardless of your location. The second relationship we signed was with Todd Klaus to secure the Terrascene 2 technology that has been so widely acclaimed by the community. Scenery and terrain will no longer be a weak spot for the Fly! product line.

AVSIM: Now to some specifics:

FLY! was a slam dunk with aircraft panel modeling, and accuracy of aircraft systems. How will Fly II top its predecessor in this area? Also, will additional aircraft be offered? And how has the external appearance of these and the existing aircraft been improved or refined?

Richard: It definitely would be difficult to "blow away" our 1.0 panels, but we are introducing new aircraft whose panels are certainly a notch above the quality of our 1.0 product. Our artists learned a lot during the 1.0 cycle, and each incremental aircraft kept getting better and better. For Fly! II our new aircraft are absolutely the best panels we have created to date.

The first new aircraft we are offering for Fly! was actually hand picked by the community. Shortly after 1.0 was released, I ran an e-mail voting poll that allowed users to vote for the aircraft they most wanted to see in a GA simulation. I was confident I knew the answer, but once I started receiving e-mails I quickly became confused. After polling was done I tallied the answers and the winner was a plane I wasn't even familiar with - the Pilatus PC-12. I had to call for sales information on the plane just to see what the excitement was about; this plane beat out every heavy and small jet, acrobatic plane, and military aircraft so it obviously had something going for it. Once we reviewed the plane, we quickly agreed the community had chosen wisely; the PC-12 is an absolute dream aircraft that anyone would be thrilled to pilot. It is as advanced as any aircraft we could have offered, but has the advantage of being flyable by a single pilot. It has incredible performance with very advanced avionics, yet clean and simple enough that beginners to experts will enjoy it.

We also wanted to make sure we added a new dimension to Fly! for the advanced users, so that meant offering a new type of aircraft. After reviewing our options we decided it was definitely critical that we introduce rotorcraft to the Fly! engine, and we picked the Bell 407 helicopter as our most wanted candidate. We've built a great relationship with Bell, and with their corporate headquarters just miles from our office it has allowed us to get very familiar with this helicopter. My co-producer Brendan Goss has already completed a complete ground school course with the 407 including simulation time to help ensure we accurately represent this aircraft.

External model visuals is an area that we've been wanting to improve for some time. While our Fly! 1.0 models certainly added some new elements to the GA scene, our planes simply did not have the crispness in the texturing that our competitors have, and it certainly showed in screen shots. We reviewed this area to see if we simply needed to take a new approach to texturing, but Mark Randel decided we needed to radically change our direction in both texturing and modeling. Mark made key improvements in our core rendering engine to give us the freedom to greatly increase our model complexity, while at the same time giving us a much higher texture resolution. For example, in our 1.0 product a typical plane was about 1000 polygons and used 2 or 3 256x256 textures. In Fly! II, we have some aircraft that are near 7000 polygons, and use up to a 1024x1024 texture (over a 500% increase in resolution). This is allowing us to put incredible detail in the models - on our new Skyhawk 172SP you can actually see the cable linkages that attach to the rudder, and they move when used. I believe everyone will agree once they see some initial screen shots that our models will meet or beat anything else out there in detail.

AVSIM: Another point on panels / systems: some have felt "what's the sense of an extremely accurate systems, if we know they are 100% fault free?" Will Fly II include a system failure capability?

Richard: We debated this for a long time, mostly because we've always been sticklers for doing something 110% if we do it at all. We had avoided failures initially because if you offer one, do you also offer the other 1000+ that can possibly occur? I think after talking with the community we got a better sense of what they expected from failures, and it would not require us to figure out how to simulate (and test) thousands of combinations of failures. So, in Fly! II we are introducing limited failures in aircraft, though they will be in categories most requested by users (engine failure, engine fire, gear damage, etc.) For the aircraft designer, we are opening up the "guts" of the system so that designers can implement failure checklists, then trigger these through a system or through the SDK. This means that a designer could implement a very complex series of cascading failures. These could be triggered to go off randomly, or they could manually be selected for training.

