Translations for our friends around the world.

Click on banner for Tornado website


Author Topic: An Interview With Digital Integration Ltd (Part 1)  (Read 5025 times)

0 Members and 1 Guest are viewing this topic.

Offline Frankie

  • Moderator
  • *****
  • Posts: 1478
  • From Dogs of War to Pipes Of Peace
    • Frankie's Moodurian
An Interview With Digital Integration Ltd (Part 1)
« on: May 22, 2016, 12:40:13 PM »
GameBytes Magazine (1994) Volume 13
by Andrew Stevens

Digital Integration are a medium sized software house/ publisher based in the UK, who have been operating in the PC games business for around ten years. They are the developers of the recently relased TORNADO flight simulation game which is distributed by Spectrum Holobyte in the US.  Their best known previous release is probably "F16 Combat Pilot'' -- a game originally released for the Amiga and Atari ST.  Tornado, according to the marketing pitch, aims to give an authentic simulation of the aircraft along with comprehensive mission planning and air-warfare simulation.

DI's main offices are located in Camberly, a medium sized town 30 or so miles west of London.  This is right in the middle of the Thames valley area.  This is the main U.K. geographic cluster for computer businesses and high technology in general.  It is a bit like Silicon valley  - highways, lots of dull looking industrial estates and expensive housing - but with pubs, brown beer, and plenty of rain.  DI's offices turned out to be in a surprisingly modest 1-story block in a by-no-means-exclusive industrial park.  The Interview was arranged with the four main developers working on the Tornado project:

Kevin Bezant --- 3D rendering engine and the simulation coding
Nick Mascal  --- 3D object design, simulation models, maps and manual
Robin Heydon --- Front-end and mission planner
Matthew Smith --- 2D Graphics

GB: What kind of background did you have before you worked on Tornado? Were all already experienced flight simulator programmers?

DI: (Robin) No not really.  Kevin joined the company at the latter end of the "F16 Combat Pilot'' project but the rest of us were really just looking for some interesting programming work.

GB: What made you go for the Tornado and the authentic/air warfware sim angle...

DI: (Kevin) The first thing we did was to develop the 3D engine.  It didn't actually become Tornado until Kevin had been working on it for the best part of a year.  This was just post Gulf war and the Aircraft had got quite a high profile.  We were also keen to do a british or European aircraft.  That and the fact that the Tornado hadn't really been done well before, and it being a British aircraft we ought to have reasonably good sources for it.  Authenticity also wasn't something that had really been done that seriously before.

GB: What about Falcon-3?

DI: (Nick) Well, for us, it is very much "the best of the rest''.  All kinds of things we feel are quite wide of the mark: stuff like being able to load up far in excess of anything, or the flight model.

GB: What about Falcon's "HI-FI'' flight mode?

DI: Well, according to the Spectrum Holobyte people we talked to the only significant difference over the standard model is the roll rate. It is still quite a long way from reality.

GB: As a European (Brit) it is good to see a decent quality aircraft simulation coming out from the UK.  Did you find it was an extra difficulty for trying to bring this out from the UK?

DI: (Robin) I don't think launching the product from Europe was a problem for us.  DI's has been going about ten years, and there are quite a lot of people out there who had F16 Combat Pilot.  It still sells and sells in the budget range.  Also, Tornado is being marketed by Spectrum Holobyte in North America so that definately wasn't a problem for us.

GB: Speaking of Spectrum Holobyte, there has quite a bit of speculation regarding their relationship to the product.  How much influence did they have on the product?

DI: (Nick) None.  We have read a few the things people have been presupposing but the game was concieved by us designed by us and finished by us.

GB: What most people have noticed is that the strengths of Tornado are often the weaknesses of Falcon-3 and vice versa...  did that just come about from simulating two very different aircraft?

DI: (Nick) Well, yes.  The actual sequence of events was more or less that we had the actual simulation part of Tornado - Kevin's stuff - reasonably well advanced and at that stage we went off to CES to blow the company horn.  Our director ran into the Spectrum Holobyte people there, who were impressed by what they saw and asked him to do a presentation.

GB: Something along the lines of "Gosh - the Brits can program''?

