Ain investigation into the AI behavior of all TW series 1 simulators

[Ticket1]

Hi folks,

 

I've decided to create this thread for primarily the following reasons:

 

(1) in response to one of a threads found in SimHQ ->  http://simhq.com/for...tml#Post3055729 

 

(2) show and reveal non-documented features and characteristics of AI aircrafts

 

The purposes which naturally follow are

 

(a) help TW Strike Fighters series 1 simmers to understand better the where we all stands from a user's perspective

 

(b) provide additional information to fill the holes for TW SF series 1 simmers who may be interested in customizing aircrafts characteristics to suit their expectations and needs

 

Action items which arises from the above:

 

...pending

 

 

I will keep everyone on this forum posted as soon as I've come up with new findings.  Please stay tuned.  Thank you.

 

 

Tk1

 

 

 

[Slartibartfast]

Well SF1 was not designed to run on anything after Win XP... So its possibly everything but saying that quite a lot of software had problems with Vista along with the users. This was because Vista is very intensive on the PC so takes resources which a game could use instead.

 

So he probably means the loading etc not the game play.

[Lexx_Luthor]

Vista Shmista. I'm thinking software devs (including TK), panicked over Vista, not realizing it would fail and be replaced by ~7. SF1 is lightning fast on Win~7. I know.

Edited September 9, 2010 by Lexx_Luthor

[Lexx_Luthor]

That said, some say ~7 is just Vista renamed. I don't know and never will know, but SF1 works the shnizzle now.

 

--

 

This sounds like a good idea Ticket. I assume SF1 2008. I think alot of big AI features came in 2006 where I'm stuck at, and then some in 2008. So I'm interested in this even though I'm 06. Whta mission editor do you use?

[Wrench]

Win7 = Vista Fixed.

 

1stGEns run just fine on Vista and 7; one justs needs to cure the HUA seemingly so prevelant.

 

06 levels games require the terrain shaders to be turned off, or you CTD on mission exit

08 levels fixes ALL that (but hoses my precious WW2 FMs -oh well!)

 

And just so's you all know, NextGens run just fine in XP too. Was running them no real issues on my old XPS (3.46, 2gig ram, 512 video). Just had to turn down the eye candy a bit. At 08 levels, with the new W7 machine, can run them maxed out with no problems of ANY kind

 

------------

Personally, I think folks are getting their sphincters in a uproar over nothing. All the 'real' brains in the modding scene are HERE at CA. I wouldn't let what happens over at SHQ bother me one little bit. Seems like a standard or garden variety "sour grapes" complaint.

 

And, "everyone who's anyone" in the modding community -all the Big Names-, has pretty much moved on the SF2; as that's the way to go. (expecting for my precious WW2 mods -which I'd love dearly to get into 08/SF2)

 

All this is just a storm in a teacup.

 

wrench

kevin stein

Edited September 9, 2010 by Wrench
corrected OS listing

[Fubar512]

Lexx, Wrench and folks,

 

I could be wrong, but I'm guess the sub-par speed that PFunk was talking about actually refers to the speed of the aircrafts.  As I'm doing all these tests for the investigation, it's came to my attentions that aircrafts do fly like in SLOW MOTIONS - both from inside the cockpit as well as from external view.  Or I've got some problems in visual recognition?

 

Ticket1

 

And you base this on what, experience? The sense of speed one gets appears realistic to me.

[Wrench]

which I can't understand the rationale behind at all: why on earth would the latest and the final version of SFP1, WoV and WoE have all the DATA.INI, AVIONICS.INI and LOADOUT.INI files all hidden up in the OBJECT.CAT file? They are all present in the vanilla versions of say SFP1

 

Umm....that's absolutely incorrect .... Since Day 1, SFP1, NOTHING was ever in the aircraft folders, other than the main ini, the skin folder with it's associated textureset and decals ini, and skin bmps.

 

The game is designed to 'pull' the necessary items FROM the various catalouge (cat) files, just like Windoze has been doing for years with it's CABs.

Why do you think one of the VERY first utilities built for this game was Skypat's Extractor?

 

Because nothing was accessable before without it.

 

wrench

kevin stein

[Fubar512]

Oh! No offense, but I'm comparing it to H.A.W.K. Aircrafts in H.A.W.K. fly just like its intro movie.  I guess PFunk was actually referring to that aspect of the simulators.  I've done a small experiment.  By the way, I've downloaded aircrafts like mad for the last couple of days.  I put myself into the cockpit of an F-18F.  A ground target hangar was 20nm aways from where I was, and the speed of the F-18F was 400knots, ceiling 12,000.  It took me a real-time around 8 minutes to finish the trip  Should it not be 0.3 minutes?

