[ GOTO THE MAIN EAWHS PAGE ]

CONTENTS

INTRO

DO AI PLANES CHEAT?
AI Use Simplified FM | More AI Planes | Bomber Gun Dispersion & AI study

DECISION CYCLES | UPDATE RATES | TWEAKING THE VALUES

TEST RESULTS
AI Improvement | Wingman Response | Players Plane

-+-

ENHANCING A.I. FURTHER
Rate Tweaks and v1.1 plus ECA | Rate Tweaks and v1.2a | Rate Tweaks and v1.1 over v1.2
Rate Tweaks and v1.1PC plus Realism Patch

It's been noticed over the years that many virtual pilots seem to consider themselves experts and yet they have reported that 'the A.I. does not black-out' and it performs outrages "UFO-like maneuvers"; With their UFO like abilities they can fly their aircraft to the edges of its performance envelope (more of a computer-like ability, than human-like).

One observation for an example is when a player pilot is trying to lead-turn the A.I. and the player blacks out but the A.I. pilot doesn't appear too. An other example is the player is trying to pull the same difficult maneuvers the A.I. does such as same speed and same sequence and you'll find out - you can't and have to break off.

Even more puzzling is that you may have noticed that AI planes seem to have much better acceleration than when you control and command them. So for an example when flying in formation with Hurricanes and you have given an order to the wingman to attack a target in front you, you'll witness him speed off to attack while your left behind, even when your at full throttle. With this observed, you would come to believe that they have more power, cooler engines and are generally faster overall.

So the question to all this superior behavior, is Why?

Maybe the AI's plane performance may appear to be 'faster' than some players planes possibly because they make changes slower and fly cleaner? or yet is it your flight control dead zones and sensitivity settings incorrectly optimized? or is it that you as the player are using the maximum difficulty settings, while AI uses an easy difficulty, with stalls & blackouts turned off? Keep reading to know the truth.

AI Use Simplified FM:
Many players believed that the AI uses a different flight model. Obviously, this would explain the huge advantage they seem to have. So years later it was discovered that the A.I. indeed use what is known as the "simplified" flight model.

One of the community members when designing their own custom FM's for EAW, had a talk with Tsuyoshi Kawahito, EAW's 'Game Designer' and 'Lead Programmer'. Mr. Kawahito stated the AI in EAW uses the "Easy" flight model, but can only fly their airplanes to 85% of their capability with this FM. The EAW designers could not have the AI employ the "Realistic" flight model and still create the kind of large size air battles EAW has without imposing a severe penalty on the user's computer system.

This is apparently a common method programmers had to take too, to reduce CPU workload, by simplifying FM algorithms for the AI, so that by and large, the overall outcome is similar (not the same, but close enough). Mr. Kawahito said that at the time work began on EAW, a 200Mhz PII was high-end! Therefore, one of the reasons EAW works so well on "old, slow" computers but can still manage 100 or more planes in the air at the same time is because it does not require making a huge number of extremely complex calculations ("Realistic" FM) for 100+ aircraft at a time. Instead, things are kept simpler to ease the burden on computer systems. It also helps with reducing the modem or network transmitted data traffic...

Some community members apparently hacked the eaw.exe (even source code changes in later years) and have forced player and AI to use the same realistic FM at all times, even during multiplaying. It was said that EAW was designed with the A.I. behavior (not the FM) but the aggressiveness (or lack of) hard-coded so changing it would be difficult. If its true that it can be changed, you'll end up needing fast connections to play properly. (The stock game with default settings run at 60 Frames per second, the AI apparently switch and use the 'complex' FM due to the fact the game is performing well).

AI planes (using simplified FMs) do not:
-stall (they may sometimes)
-spin
-redout/blackout (G effects)
-suffer from engine cuts in Hurricane and Spit1 in neg-G loops (fuel float)
-use cannons unless very close at the target (or at all ?)
-enemy seem to not go just after any target.. they come after you :)
-think creatively or plan their moves beforehand... they are actually quite predictable.

