PDA

View Full Version : Collision Bubbles



Ivan
June 28th, 2009, 12:00
Hello All,

My original message was different, but the issue has been resolved.

Just out of curiosity, How close do folks here believe that aircraft should be before there is a collision?

- Ivan.

hubbabubba
June 28th, 2009, 23:37
Hi Ivan:wavey:

For me, collision should occur when your virtual nose is in his virtual tail. No two a/c models are the same but, in general, wingspan is more-or-less equal to fuselage length, so wingtip-to-wingtip flying should be possible.

Unfortunately, it also means that vertical separation will always be greater than needed. This is a compromise that has to be done. Otherwise, we would have proper vertical separation, but we would sit on the other pilot's laps before going :pop4:...

Ivan
June 29th, 2009, 06:48
Hi Hubbabubba,

What you are describing works pretty well if altitude is the same. If there is any slight difference in altitude, the radius is too high. The reason for revisiting this discussion is that my son Michael was trying to do strafing runs against my plane which was landed at an air field. He really pulled out in time, but the game didn't think so (vertical separation as you described) and we both exploded. He recognised that something strange had happened: that he had plenty of room to pass overhead.
I told him the the game was set wrong and that I would try to fix it.

BTW, he also wants me paint up a FW 190A for him! Good choice in aircraft for a 7 year old!

05 is the stock setting from AF99.
04 is 10 meters
03 is 7 meters
02 is 5 meters and is what I am using for experimentation.

I believe 02 or 03 is best, but wanted to see what the rest of you think.
I prefer 02.
- Ivan.

minuteman10
June 29th, 2009, 08:20
~S~ Ivan...Our CFG at the AAC is set to -2

COLLISION_SECS=-2

We can get some pretty tight passes....

Ivan
June 29th, 2009, 19:22
Hi Minuteman,

What does this mean as far as minimum separation goes?

Have you tried seeing how close you can get before exploding? I did this on a couple machines in my house a couple years ago by landing two planes on the same runway and inching (metering?!?) them closer until they collided.

My opinion thus far is that a 2 (5 meter separation) is a good value for a medium / small fighter such as a Zero or a Messerschmitt 109 or perhaps even a Spitfire. A value of 3 perhaps should be used for a big ole plane such as a Hellcat, Typhoon, Tempest, or Thunderbolt.

This is pretty much in line with Hubbabubba's idea as well. A 5 meter separation implies that the aircraft is around 33 feet long or has a 33 foot wingspan. The bigger planes would be around 45 feet long.... I'm still not decided yet.

- Ivan.

hubbabubba
June 30th, 2009, 06:47
The modification of COMBATFS.CFG "COLLISION_SECS=-2" is not clear. It appears to affect collision distance, but I'm not entirely convinced of that. It may just modify the frequency at which the game verify the possibility of a collision between aircraft, hence giving the impression that two a/c passing-by each other at high speed are less likely to go boom!

AAC uses the "Head to Head" (H2H) "fix". The readme file says;


Below are the sizes of the stock collision bubble and the size that they will be changed to by this file:
Stock AC Stock Size H2H Size
B-17g 15 (0Fh) 06 (06h)
B-24j 16 (10h) 06 (06h)
B-25j 10 (0Ah) 05 (05h)
bf109e 06 (06h) 04 (04h)
bf109g 06 (06h) 04 (04h)
bf110 09 (09h) 05 (05h)
C-47 14 (0Eh) 06 (06h)
Do17z2 10 (0Ah) 06 (06h)
FW190a 06 (06h) 04 (04h)
He111h 11 (0Bh) 06 (06h)
Hurricane MkI 06 (06h) 04 (04h)
Ju87a 06 (06h) 04 (04h)
Ju88a 06 (06h) 04 (04h)
Mosquito 08 (08h) 05 (05h)
P-47D 07 (07h) 04 (04h)
P-51D 06 (06h) 04 (04h)
Sopwith Camel 05 (06h) 03 (04h)
Spitfire MkI 06 (06h) 04 (04h)
Spitfire MkIX 06 (06h) 04 (04h)
Tempest 06 (06h) 04 (04h)
V1 05 (06h)??? 03 (04h)

As you can see Ivan, these numbers are more conservatives than yours.

I know that I have used "1" with my jeep project with very good results; smilo and I were making "near head to head" passes during tests.