AVSIM: Perhaps the most often mentioned criticism of Fly! has been the relatively lack-luster appearance of its scenery. How are you addressing that perception? And will Fly II support a higher resolution terrain mesh? While we're on the scenery topic, Aerosoft recently put out their Rhein-Ruhr scenery, which included highly detailed airports. Will Fly II also include accurately detailed airports? Dynamic airport Scenery? Lightmaps?

Richard: Well, our 1.0 product broke very new ground by not only offering some of the best satellite imagery, but also the architecture to easily import ortho-corrected imagery into the product. This produced probably the quickest growth of ground scenery in the history of a GA product. However, the side effects were somewhat low resolution (some of our imagery was as low as 30 meter), and a fairly crude mesh. For Fly! II, the first thing we wanted to do is improve ground resolution and imagery quality. After shopping various satellite image providers, we became convinced that it is near impossible to offer a wide range of cities at high resolution while retaining color balance and finding cloud-free images. We were very impressed by TerraScene, but resolution was still a concern. Todd Klaus then shared with us his plans for TerraScene 2, and we looked at customizations that would drive the resolution as high as possible. This would solve our cities, but not world coverage. The best person we knew in that field was Jak Fearon of How in the World?. Jak had made significant improvements to our 1.0 scenery with various add-ons from his company, so we contacted Jak to see what his ideas would be in a 2.0 product. Jak put together some samples that blew us away, and we knew if we paired How in the World?'s generic scenery with Todd's TerraScene imagery, we would have an awesome combination. We signed agreements with both companies to provide imagery and technology for Fly! II and future products. Our imagery in Fly! II will be at 7.5 meters per pixel everywhere. We are also increasing the number of elevation sample points per tile so that we can increase mesh density in areas that need it most, particularly in mountain ranges or canyons.

As for airports, our 2.0 airports will be much more detailed. We are greatly increasing the number of models at the airports, including such details as taxiway signs, trees, fences, light poles, power lines, etc. Some airports will be more detailed than others, since we still face the problem of how to populate 10,000 airports worldwide. However, users will probably want to know that Fly! II has a built-in scenery editing mode that can be activated in flight, so if a user is sitting at an airport and wants to add more detail, they literally can press a key, use their mouse to quickly click and drag new models into place, then press a key to resume their flight. And any of these customizations can easily be shared with other users. We also have added dynamic scenery, including a path editing system, so you can add moving vehicles, fire trucks, fuel trucks, etc. to your airports or cities.

AVSIM: FLY! also broke new ground by providing a dynamic weather environment using online METAR data. Will additional functionality be added for things like Winds Aloft? Will the IFR and visibility minima be possible down to 200-? Will Clouds with extensive vertical development like Cumulonimbus be possible? And will those harsh transitions be addressed between different reporting stations?

Richard: We've added some neat things to our weather system. For METAR, we've added a new automated download and updating system so that you can update your METAR information with a single-click. You can also setup a timer so that Fly! will automatically update your METAR during flight so that you can have real-time weather. We have extended our wind system to support winds aloft up to FL600. And our new terrain system supports down to zero visibility. We have added a range of new cloud types and visuals, though I'm sad to say we still haven't found a cumulonimbus solution that we consider acceptable (sorry, no "floating wigs" from TRI.) Transitions should be better because we've increased the density of reporting stations, though if you are moving from extreme weather to extreme clear you will still will see a transition. We also have extended the flight planner so the user can drag and drop weather onto their maps to create custom weather situations along their flight path.

AVSIM: How about ATC and the full IFR experience? What additions can we expect? Will ATC give holds? Will Full ATC be available worldwide? DPs (SIDs) and STARs? Will runway assignments be consistent with minimum runway lengths for the aircraft type, and with wind directions and speed?

Richard: Well, I might as well break the bad news. ATC is one of the areas that a few of us really wanted to focus an entire year on, but when you look at the other areas where our product was either weak or absolutely lacking we had no choice but to focus our primary attention in those other areas. ATC will not be getting a major update this release, so many of the items we had hoped to add (such as SID/STARS) will not be there. This doesn't mean we're completely ignoring ATC, because we are making fixes and tweaks for some of the more annoying problems such as using prevailing winds to determine the active runway, fixing the "loops" where a controller will have you go around indefinitely, and having to wait on the active runway because of an aircraft on long approach. We have also made changes within the SDK to easily disable our factory ATC, because we've seen quite a bit of interest in various ATC add-on products that will further enhance Fly! by third parties.