-bombs/rocket (external ordnance) weight and drag penalty is unknown if it effects the A.I.. It doesn't seem to matter in terms of flight modeling / performance if the A.I. wingmen aircraft still have their ordnance attached or not when engaging bandits. (However Bombs, Rockets and Extra Tanks do impose a performance penalty for the player's plane).

Never the less it is possible for human pilots to beat them, unless of course they outnumber you by more than 10:1. If the odds are overwhelming, try flying very close to the ground and keep turning. Eventually some AI planes crash on their own or you can once in a while get easily on their six (of course this doesn't work if your task is to intercept).

More AI Planes:
To get more AI planes in the air, use the ExtraSquads setting in the eaw.ini file. If you see friendly planes not in your squad flying in formation heading to some unknown destination, you can contact Ground Control with the Pilot radio to call them over for extra help. See my 'EAW.INI Help Document' for more information, under the section titled EAW, and topic ExtraSquads=

BOMBER GUN DISPERSION:

I posted this in response to a specific question on another thread, but it didn't get noticed. The original study was done more than a year ago (2000) and updated when ECA 1.2a came out. It may be of some use, although the gun settings I discuss are evidently of little use to anyone using one of the automated campaign managers. Anyway, see what you think.
******************************************


Background

EAW game players have expressed frustration with the perceived lack of realism related to bomber guns. The general consensus is that the guns are too accurate; inflicting a far higher number of casualties to attacking fighters than is known to be historically correct. After getting some input from Charles Gunst and others, I decided to investigate the dispersion settings for the bomber guns and try to conduct reasonably accurate tests of different setting.

General Information

Upon initial investigation, I found that the dispersion settings on various bombers didn’t make much sense. Some guns had dispersion settings of less than 1.0 and some were set at over 3.0, which is inconsistent with dispersion settings for fixed guns in other aircraft. In addition, the left waist and right waist gun setting for all bombers were wildly different; usually around 3.0 for one and .03 or so for the other. These guns should have identical settings.

After a review of all the EAW bombers with defensive armament I elected to begin with the HE-111 in a Battle of Britain (BOB) scenario. It should be noted that the Rate of Fire (ROF) for all flexible bomber guns is roughly one-half of the actual rate. This had been done earlier in an effort to minimize gun effectiveness by reducing the bullet throw weight in a burst. Some bombers, such as the B-17 were set up with guns of a smaller caliber than is historically correct. This was also done in an effort to reduce bomber gun effectiveness.

Methods and Settings

My general approach was to make changes, then run at least five identical single missions with the same mix of aircraft in each. I also made some initial assumptions concerning the relative effectiveness of bomber guns by position, based on my own reading on the subject, the observations recorded on the EAW forum, and on a survey of the stock or standard EAW settings. These are as follows, ranked from most accurate to least accurate, with suggested dispersion settings:

Tail gun - D = 2.5
Ball turret – D = 3.5
Top turret – D = 3.5
Ventral, rear – D = 4.0
Ventral, front - D = 4.5
Rear cabin – D = 5.0
Waist guns – D = 6.0

ROF adjustments have been made as mentioned above. There are some minor discrepancies in ECAO and ECA 3.1 stock settings. These can be found by simply browsing the gun data and referring to Charles Gunst’s excellent Notes. Generally, ROF settings are as follows:

German 7.9mm, ROF = 10 for flexible, 20 for fixed guns.
German 20mm, MGFF(M), ROF = 8.67 (all these are fixed mounts)
Note: There are variations in certain aircraft. Check the Notes.

Browning .30 caliber, ROF = 12 (all flexible in bombers)
Browning .50 caliber, ROF = 6 to 9 flexible (see below)

The .50 Browning is a problem. EAW uses a ROF of 12.5 for them in fixed installations. This assumes the existence of a special fifty caliber for aerial use. The standard M-2 Browning fires at a rate of 450 – 550 per minute, giving a ROF of about 9 for fixed guns and an EAW rate of 4.5 for flexible. All specs seem to be for the regular M-2. This is important for anyone who wants to modify their bomber gun settings because one of the first things one can do is reset the .30 guns to .50. Then the ROF, Muzzle Velocity, and dispersion settings for the .50 become critical.

