SC7500 AI Performance Comment on Talon's Skip Bombing Thread

[Devildog73]

SC7500,

Concerning your information:
How does one go about changing the Zero's aileron problems in a dive to ACCURATE?
The Zero had an inherent aileron problem in a dive, which is why our fighters were so successful later on against them. We had procured an almost prestine Zero softly crash landed in soft soil, sent it to California, repaired it and tested it and our fighters against it.

Our engineers found that in a dive, the Zero's ailerons became unstable and inoperable, sometime even flying off the wing due to the flutter. No ailerons was a big problem for a fighter!!

This is why most of our fighters, to escape a Zeke on their six, DOVE HARD. Even the early Wildcats could out dive a Zero and pull out of the dive alive.

I have noticed in my Bong and McGuire missions that the AI stock Zeros out dive me and never come apart.

THIS IS NOT ACCURATE.

 

[sc7500]

Don't Shoot The Messenger !

D-Dog et All;

I included that info last night as a SERVICE, not as proven facts - I have several dozen ancient discussion files pertaining to AI modification that I will upload, but I can neither prove nor disprove their accuracy or effectiveness.

The original documents were posted between 2002 and 2008 - like I stated, I found them on an older back-up drive - and were written by such luminaries as Goopney, Pen32Win, and other early CFS2 "mechanics" whom the more ancient of our Basement dwellers may remember.

That said, unless we develop a method of "writing in" an aileron failure at predetermined stress levels [I'm not fluent enough in AirWrench yet to know if this is even possible - might be a job for Caleb Flerk], total flight accuracy is still a Holy Grail.

That said, I will collect ALL of my new-found info and post it as a zip file for everyone to peruse and explore later today. Some of this stuff is just too interesting to keep hidden !

Best
SC

 

[bearcat241]

First of all, we have to pay due respect to the CFS2 AI engine and its AI flight modeling. Unfortunately, its a cheat, and its very difficult to program (read tweak) around it for 100% accurate enemy AI performance.... The AI engine automatically reduces weight and adds agility to any air file when applied to an AI enemy aircraft, but not so fast when the same airfile is applied to the player or his/her wingmen. And it seems to cheat more with each increase of enemy skill level from rookie to ace. That being said, the prime controller for ailerons at speed is record 518 of the air file - AILERONS CONTROL FACTOR FACTOR VS IAS (indicated airspeed). When you single left click this record you get a graphical chart which displays performance parameters as a line that gradually descends to the right. This descension indicates a control degradation factor as speed increases. In the pic below you can see the #518 from the air file of one of my Bearcats. The control drop factor is a gradual sloping as opposed to a severe drop as you see in the stock Zero's file in pic #2. So both appear to accurately display the aileron vs IAS parameters of their respective fighters. The question is how much will the AI engine override these parameters when engaging these two birds as AI enemies. In some cases the engine seems to just toss out such control factors completely and have the enemies performing in "unhistorical" manners. The hard part about tweaking here is that you can only hex edit this value as opposed to the easier method of changing numerical values in a record as seen in other records of the air files.

Now here's where it gets really interesting for me. Have you guys ever noticed how tame enemy AI perform with 1% air files? Even the player and wingies are more restricted in performance parameters, especially in control surface factors in high dive speeds - you get too fast in certain AVH models and your ailerons and elevators lock up til you ease off the gas. In a dive you just crash before you can regain control unless you use flaps or dive brakes. Its because the 1% workbook utilizes unused factors in the air file that the developers bypassed in the FDE programming for the stockers, like record 421 - MACH AILERON? EFFECT TABLE *1024 - which actually makes ailerons perform more realistically in speed dives and at level top speed (FYI this one is also a hex edit). Flap and dive brake lift and pitch are other areas the developers overlooked, but that's another thread. The third pic below shows the 421 record from a stock A6M5 Zero and the fourth shows that from an AVH 1% 1JA_A6M5a_52. Note the stocker's mach aileron effect is flat, meaning unused and therefore not a control factor. But the 1% A6M5 shows some interesting stuff. This record works in conjunction with the 518 record, but graphically displays in the opposite direction as the 518 record. As speed increases, the line goes UP rather than down to indicate a stiffening of the control factor, locking the ailerons at speed. The verical lines in the chart are speed increase intervals, but i'm not 100% certain of their numerical ranges. The good news for non-1% workbook tweakers is even though these records can only be hex edited to make parameter changes, you can easily copy/paste any individual record from a 1% air file into your stockers to alter performance, or just do a like-kind exchange of the entire 1% air file and the vitals from the 1% aircraft.cfg as replacements for any stocker - like-kind being the keyword here.

Food for thought, eh?

 

[Devildog73]

SO, IF I am understanding all of that, I could theoretically use a 1% Zero instead of the stock and have the right effects when a 1% Zero AI follows my Lightning into a 360MPH dive....??

 

[bearcat241]

Correctimoso...i strongly believe that the application of these exotic parameters in addon flight models explains why AI friend and foe perform so unexpectedly mediocre, as compared to stockers which don't use these parameters. Remember the ubiquitous high speed nose tuck experienced in almost every 1% fighter? The engine can't cheat to override the complicated stuff and just uses what you give it. I think some of us have complained in the past about AI performance in the 1% envelope. Its not a bad thing for the player's craft in terms of immersion and even fairness in online play, but the AI engine just doesn't get it. Its why you see persistent takeoff and landing crashes and flying into the ground in ground attack mode with 1% wingies.