DI: (Grins all round) Well better than the Americans anyway ... Seriously: after some toing and froing the distribution deal evolved from that presentation.  Actually, they virtually offered to throw resources at us to get the program finished by a certain date. However, there was ultimately very little they could do.

GB: The old law about not throwing more programmers at project as a deadline approaches, presumably?

DI: Right. What they DID do was provide us with an experienced beta-tester.  They also did a lot of the later beta-testing for us.

GB: They presumably had got a lot of practice from the early problems they had with the launch of Falcon-3.

DI: Yes, something like that (Smiles).

GB: Is there any chance of DI's products ever tying in with SH's "Electronic Battlefield'' concept, or is the connection still just one of producer/distributor.

DI: (Kevin) Certainly not in the near future.  The 3D technologies used really are very different.  Also, while they do distribute for us in the US we really are still competitors in the market, albeit on VERY friendly terms.  We talk to them and they talk to us almost every day, but we keep our 3D technology secret as they do theirs.

GB: How do your sales split between Europe and the US?   I noticed in the (European) packaging there is a very noticeable pan-European feel to it.   German on the box, installation instructions in Italian Spanish and French.

DI: (Nick) From what I understand, it has turned out that the US market was far larger than the European one.

GB: Are there any plans for following up on the euro-plane theme in future products?  I would guess there isn't much chance of, say, Microprose doing a Dassault or European Fighter Aircraft simulation...

DI: (Kevin) I would say the chances of us doing a French aircraft are pretty slim (more grins all round).  Eurofighter would be nice but there is unlikely to be sufficient information to do it decently.

GB: In a UK games magazine you are quoted to the effect that an important constraint on your efforts was not being able to assume largish (say) 4M area of RAM.  What amount of memory did you assume?

DI: (Kevin) Nothing really.  The thing is, Tornado is not a protected mode program - it fits enitrely in the bottom 640K.

GB: It is? From the packaging I had somehow got the impression that it popped into some kind of 386 DOS extender.

DI: (Kevin) Well, when we started the program 3 years ago all those DOS extension products simply didn't exist.  I believe I'm right in saying this: if you take a copy of FLIGHT.EXE and copy onto an 8086-based PC it will run.  It doesn't even use extended memory to stash map data or anything.  The only advantage (in Tornado) of having more memory is for a disk cache.

GB: The landscapes included in Tornado are synthetic but quite densely populated.  Was this a deliberate design or was it forced on you by technological / marketing constraints (e.g. Germany's laws regarding glorification of war)?

DI: (Robin) There's a whole set of interlocking reasons for this. Obviously, small minimises storage requirements, also you've always got the problem of what to do with the edge.  Our solution was to design the landscape so it would tile when you wrap.  Also, we wanted to keep boring flying time to a minimum, so given that the 3D engine could handle a very high object density a small but dense map seemed a good idea.

GB: One of the areas where Tornado has had quite favourable comment is the frame-rate it manages given the amount of ground detail being shown.

DI: (Kevin) Well yes, but it depends on what you're comparing.  You have to compare like with like.  Because the Tornado is geared for low-level flight we've got heck of a lot of ground detail.  We're drawing a lot of polygons.  There's thousands of trees and things all over the map.  But then again in Falcon-3, say, they're shading the hills.

GB: On the other hand in Falcon the hills only seem to materialise about ten miles out...

DI: (Kevin) Yes, one of the things that was important to us was to have a long visual range so you didn't get this problem.  The range is actually 25 miles.  Similarly, all our stuff is 1 to 1 scale - there's no over-scaling to make things look prettier.  This also deals with the issue of why there isn't a shadow under the aircraft: at 1 to 1 you'd only see it you're flying REALLY low.  I know because we did try it.  It was almost always just a single pixel, by the time you got to see it you'd hit the ground.

GB: One of the major selling points of the game is the claimed authenticity of the flight model and weapons systems.  How much stuff did you manage to get out of the RAF guys to base your code on?

DI: (Nick) Thats actually quite interesting.  Dave Marshall (the project manager - ed.) did the modelling based entirely on publications in the public domain.  Research on open shelves.  Dave then implemented that (though not without some serious head-banging).  Our close contacts with the RAF only materialised after that work had been.  Basically, they looked at it and found that it was good.  We got one or two little tips, for example, the brakes could in fact hold the aircraft against full throttle.  We did have to tweak the roll inertia a bit but that was about it.