The nature of the collision "bubble" is also debatable; is it really a "bubble"? Could it be a "box"? The fact that only one number is involved would suggest a radius measurement. But we may be wrong.

I also wonder if distances are the same on the ground and in flight. I don't know if its only me, but it looks like they're is a change, especially in MP games, between a/c seen rolling or flying from the outside view.

AF99 always sets the model at "5". Hex-editing is therefore a "must" for any serious modeler. That said, my Taifun is actually still at "5"...:kilroy:

The next a/c I release will not! But I'm still far away from any a/c release.

minuteman10
June 30th, 2009, 07:55
...during our next gathering, we'll get a measurement for you on ground and in air....

Ivan
June 30th, 2009, 10:17
Minuteman,

Thanks. Let me know what happens.


Hubbabubba,

I have been working on a A6M2 Model 21b Zero. I just finished a A6M3 Model 22a Zero. Both are based on the A6M5 Model 52a that I did many years ago. What I am finding is that I goofed up quite a few things on the original. Just about everywhere I look I find yet another error. Don't worry so much about the Taifun..... Though I DID notice that AB Squad now has a pretty cool display of the Taifun that looks a lot like yours.

- Ivan.

minuteman10
July 5th, 2009, 11:49
~S~ Ivan...measurements today..got to within 8 meters on the ground before collision...airborne...call it 25-30 meters nose to tail...closure made it a difficult measurement!!! Speed does seem to affect the range....

Ivan
July 5th, 2009, 18:53
Thanks Minuteman!

I'm not quite sure what to make of the numbers though. 25-30 meters seems like it's WAY too far. 8 meters seems quite reasonable.

I guess I need to try out these experiments also to see if I get different numbers in the air than on the ground.

- Ivan.

minuteman10
July 6th, 2009, 06:30
Agreed...I 'm almost positive that we could get tighter wingtip to wingtip....Could the collision bubble be an ellipse? Lagtime may also be a factor as the final distance hadn't updated at the moment of collision?

Ivan
July 6th, 2009, 18:33
I am guessing there is lag time involved here. Even with just two machines on a 56k baud serial connection, I get some serious lag. The propeller animation is the most obvious issue.

- Ivan.

hubbabubba
July 6th, 2009, 22:51
The only sure way would be to have a "target" a/c fly under auto-pilot while approaching verrrry slowly from behind.

Tedious, but necessary.

We only did two attempts during a 30 minutes period, and only the second attempt gave us a reading.

This is a "full-day" (if not weeks) project.:isadizzy:

minuteman10
July 7th, 2009, 06:48
Hubba and I were attempting measurements...as he called the closing distance, I noted that he was consistently +25 meters vs. my screen at about 1-1.5 second update delay....

Ivan
July 7th, 2009, 20:41
This is one of those deals where the "Autofollow Cheat" would be useful. I don't know how to implement, but if you have a slower aircraft set to autofollow a faster one and then just throttle down the faster aircraft.....

I believe I will try the autopilot idea again. The problem is being able to have such a slow closure rate to be able to see the distances ticking down. Last time I tried this, I could not get consistent readings.

Yes, any of these tests take quite a lot of time to set up. It took about 10 minutes or so for each test of taxiing planes together on the ground.

- Ivan.

hubbabubba
July 8th, 2009, 03:19
The advantage with ground measurements is that you don't have to bother about maintaining altitude (duh!:redface:).

Keeping constant slow closing speed in flight is an art!:engel016:

The auto-follow cheat would shorten the approach phase and speed-up the taking of readings. But to use it, we would have to change the COMBATFS.cfg "COLLISION_SECS=XXX" entry. One of the advantages of having it tweaked at "-2" was to make impossible that cheat. If you try it during an AAC game, you will go :pop4:.

Send me a note and I will show you how to use it. A couple of years ago, I would have received dead threats just mentioning it...:kilroy:

smilo
July 8th, 2009, 13:41
being the master of the smilo bomb,
I have noticed that if you look real close on impact,
the label showing the final distance will hold for a fraction of a second.
don't blink or you will miss it.

I have no data to prove it, but I would bet that the "collision bubble" is more of a collision cube.

one of these days, when I can find the time, I will give the auto-pilot test a go with my multi player network.
I was doing much the same thing when I was attempting a five Lancaster formation to Berlin several weeks ago.
'twas fun, but hectic.