AVSIM: When the SDK version came out some users were disappointed that the SDK didn't seem to permit access to the ATC programming. Will the new scripting feature allow all aspects of Fly II to be tweaked, or are some functions still off limits?

Richard: It's not so much that areas were off limits, it's that we didn't offer enough ways to get our factory systems out of the way. ATC itself is not really an independent system, it's a logic tree spread across multiple areas of the product and that is why there is no simple way to allow tuning of the ATC; it is not self contained. However, the FlyScripts! product by Anthony Merton that was recently released for Fly! 1.0 has allowed for some ATC alternatives to be released, and similar types of ATC will be possible out-of-the-box with Fly! II and the new SDK. The various ATC add-ons for MSFS would also be fairly easy to implement by a third party if they were so inclined.

The SDK itself is probably going to add another 100 or so functions across a broad range of areas to give even more control. We have also made critical fixes is many of the SDK functions that made 1.0 add-ons more difficult than they should have been. The biggest problem with the 1.0 SDK is that we had no idea what users would want to use the SDK for; now that we have received e-mails detailing the problems users have had with the 1.0 SDK, we have a very good idea of how to make the SDK much more flexible and friendly for 2.0.

AVSIM: You've always said that you knew that FLY! would have to be a community product, taking advantage of third party add-ons to enhance the product. However, it was almost a year before the SDK version was available. It required considerable knowledge of programming, and a dedication that was probably beyond the ability of the average simmer. Will Fly II be more of the same, or will there be (whether by TRI or through other contracts) Aircraft, Airport, and Scenery builders and editors included?

Richard: We have made two significant additions to the core Fly! product that should help correct this. The first is the built-in, user friendly scenery editor. Anytime during flight the user can press a key to enter the editor - the simulation is paused, and the user is given a palette of tools to manipulate the scenery. A simple mouse click will let them add buildings, trees, fences, poles, power lines, bridges, towers, hangars, terminals, flags, trucks, etc. They can use their mouse to drag and drop these items anywhere they want. Once they are satisfied, they simply hit a key to exit the editor and return to their flight, and their new scenery is immediately available. These changes are stored on disk in tiny (usually 10K-100K) scenery files that can be shared with other users, so you could build your hometown or local airport and easily share it with other users.

The second item we have added that I think will make the SDK become accessible to almost any use is the new Fly! scripting system. We have chosen the Python language to be the scripting system for Fly!, mostly because it is a very easy language to use, it is open source so there are many code samples already on the Web, and you can go to your local bookstore and buy many quality books to learn Python. Our scripting system gives you almost complete access to the entire SDK interface, so anything you used to have to know C or C++ to program you can now write with a much simpler script. Another key advantage is that our script system itself is an SDK drop-in, and any other designer can create their own scripting languages and make them compliant with our scripting engine interface and Fly! will treat them as perfectly valid scripts. Since there is already one script system out for Fly! 1.0, users will have a choice of scripting languages or tools to best suite their needs.

AVSIM: Again, following up: are TRI's contracts with those third-parties exclusive? By that we mean can those developers, TerraScene for example, continue to offer similar add-ons in the future?

Richard: Some of our new relationships are exclusive to Fly!. However, we do not intend to stifle or prevent the freeware/shareware areas to continue to flourish. In those areas where we have acquired tools or technology that had also originated in the shareware market, we have given specific freedoms to those companies or authors to continue to offer free or shareware versions of those products. Our only concern is that where we have made significant investments, we do expect those tools to remain Fly! specific.

AVSIM: Will the Aircraft, Sceneries, and Utilities that were designed for FLY! be useable in Fly II, even if they can't take advantage of all the features? If so, will the need to maintain compatibility limit Fly II's features?

Richard: Our intention is to remain completely compatible with 1.0 formats. You are correct that if you simply move a 1.0 add-on directly into 2.0 more than likely you will not be taking advantage of many key features 2.0 offers, such as higher resolution, better models, etc. We also are not sure yet which add-ons can simply be dropped in, and which will have to be run through converters. There may be some areas where we'll have to offer tools that can convert 1.0 file formats into 2.0 formats, but we haven't made final decisions yet. Maintaining compatibility was a concern, but was not a limiting factor in deciding the 2.0 feature set.