GB: The simulation model certainly bleeds energy in high G turns in an alarmingly realistic manner....

DI: (Nick) Yes, that's the difference between instantaneous and sustained turn rates. Dave Marshal calculated the Tornado's maximum sustained turn rate rather than its instantaneous rate and came up with a surprisingly low figure.  He then spoke to a friend of his who was actually a Tornado pilot and he said "yeah, that's about right''.  We got the cockpit layout and the way the weapons systems worked mainly from a simulator at RAF Honington.  Obviously, we simplifed a bit in some areas, the radar for example, but basically it is right.  For example, a real Tornado DOES have a maximum of 23 waypoints plus a target of opportunity waypoint.  The radar warning display we made up - it is always blanked off in photos or when uncleared people are about. The constraints on the hard-points are a slightly simplified version of what we got from talking to RAF ground crew.

GB: What about the manual control of wing-sweep?  Isn't that automated in the real aircraft?

DI: (Nick) Actually no.  We asked the same question of the RAF people. Apparently it had been tried on an experimental basis but it was found to have more disadvantages than advantages.  Basically, wing geometry wasn't found to be a significant addition to the pilots work-load, and an automatic system was just one more thing to go wrong.  Also, automatic control has the disadvantage in combat that it telegraphs your airspeed.

GB: One of the main complaints about Tornado is that the two player mode only supports head to head play.  Would allied play have been a lot harder to support or was it just a case of a looming deadline?

DI: (Kevin) It would be a major effort.  There are so many mobile objects and so forth there would be too much traffic to send across the RS232 unless there were some radical changes.

GB: Any chance of supporting allied play in a future revision?

DI: (Kevin) Not at the moment.

GB: The other funny is that wing-men cannot use the lob-toss bombing mode.  How did that come about?

DI: (Kevin) Basically, there wasn't time to do it for the release date. One of the difficulties was that it would cause a bad way-point timing problem - instead of flying to the way-point you break off some variable but hard to compute distance early.  It was a can of worms...

GB: There was much wailing and gnashing of teeth across the net by users when they found missile hits were invariably fatal.  Was this a bug or a feature?  The game engine can certainly simulate damage for AAA hits...

DI: There's quite an interesting story to this. Basically during testing the developers and beta-testing got so good at avoiding missiles they didn't notice it as a problem.  We've fixed this in revision A.

GB: The relative absence of bugs in Tornado and the products general stability has been favourably recieved.  What, do you feel, explains this - were the time pressures less than for some of the others, or was avoiding the complications of EMS or DOS extenders etc?

DI: (Robin) The major reason for that was that we had exactly one person writing code for each of the two sections of the program: the simulator proper and the front-end/mission planner. Kevin wrote all FLIGHT.EXE bar the sound drivers, Robin wrote the front-end apart from some last minute additions for the modem code.

GB: Yes, what happened with that bug in the Soundblaster driver?  Is there a problem with some of the clones not being quite clones?

DI: (Kevin) Well, the story there is that during beta-testing nobody found sound problems.  We then left all the machines running in the office overnight, and found that very occasionally things could hang. Eventually the problem was traced to DMA handshaking sometimes hanging, a hardware bug, and the sound guy (a contractor) programmed in a time-out that automatically switched to AdLib mode if that happened. However, once the users got using it that too turned out flawed.  We're thinking of completely rewriting that code for a later revision as the contractor has left...

GB: What kind of tools do you develop on?  A Novell network?

DI: Frisbee net!  We throw the disks... (chuckles)

GB: Can you say anything about the AI? For example, how the computer-controlled Interceptors work - Do they actually pull maneuvres, say yo-yo's to reduce turning radius, or change maneuvres in response to events?

DI: (Nick) Well, yo-yo's, scissors and the like are actually a natural consequence of appying even a very simple maneuvering algorithm. Basically, just try to point at enemy all the times and you get into a Scissors, that kind of thing.  The actual algorithms are more sophisticated than that though.

