MrFlibble wrote on 2021-02-17, 19:49:
As serendipity would have it, I've known about Great Air Battles by Charles B. Law for quite some time after having found two ea […]
As serendipity would have it, I've known about Great Air Battles by Charles B. Law for quite some time after having found two early demos at cd.textfiles.com, but only recently started researching the topic of FST games more after fstumbling upon the Wings of Thunder entry at MobyGames. As a mater of fact, I was going to start a thread with my findings here but you guys spared me the effort 😀
So here's what I have found so far, mostly via the Wayback Machine:
Additionally, I found shareware versions of several Viper Software games at archive.org, download links here (file names not original). The full list of Viper Software titles is available here, however without file names or places where these used to be found. A German shareware CD preserved what appears to be original file names for three of the games: Private Pilot Simulation (PPILT145.ZIP), BERLIN 1955 (BERLIN10.ZIP) and F-16 OPERATION BLACK DIAMOND (BDIAMON1.ZIP). I'm also tentatively assuming that EUROFIGHTER 2005 (not mentioned on the Wayback Machine Viper page) file was called EF2005.ZIP based on the folder name with the unzipped game that was indexed on a French magazine coverdisk.
Using Hallfiry's magazine coverdisk catalogue, I also found a newer demo of Great Air Battles dated 24 March 1995 (see attachment), but this is not the latest release. In fact, the Wayback Machine preserved a later demo v1.5a dated 3 June 1995 (gab-d15a.arj), but the archive is damaged, so I was only able to extract FILE_ID.DIZ, README.NOW and ORDER.FRM. An even later version, 3.0 is known to exist and was available here (file nor archived).
Not found in any form (yet)
Wayback Machine snapshots of mcnett.org actually contain a lot of preserved files (including some of those listed here), but unfortunately they were indexed in 2000 or roundabouts - and files from those dates are only preserved in the WM up to the first 1024 KiB, and everything above that is lost. I already downloaded a few ZIP files that were larger that this and now preserved in a broken/incomplete state. The best way is to list Wayback Machine contents for mcnett.org to retrieve them, but the downside is that you cannot tell which is a complete sim and which some kind of component (model, music, sound effect or even just screenshots), and also there are official FST patches released by Domark/Simis.
Some files appear to have been preserved here.
I also found a sub-site by Adrian Brown which had a very big collection of FST sims including files, but only lists of projects appear to have survived:
Since individual project pages are missing there is no telling what state was which project in, or who the author was etc. But at least it gives you some idea.
Oh, ask someone who posted some of that stuff on McNett (i.e....me). I could tell you the following:
a) Most of the sims referenced by Adrian Brown did exist at one point - some were in that incoming directory, others were his cached copy of the stuff that we were working on. Just assume that anything over, say, 240k is an archive containing sims (potentially). Some of the stuff mentioned were dead ends or only existed as artwork + ideas. The carrier landing demo for example is the same as Cavanugh....which was kinda weird because I originally named it Cavendish (after the banana). What was it? A really tricked out flight model (FMD file) attached to an Su-33 shape which is being flung onto a carrier deck. You'll have to land it (and I made the brakes ridiculously overpowered it simulates an arrestor hook. The same trick was used to implement STOVL flight models like the Harrier). It was quite a bit of fun to play (but twitchy as hell with a joystick), but given the time and resources, it can be easily duplicated.
How would you tell what is what? FSDs are polygon meshes/3D object files. PCX are likely cockpit backgrounds with an alpha channel, there are some artwork (often references in discussions) and there were some archives (anything over 1MByte is a lost cause). Jans B converted a few of his ridiculously high polygon shapes into VRML, and there were code to convert Quake MDL 3D models into FSDs, FSDs into VRMLs, and vice versa. You might see some snapshots of sims and 3D objects that were released or were mentioned in passing
b) The Blade Runner stuff was originally DUO (Lucas Olestri)'s idea.
Some new art assets were added by others. It's a fun one - flying a Blade Runner spinner in a cityscape even if there were no purpose...makes for a rather relaxing experience. The original version was nb.zip (which was unfortunately truncated by archive.org) with many of the assets borrowed from his SHADO UFO sim (which thank goodness was cached). The assets from the BR Revolution "patch" were retained, which was small comfort.
Heh. That train in the Blade Runner Demo were based on the Tyderium Shuttle (trn.fsd) from one of the Star Wars sims (Coruscant Sweep?). Luckily for McNett the IP holders still believed in the principle of fair use back in 2000.
BladeRunner 2 was more of a dense cityscape demo...
and BR2025 (see RAR below) was supposed to be a more modern version of the demo with a different art direction.
c) Most of the star wars sims don't/can't run because they use the later 1.4x version of fly.exe, and those require DOS4GW (not bundled). FST is terrible emulating a space sim (they had a "brick" flight model that ignores the usual physics calculations, but the sim still subject the projectiles to gravitational pull, so firing simulated "lasers" don't really work here. There were supposed to be builds of fly.exe (after the sourcecode release) where new features were implemented...like no-drop bullets and rockets (lasers and particle cannons)?
d) The one you really want to look for (in addition to the Avro Arrow sim or the updated Gripen sim) is the Nighthawks in Baghdad sim - it featured a very good Iraqi cityscape. I could tell you that Pearson's sim features Avro Arrows intercepting Tu-16s and Tu-95s traversing Churchill, Manitoba and ends with a pretty somber scene of the Arrows getting scrapped within some Flash-like menu thing. I only wish that someone kept a copy - It's a good one. If you try to play the J29 Congo sim, it'll fail due to a screw-up on the cockpit FGD file. The easiest way is to copy a good FGD (and its associated background PCX) from a working sim. If you pulled the Gripen sim, look inside for COCKPIT.FGD/TILLF.PCX, copy them to the Congo directory, then edit world.fst there. Look for the player object and swap J29.FGD for COCKPIT.FGD, save, and the sim will start to work.
