Jump to content


Photo
- - - - -

Spad AI: calling out HT


  • Please log in to reply
59 replies to this topic

#41 gavagai

gavagai
  • Posts: 15542

Posted 17 October 2013 - 22:25

Again, by HT's argument the AI N28 should be turning as much as possible, and the AI Albatros D.II should be using energy tactics.
  • 0

#42 HotTom

HotTom
  • Posts: 8177

Posted 17 October 2013 - 23:04

Again, by HT's argument the AI N28 should be turning as much as possible, and the AI Albatros D.II should be using energy tactics.

There's the rub: Turning as much as possible. I'm not sure the N28 can be modded to turn without huge energy loss. Some coding changes are required. According to Harold Hartney, who liked the N28 better than the SPAD, the N28 should be as agile as a Camel but slightly (4 mph) faster.
  • 0

#43 Thaatu

Thaatu
  • Posts: 740

Posted 18 October 2013 - 18:12

I found something interesting.

Grattan, Robert F. 2009: Origins of Air War : The Development of Military Air Strategy in World War I, I.B. Tauris. London, GBR


I highly recommend this to everyone interested on the subject. Annoyingly it doesn't provide a conclusion for the argument, but instead reinforces both views depending on interpretation. Only a small part of it is about fighter vs. fighter tactics. Nevertheless it contains a 29 page British GHQ memorandum from February of 1918 titled "Fighting in the Air". Hopefully it's available somewhere else also.

On the other hand the memo says: "Fighting tactics may vary with the type of machines and with the powers and favourite methods of individual pilots." But then again it at least twice emphasizes that faster planes have the advantage of disengaging by climbing and thus controlling the battle. The main point of the memo is that the most effective tactic is an ambush. It advices formations to attack in pairs, so that when one pair disengages the second pair attacks, catching those who try to chase the first pair. It's pretty safe to assume we'll never get an AI which does that…

At some points the memo does seem to advocate what might be described as "BnZ", but it does so for all the single seater fighters. It doesn't however restrict attacking to be simply a single pass from above followed by a climb. I'm sure it was done, but the chosen style was dependent on both the type of machines and favourite methods of individual pilots. That an attack must be followed through is one of the basic commandments laid out, but this can be interpreted either way.

Actually, judging by the book and the memorandum, the best way to produce a realistic AI would be to combine all the three AIs we have. It may not be easy, or even possible. Vanilla AI dives and shoots best, while Gav's AI climbs best and Criquet's AI gains distance best. Still, I would not exclude maneuvering especially for targeting from a BnZ AI.



Btw, Gav, thinking beyond Spads, would it feasible to try the current Spad AI with fast climbers, such as Camel, N.11 and 17 or Dr.1?
  • 0

#44 gavagai

gavagai
  • Posts: 15542

Posted 19 October 2013 - 01:52

Hadn't thought of that. I will give it a try this weekend.
  • 0

#45 gavagai

gavagai
  • Posts: 15542

Posted 20 October 2013 - 13:13

Damn, the Spad AI will actually idle the engine when it should be running away or zooming at full throttle. No wonder it is so easy to catch.
  • 0

#46 gavagai

gavagai
  • Posts: 15542

Posted 20 October 2013 - 14:53

New Spad 13 available for testing in the OP. I'm hoping it's a good middle path. Please use the "ace" setting for tests.
  • 0

#47 HotTom

HotTom
  • Posts: 8177

Posted 20 October 2013 - 18:12

Fool me once, shame on you, fool me twice (you know the rest, sing along!). ;-)

It ain't broke, Gav. No need to fix it! :S!:
  • 0

#48 gavagai

gavagai
  • Posts: 15542

Posted 20 October 2013 - 18:29

:roll:

I have to disagree HT. I think the Criquet Spad 13 AI is one of the easiest to shoot down in the whole sim.
  • 0

#49 HotTom

HotTom
  • Posts: 8177

Posted 20 October 2013 - 19:16

Dunno. I'm usually flying the SPAD, not shooting it down. But I usually bring all my SPAD wingmen home at the end, so they must be doing something right.

I think we use a different yardstick, too. You're looking for kills, I'm looking to get my flight back alive (some have really good scores) after the engagement.

You just don't rack up big scores in BnZ planes. It takes patience to use them correctly. Wsa true historically, too. :mrgreen:

Trust me. If it were not right, I would be complaining about it. Reticence isn't one of my virtues. ;-)
  • 0

#50 gavagai

gavagai
  • Posts: 15542

Posted 20 October 2013 - 20:43

Fine, but there is always room for improvement. The Spad 13 is far superior to most of the German scouts in RoF, so it damn well should be scoring kills!
:x
Please, try this. I think this Spad 13 does both. It runs, it maintains a bit more cohesion, and it shoots stuff down.
  • 0

#51 HotTom

HotTom
  • Posts: 8177

Posted 20 October 2013 - 21:04

Welllllllllllllllll…okay…My Tigers are out (friggin Red Sox have a great bull pen; Tigers not so much), so what else do I have to do? ;-)


  • 0

#52 HotTom

HotTom
  • Posts: 8177

Posted 20 October 2013 - 22:12

Wow! Mikey likes it! You're getting there! From a "staying alive" standpoint, it's a big improvement!