GB: Is there some kind of state machine ticking way in response to threats or changes in relative attitude and so on?

DI: (Nick) No it's not like that.  Effectively it is a bit like an analog feedback control system, its a completely general stateless algorithm which attempts to fly the aircraft.  It does behave according to the basic rules though.  It will always roll and then pull and so on. It also tries to stay within sensible G limits maneuvering rates for the aircraft involved.  If you go into Drone view mode you can occasionally watch two fighters in a dog-fight.  Go to one of the fixed missions and leave your own aircraft on the ground.  A good example is mission 6 in Warzone 3: there a pair of Tornado's take on some Mig-31's to clear a path for you through enemy airspace.

GB: Can you give some idea of how allied wing-men are handled in Tornado?  They seem to be given a little bit of grace because they are a bit dumb and don't maneuvre much against ground threats.

DI: (Kevin) What actually happens is that the game doesn't simulate stuff being fired at them.  The reason is that there are only so many slots for weapons available.  If you imagine drones someplace are dropping JP 233 sub-munitions and tracers are firing at them you might very soon find there are no slots when you want to fire.

GB: So how is the end result work out?  Sometimes wingmen don't make it - especially when the users mission plan is bad.  Presumably there must some kind of war-game style combat resolution depending on the surface AA threats?

DI: (Kevin) Yes, basically thats right.  If a wingman is exposed to AAA or radar or whatever that just increases its chance of being "shot down''.

GB: How are the ground forces handled in campaigns?

DI: (Robin) There is a war model running behind the scenes in the front-end.  It is based on supply and demand, forces and so on demand supplies supply routes and airbases can supply them.  If you knock out bridges and airbases supply levels go down and with it combat effectiveness.

GB: The mission planner has come in for a good bit of praise.  Did you have a local GUI expert or was it just inspired experimentation?

DI: (Nick) The latter really.  I specified it, Robin implemented, Mat generated its look.  We sorted out its actual functions by consensus and by building on existing stuff.

GB: What about the transparent windows?

DI: (Robin) It was a technique we'd seen on the PC, so we implemented it and it worked well.  The screen is 640x480 which is really too small to have a map and all the windows open.  So, having see-through windows means you can see the map without feeling too claustrophobic.

GB: Why is there the requirement that all aircraft must have the same number of waypoints between a split and a formate?

DI: (Smiles) Simpler code basically - we thought about it removing and basically it opened up a can of worms otherwise.

GB: One eccentricity is the complete absence of an in-cockpit rear view.   I know you can't see a terrific amount looking back in the Tornado, but I'm sure you can see something in the rear quadrant...

DI: (Kevin) Well, you may have noticed that in Tornado there is no momentary freezing of the program when you change view.  The reason for that is that we use a non-standard VGA screen mode which enables us to store all the cockpit graphics on the display card.  This means on a standard VGA card I can hold 128K of graphics.  However, for such a large screen, with the back seat and so forth there just wasn't room.

GB: That presumably was so it would run on even the worst el-cheapo VGA cards.

DI: (Nick) Yes, it will run on a completely standard VGA.  We tested it on the original IBM PS/2 with 256K of screen memory.

GB: What machines did you target as you developed the game.  It says 386 with 1M of RAM on the box...

DI: (Kevin) Well, when we started the program 386es were not the norm. In fact we only had one 386 (a 20Mhz model) in the whole of the company. Obviously, things changed rapidly.  It was only towards the end that we decided it would need to be a 386 or above.  Until quite recently we still had drivers for EGA,CGA,Tandy and even Hercules.

GB: So can Tornado actually be tortured into life on, say, a fast 286 with VGA?  It certainly doesn't say so on the packaging...

DI: (Nick) You can run it on 286.  We ran it on old 286 in the other room.  You would probably need DOS-5 though, to get amount of memory and things like that.

GB: Did you ever consider including a "padlock'' view like Falcon-3?

DI: (Kevin) We didn't consider it because we didn't think of it.  We knew there was something called padlock on Falcon-3 but at the relevant time during development we'd never played it.

GB: If you'd had "just six months more'' what would you have added?

DI: (Kevin) No, no no... it wasn't time it was memory - we'd filled it up!

GB: O.k. so what if you'd had "just 128K more''...