So, was FST good? Eh, it was for a time. Out of the 5-and-a-half utilities (world.exe (for making the game arena map), model.exe (flight model generator), colours.exe (for working with the limited palettes supported by fly.exe - it has a limit of ~239 colors that were gourard shaded, and 10 or so that were not shaded (so they can act as "light panels", so to speak) and a few reserved for explosions and cockpit instruments, cockpit.exe (for putting instruments onto 256 color PCX files (with an alpha channel for simulating cockpit windows), shape.exe (a 3D shape designer/editor)+view.exe (the Windows GDI based 3D object viewer, very clunky by modern standards), it was shape.exe that as considered well designed. It actually worked quite well to design 3D meshes of objects for realtime 3D use, and with good planning and modularity you can design some excellent meshes for it that were more precise than, say, early versions of blender. It supports grouping, extrusion, and the usual rotate/mirror/skew functions that people take for granted...and this was made back in '93!
View.exe had the unfortunate distinction of being designed with a limit of 511 vertices per level-of-detail. If your 3D mesh has more than 511 vertices the shape will not render correctly. The game engine (fly) had some good attributes but it was limited in what modules were implemented. It does not have a radar module with selective targeting or lead calculations for gunsights, situational awareness aids used in popular flight sims (like the multi range radar in USNF or the 3D hemispheres in TIE fighter) do not exist, and the pathing/AI logic were rudimentary. The collision box implementation and damage modeling were...primitive. The polygon drawing routine implement the painter's algorithm, which means that overlapping polygonal segments do not get obscured correctly (no z-buffering)...which is a pain to design 3D shapes to work with. You almost always have to break long thin tubes (like fuselages) apart so the intersections with wings or intakes do not cause issues (you see some in early FST designs). The other issue is to be aware of non-coplanar polygons - FST Shape uses quads, so it's possible to create non-coplanar polygonal meshes which causes drawing errors. What's a good example? The Spinner cockpit surface in the BR2025 demo, or the nose on the Imperial shuttle. That's why the later FST shapes tends to use triangle strips to prevent this from happening (since triangles can never be non-coplanar)
Why did FST fall out of favor? I blame Simis for not investing further into FST. When people bought FST back in 94-96 most PCs were 486/early Pentium based, and you are lucky if you had an S3 to accelerate line drawing. So back then a 3D object with, say, 4-5 levels of detail, a max ~260 polygons per level (co-linear rectangles are allowed), flat shaded/gourard shaded polygons, limited color palettes and confetti damage/destruction model with less than 12000 vertexes total were perfectly acceptable. When people move to perspective corrected bitmapping in 256 color palettes (or 16/32 bit color later), this was never added to a marketed version (well, there was that experimental all-elbows FST2 utility that let you map 256 color textures to the mesh, which is supported in fly 2.x engines). The guys in the FST mailing lists spent hours ripping apart Simis' Absolute Zero demo (which was FST/fly 2.x engine based and had texture mapping) so they can figure out how to implement sims based on it, but in the end, just gave up. We knew how it works but putting in the effort to make new sims based on it was...simply too much work. Greg O and McNett got versions of Fly that takes advantage of Rendition RRedline for Greg's Verite 1000 (Redfly) and OpenGL targeting McNett's 3dfx Voodoo (Openfly), respectively. Those could support non-palette textures and ran on Windows...but you’ll need non-FST tools to author for it. It's really too bad that even after the fly.exe sourcecode was released that many features were not backported for it, nor were anything new besides eye candy. The fly.exe sourcecode was a masterclass for someone who wants to implement a retro flight sim, but lots of work is needed to make it fun. If I were to spend time on it today it would probably be going towards adding IK/modular components to the FSDs, better damage models, improve the current sim modules, like the radar functionality....and integrate LUA support for better scripting/AI capabilities. We kinda saw that in Simis’s Terracide (which still uses FSD files to define 3D objects)...but it’s too bad those features were not sold to us.
Simis talked about releasing FST2/FST98 back about 23 years ago, which never happened. Domark got bought out by Eidos, Eidos cancelled Confirmed Kill (later discussion), Simis left Eidos (Simis had a side business doing medical imaging and was not dependent on EIDOS to survive, so when EIDOS went south they did a management buyout and left) and then being hired by Microsoft to do Train Sim/TS2 were a big part of it. The game engine wasn’t entirely the issue (it could definitely use some work there though) - there were need to improve the authoring tools. I think some of the guys got sick and tired of waiting for a future that never came, and went onto other things.