FM Testing Gamefilm

[ShrikeHawk 384]

I don't know the testing routine that otheres use for their FMs (Flight Models). I flight test a LOT. I like to create a mission of 8 versus 8. This gives me a better sense of how the AI is handling the FM. In 4 versus 4 engagements the AI tends to hold back and let the User  run the fight. In 8 versus 8, the AI steps up and fights harder.

In the vid below, I had been resisting changing the CDL and the CMQ values on the Yak-9 outerwing. The speed was good, but turning was lethargic, I finally gave in and changed it slightly. Got the combat results I was looking for. Of course FM mods is a constantly balancing act of many parameters. Combat results were good, but watch the speed of the User aircraft carefully. It's awfully high.

 

I felt the Yak was now too fast. Normally, I only make small changes and immediately flight test. This time I increased the CDO in the fuselage. The better turning would remain but the average speed through the flight decreased. I also pulled back a bit on my highly energetic Mustang. I increased CDL in the wings and outerwings. Yaks and Mustangs matched up pretty closely before. I was concerned the Mustang might be overmatched after these changes. I was wrong. So far, the Mustang is a more stable gun platform, and its more guns means more hits.

 

Lastly, I jumped back in the Yak after slowing it down. This change didn't diminish combat capability. In fact, as I sometimes find, the AI is better able to line up targets if the closure speed isn't too high.

[r2dter 0]

How about letting *cough* tye combatace community judge your FM handiwork for themselves.

[Do335 382]

ShrikeHawk,

 

If you got the Yak-9 FM from wings over korea pack Baffmeister had pointed out to me that I forgot to use the latest version. Basically CLa on the wings needs to be increased, although i reckon some CDL reduction would have the same effect and there's nothing like enjoying own work. But mostly I see it has prompted you to make 3 videos so I rejoice:D Thanks for visiting Korea ShrikeHawk!

[ShrikeHawk 384]

ShrikeHawk,

 

If you got the Yak-9 FM from wings over korea pack Baffmeister had pointed out to me that I forgot to use the latest version. Basically CLa on the wings needs to be increased, although i reckon some CDL reduction would have the same effect and there's nothing like enjoying own work. But mostly I see it has prompted you to make 3 videos so I rejoice:D Thanks for visiting Korea ShrikeHawk!

 

Thanks, but I'm using Pasko's Yak-9 from way back when. I believe the one from the WoK is Cocas' work. I always liked Pasko's Yak and had gotten used to it before so I've been flying it lately. I thought the FM is very good on it. Trouble is, I've tweeked the TW Mustang into a pretty vicious killing machine. In the vids, you might've noticed the Mustang AI pulled some very sweet maneuvers. I've played around with the FM until the AI fly it brilliantly and very aggressively. So even though Pasko's Yak FM was very good, the Mustangs were killing it far too easily. I could've dialed back the Mustang to match appropriately or change the Yak. I chose to change the Yak. I probably fiddle with the FMs too much, but that's where I get a great deal of fun out of the game.

[Do335 382]

Thanks, but I'm using Pasko's Yak-9 from way back when. I believe the one from the WoK is Cocas' work. I always liked Pasko's Yak and had gotten used to it before so I've been flying it lately. I thought the FM is very good on it. Trouble is, I've tweeked the TW Mustang into a pretty vicious killing machine. In the vids, you might've noticed the Mustang AI pulled some very sweet maneuvers. I've played around with the FM until the AI fly it brilliantly and very aggressively. So even though Pasko's Yak FM was very good, the Mustangs were killing it far too easily. I could've dialed back the Mustang to match appropriately or change the Yak. I chose to change the Yak. I probably fiddle with the FMs too much, but that's where I get a great deal of fun out of the game.

Ah I see, I switched over to Pasko's as well due to it more fps friendly. From my own experience the mustangs are surely a bit weak also because of a small airframe... so it's more vulnerable to gunfire. The Yak is no better though, it's quite fair to balance them I reckon.

[ShrikeHawk 384]

Ah I see, I switched over to Pasko's as well due to it more fps friendly. From my own experience the mustangs are surely a bit weak also because of a small airframe... so it's more vulnerable to gunfire. The Yak is no better though, it's quite fair to balance them I reckon.

 

That's the downside to inline engines. Much more susceptible to battle damage. In the case of Yak-9 versus P-51: my read on this is that the Mustang has a speed advantage at most altitudes depending on which Yak-9. Mustang has the turn advantage at high altitude (mostly because it can build up a lot of speed for a fast turn). At medium and low altitudes, the Yak-9s turning ability outshines the Mustang. These are really two different airplanes built for different roles. But in Korea, they were holdovers from the previous war. They would meet, not in their originally intended, established roles. That's why this fight, and modeling the relative FMs right for both, is so interesting to me.