DI: (Kevin) (Laughs) A rear view of course!  Well, two rear views actually, one for the pilot and one for the navigator.

GB: Confessions time. Is there anything you'd really liked to have changed in Tornado but were essentially stuck with for the release version?

DI: (Kevin) Well, if there was one thing I would have done to cut out an immense amount of wingeing, it would have been to make the keyboard completely redefinable.  Every key was carefuly considered as it came along and we tried to find good reasons why it was where it was.  But when the Americans got hold of it they were completely baffled by our choice - basically because it just was not Falcon-3.  The classic example was the comment we got from Spectrum Holobyte that they felt it was a bug that you could use the U key or the G key to raise and lower the undercarriage.  They don't call it undercarriage, for them its the gear.

GB: Speaking of keys.  Is there any way to change the radar range without having to resort to a (shifted) mouse-button?

DI: (Nick) Not at the moment no.  However, in the revision after next we're going to make it so that when you turn the radar on the range corresponds to the current weapons you have selected.

GB: So you just select the weapon and then hit the "R'' key?

DI: (Nick) No, you press the fire button. The real thing has an air to air over-ride switch right by the throttle, hit that and everything jumps up ready with the right air-to-air radar mode and so forth.  That's what we're aiming to do with the fire button.

GB: Can we expect to see Tornado on other platforms?  A lot your stuff was, after all, originally launched on the Amiga...

DI: (Kevin) The Amiga/ST version is being worked on by a couple of guys right now, and should be out before Christmas.  Basically it has continuously followed on a couple of months behind the PC version.

GB: That would take some shoe-horning on a standard Amiga.  Does an 8Mhz 68000 really have enough power.

DI: (Kevin) Well no.  We are cutting things down a bit with more simplify options, but you have to remember some people do have fast Amigas.  Basically the simulator is identical.

GB: In the manual you state that "...designed to be the first of many new leading-edge products''.  Can you give any hints as to the likely shape of those leading edge products?

DI: (Kevin) Well the feeling is that we will need to go through line by line and put the 3D code into 32 bit protected mode.  Obviously, we will be doing flight simulators but we haven't figured out precisely what. We're still arguing!

GB: There is, apparently, a patch soon to be released to fix the first release version of Tornado.  Will this enhance the functionality in any way or will it be purely a bug fix?

DI: (Kevin) Yes, revision A is done and should be out in a few days.  It fixes the SAMs of course (see above -ed.) plus now you will be able to use Thrustmaster rudder pedals, and we've extended joystick calibration to all joysticks.  Originally it only did that for Thrustmaster because they're quirky.  Also, you'll be able to save the way-point plan for a campaign.  Plus a couple of fixes for little niggly bugs.

GB: So whats happening in revision B.  The soundblaster problems presumably?

DI: (Nick) Yes, the sound will a major issue on revision B, but there'll be quite a bit more.  The drone view will give information on the drone rather than your plane.  Then there will be the modification to the radar buttons, and a missile proximity alarm that goes off when a missile gets within a mile or two.  Also, the mobile SAM launcher is going to be changed to a SAM-13 launcher rather than Romb. Somehow during development we forgot that the "Romb'' launches radar guided missiles rather than the infra-red guided ones modelled in the game.

GB: Any idea when you'll be releasing it?  Two or three months?

DI: (Kevin) Oh no, three or four weeks, no more.  It depends if we decide to wait for the sound or put that on a later revision.  Oh, and there will be different versions of the patches for the European and US markets.  That's well worth knowing.  It's because the Americans have a different protection scheme.

GB: Any chance of mission disks and such like coming out.  High mountains perhaps, or exotic locations?

DI: (Robin) There seems to be virtually irresistible pressure to do Gulf War scenario.  Trees are not in, sand-dunes are!  I can show you one or two sample objects we're working on.  We've also got an idea to extend the mission planner so you can create complete missions, allied and enemy aircraft, everything.  The idea is you can go to a friend and say "this mission is hard - why not try and do it''.