.50 Browning Supplemental Information:

After this report was done, further information was obtained concerning special “Aviation” models of the Browning. Various models were supplied with shorter barrels, modified actions, and electrical feeds, ‘forced’ the guns to higher ROF numbers. The EAW 12.5 ROF can be considered reasonably accurate for these guns. One-half that rate would be 6.25. Other considerations concerning ROF are included at the end of this report.

Test Series

The HE-111 vs. Hurricane tests started out poorly because I forgot to run a test series with the stock settings. A total of twelve missions were flown, pitting 6 – 12 Hurricanes against 12-21 HE-111 bombers. The variation is because a few missions were flown in campaign mode. An average of 5 bombers were shot down against a loss of 1-1/3 fighters on each mission. Totals: 60 bombers and 16 fighter losses (3.75 to 1 ratio). This is an acceptable ratio and probably close to realistic, given a total combat time of about fifteen minutes each mission and no reinforcements for the fighters.

Five missions were flown with HE-111s pitted against Spitfire 1a fighters. The Spitfire is a bit more fragile than the Hurricane and the numbers bear that out. 22 bombers were shot down for the loss of 8 fighters, or 4.4 bombers vs. 1-2/3 fighters per mission (2.75 to 1 ratio).

Then I tested the FW-190 against the B-17F. Five missions were flown using standard settings. Seven bombers were shot down for a loss of nine fighters. Clearly, the Forts are harder to down than the HE-111s.

Five missions were flown in the FW-190 after changing the B-17F guns to .50 caliber and setting the dispersion as given above. Interestingly, I flew a couple of missions in the ME-110 against B-17Fs with stock dispersion settings and guns set to .50 caliber. The change to the heavier guns didn’t seem to have any effect on the AI fighter losses, but the difference for my own plane was obvious. I sustained fatal damage to my 110 after a single pass through the bomber formation on the first mission and on the second pass on the second mission.

But, back to the FW-190. Five missions resulted in 14 bomber kills against 6 fighter losses. In the last mission, I took no part in the action, but merely observed and directed my squadron. This mission resulted in 2 kills and 2 losses.

In order to determine the AI effectiveness on their own, I flew five missions using the FW190D against the B24D. Total bomber kills were 20 against 11 fighter losses. Average combat time was 20 minutes. Then I flew five missions pitting the FW190A against the B17F. 16 bombers were shot down and 13 fighters were lost. I observed the attacking AI in detail. More on that in the next section.

Artificial Intelligence?

By this time I had spent a good deal of time observing my squadron mates attacking bombers. I noticed one thing in particular. AI pilots are not very good. Period. No big surprise, except for the depth of their awkwardness. The intercepting fighter pilots tended to make their attacks in only three different ways: left quarter, right quarter, and from the rear, with variations in height. Some attacks were made from about the level of the bomber formation, but the majority seem to be made using a quartering attack from slightly above the main formation. The AI tend to shoot from long range, 400 – 500 meters, and close to within only about 300 meters before breaking off. The problem is that they tend to target aircraft in the forward center of the formation, which means they are often exposed to defensive fire from closer bombers. Upon receiving defensive fire, they break off.

Of particular consequence are the attacks from above the bomber formation. For some reason the AI end up too high for an attack, and break off without firing a shot. That’s bad enough, but they tend to break across the front of the bomber group, frequently in a gentle descending turn. AI get hit 3 –5 times in the course of an attack during which they fire at the bombers and often don’t get hit at all during non-firing passes. For some reason, when an attack ends with a vertical zoom away from the targets, they seem to get hit more often.