 

0.3 minutes? More like 3 minutes. Around 8 minutes? Did you measure it? If you make a rough guess like that, then you cannot blame someone for dismissing your observations.

[malibu43]

Oh! No offense, but I'm comparing it to H.A.W.K. Aircrafts in H.A.W.K. fly just like its intro movie. ...

 

 

 

Ah yes... H.A.W.X... The gold standard to which all other arcade - I mean flight sims - should be compared to.

[Fubar512]

 

No Sir!     I measured it with my stop watch in my Nokia 6700 Classic phone. Sorry Sir but there it is.

 

Then obviously, you don't know how to read a watch. I used a Phantom with it's radar in ground map mode, and waited until a ship target appeared on the screen with the range set at 10 nautical miles. It took 101 seconds at 390 Kts ground speed, before I overflew the target, which was sailing at 30 knots on the same heading that I was.

Edited September 9, 2010 by Fubar512

[FastCargo]

Oh! No offense, but I'm comparing it to H.A.W.K. Aircrafts in H.A.W.K. fly just like its intro movie.  I guess PFunk was actually referring to that aspect of the simulators.  I've done a small experiment.  By the way, I've downloaded aircrafts like mad for the last couple of days.  I put myself into the cockpit of an F-18F.  A ground target hangar was 20nm aways from where I was, and the speed of the F-18F was 400knots, ceiling 12,000.  It took me a real-time around 8 minutes to finish the trip  Should it not be 0.3 minutes?

 

Your math is WAY off.

 

Lets assume you meant you were at 12000 ft MSL, 400 knots IAS/CAS. For a target that was 20 NM way.

 

Assuming that the CAS/TAS difference on a standard day is 5 knots per 1000 ft, that means you were doing 460 TAS.

 

I'll even give you a fudge factor and make it an even 480 TAS. Assuming no wind, that means your ground speed (GS) was 8 NM per minute (easy math).

 

That means for a 20 NM target, it would take you 2.5 minutes to get there.

 

If you were trying to do it in 0.3 minutes (about 20 seconds), that means your GS was 60 NM per minute, or about 3600 Knots.

 

You would have to have been traveling at about Mach 6 to make it in 0.3 minutes.

 

 

Now, you could argue with me, but considering I've flown at Mach 1.1 at 300 feet in real life, and I can do these kind of speed calculations in my sleep because I did them ALL THE TIME, I'm pretty sure I have a little credibility on the subject.

 

FC

[FastCargo]

Then obviously, you don't know how to read a watch. I used a Phantom with it's radar in ground map mode, and waited until a ship target appeared on the screen with the range set at 10 nautical miles. It took 101 seconds at 390 Kts ground speed, before I overflew the target, which was sailing at 30 knots on the same heading that I was.

 

Ed, sorry man, but you're wrong.

 

390 Knots GS - 30 Knots GS = 360 Knots GS = 6 NM per minute.

 

10 NMs should have taken 1.66666 minutes or 100 seconds...not 101 seconds.

 

 

Okay, you can smack me now.

 

FC

[Wrench]

One is also forgetting that all the stock maps are 63% Real World ™ size, and there are MANY inherent scaling issues with them.

 

it's been so long since I had a "stock" 06 or earlier (excepting my beloved WW2 installs -but all stock aircraft are deleted), I did't actually remember that.

 

wrench

kevin stein

[FastCargo]

Wrench,

 

The scaling issues shouldn't matter really, because the units don't change. The only difference for the most part should be the comparison of the real world distance vs the map counterpart (ie a target that is 100 NM away in real life shows at 63 NM in the sim, but the aircraft should still cover the 63 NM distance in the appropriate time that it's real world counterpart would cover 63 NM).

 

Mainly what the 63% affects is fuel usage and total time airborne, since you're not traveling as far as in real life.

 

Also, just as a test, I did a timed run in an RF-111A, 6000 MSL, 480 GS (using the enhanced HUD info available in Debug mode). On a target that showed on a 25 NM ground map scope...it took just a tick over 3 minutes to get there. This was verified on my watch and on a kitchen timer. I'm not seeing any sort of sim slowdown at all.

 

I'm running on an Intel T8300 (2.4 GHz dual core), 4GB RAM, Win7-64, NVidia 8600GT, SF2 at June 2010 patch. Hardly a fire breather, but runs well enough.

 

FC

[FastCargo]

Final note, and this has been discussed before reference the 'sensation of speed'.

 

The problem is one of resolution and FOV.

 

The default view of most of the cockpits is so that the instruments and HUD can be read on an average resolution computer screen.