GB: One of the real limitations of the PC for simulations is that ISA-bus bandwidth problems pretty much restrict you to 320x200 video.  I remember real pain moving from simulators at 640x256 on the Acorn Archimedes (ed. note - a mildy exotic British machine based on a RISC processor) to PC's.  The PC games were far deeper, but the display... What would you see as the minimum configuration needed to support a full SVGA (640x480) display on a PC?

DI: (Kevin) Well, basically local-bus and a fast processor. 

GB: A lot of people have big investments in ISA.  Do you see ISA plus an accelerator card as a viable alternative?  There are rumours that one of the Harrier Sims will support 640x480 on S3 and ATI-based accelerator cards...

DI: (Kevin) No not really.  The problem is that basically all the accerator buys you for simulators is a horizontal line-drawing function. That, however, can take a surprising amount of time to set up so that if you have lots of short lines, as in a simulator, it doesn't save nearly as much time as you'd think.  We've done some experiments with the S3 chipset.  Things are MUCH worse still if you try to do any kind of shading.

Also, one of thing we learnt from Spectrum Holobyte was that it is precisely the buyers of computer games that buy fast modern machines. People actually buy expensive games machines to run games and not the other way around.  For our next projects we'll be keeping that quite clearly in mind.

GB: Can we expect a SVGA-Tornado?

DI: (Kevin) No, almost certainly not.  For the existing Tornado 3D engine we could do it, but it probably would not be worth it without a new game.  But on a new game we'd almost certainly want shading.

GB: On a lighter note: are there any "Easter Eggs'' hidden in the game anyplace.  Things like the tap in Falcon-3...

DI (Nick): No there aren't.  We did toy with the idea of signing names in trees someplace on the map but basically we thought it would probably detract rather than add to the game.  What I did do in moment of madness was generate the night-time star pattern from the catalogue of visual stars in the Northern hemisphere.

GB: I had noticed some of the constellations looked rather familiar...

DI: (Nick) Yes.  The colours are right too - I wrote a little program that not only translated positions but also worked out an approximate visual colour based on spectral class and magnitude.  The only thing that is incorrect is that the stars are as seen from the celestial north pole.

GB: Thanks very much for the Interview.

At this point the Interview wound down and coffee was served, along with some demonstrations of the development tools built for the game plus some the draft objects for a possible Gulf War mission disk.  An interesting tidbit that came out of this was the extent to which little details in ground objects and aircraft had been modelled based on photos of the real life counterparts. Aircraft shelters, bunkers, etc have little doors or open entrances in the right places, and many of the aircraft have under-carriage down/up and wing forward/back variants. Some explosion craters actually have faces reaching below the ground surface.  A little play with "explore mode'' is highly recommended.

This interview is Copyright (C) 1993 by Andrew Stevens for Game Bytes Magazine.  All rights reserved.

« Last Edit: July 09, 2016, 03:43:07 AM by Frankie »
No reactions
No reactions
No reactions

After the Dogs of War are let slip, let us smoke the Pipes Of Peace.

Offline Asid

  • *
  • Posts: 25125
Re: An Interview With Digital Integration Ltd (Part 1)
« Reply #1 on: May 22, 2016, 03:54:28 PM »
Very interesting. Thanks for posting :)
No reactions
No reactions
No reactions

I stand against Racism, Bigotry and Bullying


    Actor practiced on "Tornado computer game at home" as part his TV role

    Started by Frankie

    Replies: 1
    Views: 3667
    Last post May 12, 2017, 08:21:03 AM
    by Lusik
    Flying the Tornado GR1 | John Nichol (Full July 2021 interview

    Started by Frankie

    Replies: 0
    Views: 537
    Last post January 05, 2022, 03:16:31 PM
    by Frankie
    Italian review of Digital Integration's 1993 Tornado. Blast from the past.

    Started by Frankie

    Replies: 14
    Views: 12283
    Last post February 23, 2016, 02:02:47 PM
    by Frankie
    In Memoriam of David Keith Marshall Co-Founder of Digital Integration

    Started by Frankie

    Replies: 3
    Views: 4046
    Last post February 01, 2019, 03:28:55 AM
    by Speedwagon
    Recent articles on Digital Integration's Tornado

    Started by Frankie

    Replies: 2
    Views: 2366
    Last post March 08, 2019, 03:18:11 PM
    by Rinix