During my early tests I had been making high-speed slashing attacks. In the Hurricane this means 200+ mph and 320-350 kph in the Fw-190. However, I was getting hit too much when attacking the bunched formation. While I accounted for about 2/3 of the bombers, I also got knocked down on nearly every mission. Indeed, if I kept making passes, I either got damaged too much to continue or I ran out of ammo. Usually the former. In my normal BOB campaigns as a British pilot, I tended to hold off, attacking escorts only, until the bomber formation had been damaged and a bit scattered. This kept me out of the highest concentration of defensive fire. After the first five missions in the Focke Wulfe I started hanging back, shooting from long range, and trying to cripple bombers so they could be attacked in greater safety. Unfortunately, the FW is better at slashing attacks, being a poor gun platform for sniping.

My patience would often wear thin during these missions. I tried to keep up the attacks for about the same amount of time for each mission, in the interest of making apples to apples comparisons between missions. Often, though, the AI would simply not be able to do more than inflict minor damage to the formation after 15 -20 minutes of combat. So I’d go in and shoot up a few Forts. And get damaged. And crash land. It got very boring and predictable.

Further, AI are not aggressive. Unless ordered, they seem to concentrate on the main bomber formation instead of attacking cripples or even bomber elements which have strayed away from the main group (see Conclusions). This is especially true of Hurricane AI. ME-110 AI seem to be more aggressive for some reason. I saw many of them attacking crippled B-17s, although they still tended to make most attacks on the main formation.

Much direct (and boring) observation has convinced me that most bomber kills are made by us humans. I know that the AI can be relatively effective in combat with other fighters, so this is a bit odd. It appears that the AI programming is geared toward fighter vs. fighter combat. This would explain why they are less effective against bombers.

So, the AI are not particularly effective at killing bombers and they don’t expose themselves to defensive fire enough to suffer badly from it. In the course of doing these tests, I have varied all the settings for ROF, MV, and dispersion (and gun size for American bombers) and have noticed little difference in the way AI conduct attacks or in how close they will go before breaking off an attack. After completing most of my tests, I set the HE-111 guns to max dispersion, a setting of 15, and flew three complete missions using Hurricanes. The result? Nine bombers downed and six fighters lost. Not very dramatic. And the AI didn’t behave any differently. Then I set the MV for the defensive guns at 1000, or roughly half the real value. That resulted in fewer fighter losses due to the lower damage effects from hits, but there was no obvious difference in the way the AI conducted attacks.

During the FW190A/D testing, with AI only making attacks, I spent some time following individual AI around as they made attacks. This is what that looked like:

Electronic Pilot “A” – began by attacking from the rear and above the formation. As is usual for these attacks, he throttled back and sort of hung there, up on one side before breaking off. Did not fire. Followed with the same type attack, except he dived sooner, and actually fired his guns. Then another high, rear attack (HR), no firing (NF), followed by a high, quartering (HQ), NF. In the next series of attacks, he made an HR, NF attack, breaking very close across the formation, just above a cell of three bombers. This was followed by two high, crossing attacks, both NF, and both of which involved his driving through the bombers from one side to the other. After the second such attack, he wound up behind and below the formation and commenced a low, climbing attack from the rear, NF, made a climbing left turn, broke away from defensive fire, turned back into the bombers and was driven off by bomber fire twice. Then he made a sweeping turn to a medium altitude, quartering attack, broke off in a climb, rolled away to the left, right turn back to attack, finally FIRED, made a left break into the bomber formation, and collided with a B17. Had the nerve to survive and bail out!

Electronic Pilot “B” – This pilot made a total of seventeen beam, stern, and quartering attacks, actually using his guns on eight attacks. He converted many of the beam and quartering attacks into stern attacks and often broke right across the back of the formation, in the direction of the larger number of bombers and twice flew through the formation after an attack. After the eighth attack he began combining a break away with a hard turn back in to re-attack, varying that with a large circle back to re-attack. One his last attack, he fired on a cripple, behind and below the main formation, then went on through the formation, was hit, bailed out.