 

The problem is that the FOV to do this is about 2/3 to 1/2 of the FOV the human eye naturally sees. Because of this 'zoomed in' effect, ground rush is significantly slower than it should be.

 

To get a truer sensation of the proper FOV, zoom out. As a technique, I zoom out until the canopy bow takes up about 1/2 of the lateral length of the screen ie:

 

Left edge of screen - 1/4 screen - Left canopy bow - windscreen - Right canopy bow - 1/4 screen - right edge of screen.

 

You'll find that will give you an increased and more representative sensation of speed without resorting to arcade trickery.

 

FC

[Lexx_Luthor]

He/She may not know about the game's debug data.

[Slartibartfast]

Folks,

 

I've found this video clip on Youtube.  

 

It does give us some "sensation of speed". And I hope this is not the real fighter:

 

 

Ticket1

 

Okay the second one is from a Radio controlled model... so the speed scale is going to be a bit messed up I think...

[Wrench]

You mean you've deleted all the stock aircrafts? For what reasons if I may?

 

Re-read the statement; it's pretty self explainatory. (ie: the hint is "WW2 Installs"). What the hell do I need Mig's and Phantoms cluttering up the /Aircraft folder for, when the 'average' WW2 install -- and there's 3, but realy should be 4 -- has 110-150 aircraft? Same goes for 98.6% of the ground objects (excpeting fuel truck, cargoships -- even the tanker is wrong for 1939-1946)

 

wrench

kevin stein

[FastCargo]

TA radar is.

 

TA is a manual mode.

 

The proper term you are looking for is Terrain Following (TF).

 

TF is not implemented, neither the radar or the autopilot (though the AI does a pretty decent job of staying out of the weeds).

 

Also, those videos are inaccurate - the first was filmed with a fisheye lens, so the FOV is higher than normal human FOV and the aircraft is flown deliberately near sonic speeds and at lower altitudes for the show.

 

The second is an R/C model...so the sense of scale is off.

 

If there is a weakness in the 'sense of speed' in the TW sims, it's the terrain. The more detailed the terrain (larger amounts of smaller objects are better), the better the sensation of speed. The TW terrain has always been a weak point in the sims because it is low resolution. It however, is not inaccurate in terms of rate of speed or distance covered.

 

FC

[+Florian]

Also, those videos are inaccurate - the first was filmed with a fisheye lens, so the FOV is higher than normal human FOV and the aircraft is flown deliberately near sonic speeds and at lower altitudes for the show.

 

FC

I second that!

 

Look at the sequences filmed from ground. Compared to that in-cockpit view, the aircraft looks like creeping through the sky. But compared with in game flybys you´ll notice hardly a difference. That´s the best proof for the speed in game is absolutely correct!

[Derk]

Wrench,

This was verified on my watch and on a kitchen timer. I'm not seeing any sort of sim slowdown at all.

 

I'm running on an Intel T8300 (2.4 GHz dual core), 4GB RAM, Win7-64, NVidia 8600GT, SF2 at June 2010 patch. Hardly a fire breather, but runs well enough.

 

FC

 

Hey FC, what are the specs on the kitchen timer you're running ? (watch is not terribly important.....)

 

Hou doe,

 

Derk

[+Florian]

Hey FC, what are the specs on the kitchen timer you're running ? (watch is not terribly important.....)

 

Hou doe,

 

Derk

They even have a picture of his kitchen timer: Look HERE

[Icarus999]

If the sensation of speed is such an issue for you try running the game at X2 time compression. Also i would not listen to Pfunk .

[Lexx_Luthor]

FC::

It however, is not inaccurate in terms of rate of speed or distance covered.

The debug data output shows TK's sim is spot on. Combine that with free camera distance measures and we can show this. Most however, never think of doing this.

 

There is no AI terrain avoidance radar. Infrared perhaps, or night owl vision, whatever. AI avoids terrain no matter the airplane and day or night, but does it very well indeed in this game! However, it can be called "TA" for purposes of, say, a B-1 campaign....personally I would pefer a low level B-47 campaign.

[Fubar512]

 

The DATA.INI file controls the AI behavior whereas the AVIONICS.INI file doesn't play a role in that part of the simulator, am I correct?  Thanks in advance!

 

Ticket1

 

Aircraftobject.ini = global effect on AI (I advise against modding this file, as it's not been necessary for over two years now.

 

AircraftX_data.ini = effects only that model, and can be tailored specifically to take advantge of that model's capabilities (or to minimize a model's shortcomings).

 

The AircraftX_avionics.ini is strictly for the player's benefit, controls how the avionics behave in the player's cockpit, and has absolutely nothing to do with the AI.