[+streakeagle 871]

My preferred method of testing FMs is simply to have known performance data points, better yet complete charts, and then turn debug mode on and see for myself how the net lift and drag values compared at that those points.

 

My AIDE tool allowed me to convert SFP1 flight models into standard performance charts found in flight manuals. I would simply tweak the values until the output of my program fit the charts. Unfortunately, the last version I made/released to accommodate some changes TK made to the flight models introduced bugs that I could never fix making the program unstable, most likely a memory leak or worse. The program was large and my free time had dwindled. I always dreamed of a day when I could start from scratch and make a newer better version.

 

NASA has CD0 plots for many aircraft of the 1960s/1970s, so it is fairly easy to get that value correct. Many flight manuals for jets have V-n diagrams that let you get Clmax vs speed and altitude. Many have level acceleration times from one mach number to another (i.e. Mach 0.5 to Mach 1.2), which provides T-D/W (specific excess thrust). If the sustained g performance charts are available in standard height vs Mach format, you basically have everything you need to get complete lift, drag, and thrust tables. The difficultly is in proportioning those numbers to various parts of the aircraft.

 

There is a program called DATCOM based on all USAF performance prediction techinques that can be extremely useful for SF flight modeling: http://www.holycows.net/datcom/

 

Of course WW2 aircraft performance was never documented as well as jets. The pilot manuals are generally devoid of performance tables. Military testing was sometimes very subjective... unclear what fuel was used, weights, and other critical test conditions. The performance of particular engine and prop combinations are hard to determine. Jets are generally easier to model than prop planes... at least at subsonic speeds.

Edited August 7, 2015 by streakeagle

[Do335 382]

My preferred method of testing FMs is simply to have known performance data points, better yet complete charts, and then turn debug mode on and see for myself how the net lift and drag values compared at that those points.

 

My AIDE tool allowed me to convert SFP1 flight models into standard performance charts found in flight manuals. I would simply tweak the values until the output of my program fit the charts. Unfortunately, the last version I made/released to accommodate some changes TK made to the flight models introduced bugs that I could never fix making the program unstable, most likely a memory leak or worse. The program was large and my free time had dwindled. I always dreamed of a day when I could start from scratch and make a newer better version.

 

NASA has CD0 plots for many aircraft of the 1960s/1970s, so it is fairly easy to get that value correct. Many flight manuals for jets have V-n diagrams that let you get Clmax vs speed and altitude. Many have level acceleration times from one mach number to another (i.e. Mach 0.5 to Mach 1.2), which provides T-D/W (specific excess thrust). If the sustained g performance charts are available in standard height vs Mach format, you basically have everything you need to get complete lift, drag, and thrust tables. The difficultly is in proportioning those numbers to various parts of the aircraft.

 

There is a program called DATCOM based on all USAF performance prediction techinques that can be extremely useful for SF flight modeling: http://www.holycows.net/datcom/

 

Of course WW2 aircraft performance was never documented as well as jets. The pilot manuals are generally devoid of performance tables. Military testing was sometimes very subjective... unclear what fuel was used, weights, and other critical test conditions. The performance of particular engine and prop combinations are hard to determine. Jets are generally easier to model than prop planes... at least at subsonic speeds.

eh I like all good FMs, especially ones made with love, some may not be extremely extremely historically accurate but can't really complain... of course hi-fi FMs would be very awesome. In the (likely very rare) event that you get some free time, streakeagle...

 

 

And Shrikehawk please keep the vids coming man, SF videos are so scarce these days...

[ShrikeHawk 384]

Gawd, I would love to have the CD0 plots for the ww2 aircraft. As it stands, I'm going to wwiiaircraftperformance often to get the speeds at various altitudes. As it stands on many factors I'm trying to get aircraft performance, relative to another correct. And certainly recreate the combat environment as authentically as possible. My Spitfire IXc can easily turn away from an Fw-190A-4, but the Fw190A-4 easily out rolls the spit. In flying the Spit, an angles fight is its best fight, but trying the same with the Fw-190 against Spits is just about certain death. In the Fw-190 I'm doing BnZ and in that realm it can eat Spits for breakfast. I just gotta figure out how to make the AI do that.

 

By the way, there's an ini file that lists AI parameters like chance to make a hard break, chance to attack vertical and so on. Does anyone remember what the name of that file is?

[Wrench 9,850]

AircraftAIData.ini

 

it's in Object001.cat

[ShrikeHawk 384]

Awesome. Thanks Wrench!