In general, I noticed that the AI pilots attacking a bomber formation often get in each other’s way. This is obvious at times and may be happening even more than I was able to observe.

Conclusions

Dispersion settings alone make little difference in the behavior and effectiveness of AI attackers. My American bombers are currently equipped with .50 caliber guns and that change by itself also did not have an observed effect on AI behavior or casualties. The change to .50 caliber guns did have an effect on my survivability. It made it very dangerous to approach such bombers closer than 300 meters, especially if my speed was under 250 - 300 kph. With the .50 rate of fire set at 6.0 and dispersion settings as given above, I was able to successfully complete about the same number of attacks as I had when the settings were for .30 guns. But, the secret to survivability against the Ami bombers is to make high speed, slashing attacks, or to break all attacks outside the 300 meter range and do a lot of long range shooting.

Once the bomber formation has broken up a bit, AI in all types of planes seem to become more aggressive. This is especially true when attacking bombers on the outbound leg, after they have struck the target. At some distance, perhaps 50 kilometers from the target, they seem to spend more time in pursuit of cripples. This is an unproven observation. It may be that they are going after the smokers more simply because there are more smokers. Nevertheless, a good many kills happen well out on the outbound leg. However, one must also consider that the AI have by then taken a good many hits. At least half my fighter losses in the AI-only test came late in the mission. Pilot “B”, in the above example, made seventeen passes before being shot down. AI have to be conservative in attacks just in order to last awhile during the game. If we ever succeed in making them more aggressive, they won’t live long.

A successful strategy against bombers involves the human pilot making cautious attacks aimed at disrupting the formation. Once bombers begin to leave the formation, there will usually be opportunities to make attacks against outlying groups and against cripples.

Since I conducted most of these tests in Single Mission mode, it is also likely that AI in Campaign mode will learn to be more effective with experience. They do seem to gain skills in other facets of the game.

Finally, I think that we sometimes fall into the trap of expecting too much from a single attack. Even make believe fighter pilots, such as we, have to balance aggressiveness with caution, and need to exercise good judgement and planning in carrying the fight to the enemy. The brass paid good money for that plane of yours. Use it wisely!

Additional Information and thoughts on V1.2a:
If AI are more aggressive and effective in V1.2a, then thought should be given to increasing the ROF for defensive guns back to the EAW normal levels. Generally, based on my reading of historical accounts, especially in books like “Luftwaffe Fighter Aces” by Mike Spick and “JG-26” by Donald Caldwell, a realistic loss ratio of B-17 and B-24 bombers to attacking fighters is probably about 3 or 4 to 1 for 1943, and probably closer to 1 to 1 in 1944. That includes losses to escorts. Naturally, this is a game and one can introduce variables suited to one’s style of play and personal inclinations in a manner not available to the actual combatants.

Jim Hume (OG) 5-2001

EMIL: The soon to be released Pacific Tide : Solomons War with the 1942 Guadalcanal and 1943 Solomons Campaign, the recent 1942 European Theater of Operations and the campaign I'm working on right now : the Battle of Brittain include updated armament. All the guns have correct caliber, rate of fire, muzzle velocity and range both for Axis or Allied defensive guns. The dispersion for all the defensive guns is set to 4.
However, I'll use your dispersion values in the BOB on the He-111, Do-17, Ju-87, Ju-88 and Bf-110.

A Solution to these problems:

In my own installs, I set the gun dispersions as set forth in the study, made all guns the correct caliber, ROF, and muzzle velocity. Emil's setting of all dispersions at 4.0 is a good compromise, but overlooks the differences in gun mounts (powered mount vs. flexible) and the inherent variation in dispersion due to physical position on the aircraft. Waist guns, being flexible mounts and on the side of the aircraft, should probably have dispersion settings even higher than the 6.0 I settled on.