AVSIM: Many Airport/Airspace simulations in the real world aviation industry use basic text arrays to enter the simulation variables. For example an 'Official Airline Guide' file can be loaded to provide data for airport activity; a list of fixes, altitudes, and Nav facilities can be manipulated to simulate a SID or STAR. Will FLY!2 provide any database manager where users can modify, update, and share this information? Obviously TRI cannot be expected to be intimately familiar with common procedures at every airport worldwide; this would permit users to tweak and improve the reality of the sim.

Richard: We had this idea over a year ago when we were contemplating SID/STAR support in the ATC engine. Most of what you are describing could be accomplished simply by having a user-friendly database editor for our existing database files; most of the variables (such as traffic density, navdata, etc.) already exist in our databases. We've added no specific support mostly because this isn't an area I had received a lot of feedback, but if this is an area users want a friendlier tool we can certainly consider it.

AVSIM: Given Fly!'s systems complexity, some have found it difficult at best to master them, and we've seen that this may be even more of an issue with Fly II. In a few cases users posted tutorials with step-by-step procedures, including setting up the FMS, finding their way around the airport taxiways, and using ATC facilities, and so forth. Have any arrangements now been made to provide tutorial-style documentation for using Fly II with specific aircraft types?

Richard: Yes. Our 1.0 manual did a pretty poor job of showing the user how to fully use our product. For Fly! II, our printed manual will focus on how to actually use every major feature in the product. We will then offer HTML documentation on the CD that will delve into the more detailed systems, such as GPS, FMS, using autopilots, etc. I certainly hope we can do a better job with Fly! II; I can't imagine doing worse!

Another thing we've also added is flight tutorials to walk users through some of the basics, and we've also added things like audible callouts so you can perform your checklists manually without callouts, manually with callouts, or automated with callouts. This should help users slowly gain skill in checklists and take over more of the startup procedures themselves.

AVSIM: Very recently there have been questions on the Forum related to FLY!'s marketing. Let face it, you guys are up against a '500lb gorilla'. Will there be some changes in the way Fly II is marketed? We think the FLY! community really wants to see this product be successful.

Richard: Yes, we have also had major concerns. We have been working with our publisher to build a detailed marketing plan so that everyone is on the same page for things like promotion dates, interviews, articles, and trade shows. We also don't intend to repeat the mistake of running most of our advertising months before Fly! actually shipped, which hurt us. We've also identified ways to get the most bang for our buck by concentrating in areas we know we can attract aviation enthusiasts. For example, having a presence at this years Oshkosh is important to us. But, as with 1.0, I think people would be amazed at the power of word of mouth—our users have done a fantastic job of spreading the word about Fly!, giving demos to their friends, or recommending us to others. I'm sure this type of grass roots will still make a significant impact to our acceptance.

AVSIM: Is anything being done to encourage commercial third party support of Fly II? For those that are not online, it would be nice to see as many commercial add-ons next to the Fly II retail box (or at least by web download) as with your major competition.

Richard: We have been making relationships with many vendors, and though I can't mention names I can also say that almost every major flight sim publisher out there has contacted us at some level. Some are still sitting on the fence waiting to see how Fly! II is accepted, and others have already made firm commitments to develop for our platform. We're doing what we can as a small team to give all the support third parties need to get started with Fly!.

AVSIM: Will there be anything like an upgrade offer for current FLY! users, or a competitive offer for other flight sim users?

Richard: We might do a competitive upgrade, but we also intend to price Fly! reasonably so every user will feel comfortable they are getting good value for their money. I pointed out well over a year ago just how affordable GA flight sims are compared to other forms of entertainment. For the hundreds or thousands of hours users spend with a product, their cost is pennies per hour. There aren't too many ways I can think of to keep yourself entertained that cost so little.

AVSIM: And finally, what is the expected timeframe for release? When can we expect to see some screenshots?

Richard: We are shooting for a release in the second quarter of 2001. If things go well, you'll start seeing screen shots just about the time you publish this interview!


Tell A Friend About this Review!


© 2001 - AVSIM Online
All Rights Reserved


Click Here!