Think of it. The waist gunners never have a zero deflection shot. Theoretically, the nose gun and tail gun positions should have the least dispersion and be most effective against attackers, having at least the possibility of zero deflection shooting. My reading has convinced me that the top turret was generally more effective than the nose gun, due probably to its being operated by a regular gunner/engineer.

In the Ploesti missions, a superb British gunner flew top turret on at least one strike. This guy was so good he would have the pilot trim the B-24 nose-down so he could engage flak towers along the route with his twin fifties. Knocked a couple out, too!

The issues of AI performance have to be addressed in other ways. Adjusting anything related to the defensive guns doesn't seem to have much effect on their skills.

Jim

Tannethal: Way back had done similar tests with clean V1.2 but increased the dispersion setting for the def-guns to around forty(40,0) this resulted in some nice midair collisions as the AI broke only at the very last moment and crashed into other planes of the formation frequently. This also resulted in a lot more cannon use on the German side since these are used by AI at relative close range only, as commonly known. The collision effect was somewhat reduced if the dispersion was set to around twenty.
In retrospective these values might have been to high since the bombers were dangerous only if you flew right beside them.

EAW uses a kind of a decision cycle that you can modify to affect the A.I.'s flight performance, behavior and command response by changing two or three settings in the eaw.ini file under [NETOPTION]. While changing these update rates to new values they can actually make the game easier in some ways, however in many different ways, others have found it will produce a fair fight. The real goal of these update rate tweaks is to at least prevent the enemy A.I. from having the edge all the time by eliminating their SUPER Enemy behavior and maneuvers. The rate changes can be done for offline and online play. (Note: Changing these rates in the eaw.ini on the Client side will not mess up anyone that goes online flying, it appears that these options are set by the Hosts ini and used for all players.)

Back in 2001 when this information was discovered, then it was used and tested, many players where a bit skeptical at first if it really effected the AI and you may as well be today, but the more you read on and test things out yourself the more you will come to see the differences. Because its just a concept and not about an absolute "fix", much more testing with EAW's default Flight Model (FM) and Difficulty Settings would be needed to find the complete and thorough answers. But because of the reasons mentioned here, this help document was created to try it out for yourself.

It would be a good idea to consider what Computer you are running EAW on and how many AI pilots are 'in the air' before making adjustments since EAW uses a decision/thinking cycle that is based on the speed of the computer; If the PCs CPU and Memory speed is not up to running all of the AI's at full quality speed (at least 15 Frames per Second or smooth 'Flight') you could have faulty performance in the game with the AI. Slow and choppy game performance will make the AI behavior poor and won't exactly reflect the numbers you put in. You may also find in a slow performing game that the player's (yours) plane controllability may become delayed or lagged.

The AI_PlaneUpdateRate seems to have a restriction, with the lowest possible value of 140. The other two settings, PlayerUpdateRate and ClockSyncRate can be set for much lower values.

• The lower the AI_PlaneUpdateRate the smarter the AI becomes but in turn it increases the load on the system times the number of AI planes in the scenario. (on a slow machine an AI value of 140 can make the A.I. a bit lazy as the computer can not meet the demand of the lower value). If the game slows down with really low numbers, use a higher A.I. value to keep the speed of the game up.
• When you increase the PlayerUpdateRate this makes "Flight Errors" more forgiving , it decreases the load on the system times one for the player as the game checks less often; At the suggested new value of 140, EAW gets your inputs at about seven times a second.
• The ClockSyncRate, was tested at higher values of around 200 or 210 (the game would Sync less often), but findings conclude it seems best to stay with the default 140. Although a lower value would Sync more often, in turn increasing overhead.

For online use, One Virtual Squad found the values 140, 16, and 100 respectively where rather enjoyable. Again depending on your computers speed and connection quality, You may need to tweak them out a bit.

In the eaw.ini file go down to where it reads "NETOPTION" And the Last three lines there read:

AI_PlaneUpdateRate=350......Change to 140 (alternative 140)
PlayerUpdateRate=35.........Change to 140 (alternative 16)
ClockSyncRate=140...........Stays the same. (alternative 100)

A.I. Improvement:
When the values are properly tweaked they invoke the AI to use different flying abilities while still putting up a good fight. The enemy AI fighters will now pull fewer "impossible" maneuvers (UFO's) like those impossible outside loops and vertical 'zoom' climbs as they tend to do in the default settings, thus bringing their skill level closer to the Players, even when the hardest difficulty is set in game. When your on their six, their defensive maneuvers include doing more barrel rolls, turns, climbs, dives and loops rather than the standard mad dash to the deck where they roll inverted and zoom out away from you, however they still split S when they have to. They sway back and forth and try to get you to overshoot by using scissors. Overall the dog fighting just seems to have a better feel as all AI fly more smoothly.

Wingman Response:
A.I. Rate Tweaks effect the performance of your Wingman as well. When the tweaks are done to improve AI, your Wingman's aggression level seems better, they will take initiative and attack on their own accord. They become so intent they can nearly shoot you down if their after the same enemy plane as your chasing. So at times, you may need to disperse the formation a bit and/or tell them to just cover you to avoid friendly fire. In careers your wingman stay in formation with you rather than taking off ahead of you. Unfortunately with the aggression level increased, this will make your Wingman more un responsive to some orders. For an example, when Wingman are told to disengage, they may NOT comply until their current target is destroyed.

Players Plane:
When increasing the value from the default, A few things you'll notice is that your plane runs well at 3/4th throttle setting; You don't stall as fast or burn out your engine as quick! For an example, the Me-110 doesn't stall every 8 seconds like it usually does; You can actually make a turn now without spinning. A Hurricane has the capacity to keep up as well.

To further enhance the the A.I. read the following section combined with various Help Document for more information to learn other ways than adjusting the delays.

Rate Tweaks and v1.1 plus ECA:
Once you have your EAW game patched upto v1.1 (which has very aggressive AI); Using Charles Gunst's ECA (Enemy Coast Ahead) with tougher Armor and improved flight models, makes for some very interesting fighting. Shooting down an enemy plane is very difficult, but doable. Adding the use of the A.I. Rate change tweaks on top of those two options will even increase the difficulty and deadliness even further.

Rate Tweaks and v1.2a:
You can take the custom made v1.2a where the A.I. was improved (over 1.2) and further enhance it with the A.I. Rate tweaks. This will make the skill of the A.I. comparable to the old v1.1 aggressive A.I., (this is in comparison with the relaxed A.I. found in the stock v1.2). Overall AI behavior on both sides will appear less erratic, more aggressive, realistic and intelligent.

Rate Tweaks and v1.1 over v1.2:
Once you have your EAW game patched upto v1.2, you can now swap out the eaw.exe for a v1.1 exe while using the newer flight and damage models of the v1.2 patch. This will provide you with more aggressive and better aiming A.I. than you would normally find in the v1.2 exe. With the Rate change tweaks this will make the A.I. even more deadly.
However, you will loose the ability to use the ExtraSquads eaw.ini setting with v1.1 since it wasn't made part of the game until v1.2 came out. Also the bullet drop is greater in v1.1 than it is in v1.2 where bullet velocity was corrected.

Rate Tweaks and v1.1PC plus Realism Patch:
Once you have your EAW game patched upto v1.2; Using Paulo and Andy's v1.1pc hex edited eaw.exe and Knegel's Realism Patch v3.0(with tweaked FM's) will give you an extremely dangerous realistic game. This will give you the best combat challenge against as realistic AIs as possible. This arrangement will keep you on your toes. Make a miscalculation (ie: too low, too slow) and the AIs will be on you like flies on sugar. You will also notice that AI leaders and wingmen will also work together against you. So it's best to look for an isolated pair and go for the wingman first but don't take too long or the leader will be on your six really fast for a kill. Adding the use of the A.I. Rate change tweaks on top of those two options will even increase the difficulty and deadliness even more.

[END]