Flew two missions: 3 ace SPADs vs 3 ace D.VIIs and then vs 3 ace D.VIIFs.

The first against D.VIIs was inconclusive but the SPADs extended very nicely. After about 20 minutes no one had shot down anyone.

I figured the D.VIIFs would overpower the SPADs but it wasn't true. Two Fokkers shot down one SPAD but then it was 3 Fokkers versus 2 SPADs and it went on forever. I was surprised the SPADs outran the D.VIIFs. And I was very impressed my AI driver ignored a close shot on two Fokkers chasing my wingman because there was a D.VIIF a hlaf click behind me. My guy just kept on running! Bravo! With your Mod, the SPAds stay pretty much co=alt throughout.

So, then I tried SPADs vs D.VIIFs with Criquet's Mods. It was quite a similar fight to using your Mod. In fact, I was pleasantly surprised when my driver got down to below 9,000 feet (the D,VIIs were at 12,000 and had altitude superiority throughout) and my guy then broke off, did a series of climbing turns and got back up to 12,000 feet. In the end, the SPADs lost. One SPAD and one D.VIIF collided. The other two D.VIIFs then shot down my wingman and next me (why my guy didn't try to dive away mystifies me). The Fokkers had regained their altitude superiority (which they didn't with your Mod) and that made the difference.

Nice Work. Neither is perfect but both are effective. I would use either one now!

:S!:

HT
  • 0

#53 gavagai

gavagai
  • Posts: 15542

Posted 20 October 2013 - 22:20

Ok, thank you for the feedback!
  • 0

#54 Thaatu

Thaatu
  • Posts: 740

Posted 25 October 2013 - 18:49

I've been too busy for RoF this week, but I did a few tests (5 Ace D.VII vs 5 Ace SPAD XIII). The SPADs flew nicely, managed to escape the enemy into an advantaged position, usually with a split-S followed by a climb. Their initial dives from above weren't as well executed as Vanilla AI's. Usually the SPADs started the fight by diving slightly above the enemy's level and making a barrel roll as they fired during the head on approach. Mostly the Fokkers won the matches, but predominantly not because of firing at SPADs that were maneuvering or escaping, but during head on approaches. The SPADs usually went down after long fuel leaks. I got the impression the SPADs constantly had the upper hand, but they just couldn't hit the Fokkers although they had lots of opportunities. Something is off with their shooting behavior and accuracy. A few times I witnessed a SPAD shooting at a target that was 1000 meters away, with a long and determined burst. Usually they began firing at a distance of 500 to 400 meters, which is too much even for Ace AI. Even when they fired snapshots from closer range, the Fokkers were visibly more accurate. Can accuracy be improved without affecting the flying parameters?

I'd say they fly really great now, but they need to control their fire more and preferably hit as much as other Aces. Excellent job Gav! :S!:
  • 0

#55 gavagai

gavagai
  • Posts: 15542

Posted 25 October 2013 - 19:10

Yeah thaatu, I was experimenting with PID controllers, and that might have reduced their shooting ability. For now I'm going back to the default PID controllers because my adjustments only seem to make things worse!
  • 0

#56 xjouve

xjouve
  • Posts: 70

Posted 18 December 2013 - 14:30

Interesting thread so far.

From my understanding, the tactics have almost certainly evolved as a consequence of aircraft evolution.

Quick comparison:

Although Guynemer has been intensively involved in the development of SPAD fighters, he learnt how to fight during a period, when lone wolf attacks and circling was a good option. He truely excelled in that particular method: (1) Udet reported after the war to have once fought against Guynemer who even with a SPAD proved to be an excellent dogfighter and better than he was (Udet was not yet the ace that he later became), (2) Risacher remembered this story: a Camel ace once visited the Storks and offered to have a mock dogfight with Guynemer and Fonck. Guynemer took him on head-on and outmaneuvered a Camel in his SPAD. The Camel ace said something like that he would have died 7 times in real life. :roll:

Fonck defied 3 of the british pilots, and he would attack them without being seen. That’s exactly what he did. Risacher added "Fonck was a very good pilot, of course, but he never made a dogfighting maneuver in the air - he always flew flat. Not to be seen by anybody - that was his style.” To add something interesting, according to Fonck himself (well, via Cne Fourcade who is the real writer of Fonck's book), he loved doing acrobatics, but alone, playing with the clouds; and he eventually used some during combats but in a very limited use.

Contrarily to Guynemer, Fonck learnt to fight with single seaters by mid 1917, as he joined the Storks on the 25th of April, when SPADs became the norm, especially in elite squadrons like the Storks. This speaks a lot. Guynemer finally died, most likely because he didn't adapt his style. According to Boyaux (another French ace) and according to several other sources (Mortane for example), Fonck has never been hit during his career as a single seater pilot.

Guynemer is the most well-known French example, but I can confidently add Dorme as well who, whitout being as a tenacious as Guynemer, was seen by his comrades as the best Stork and a top-class maneuver. To be taken literally or not, Guynemer once said that Dorme used to shot down 1 enemy per day and that his preferred game was to slide behind his 2-seater opponent to get him firing all his ammunition prior to shot him down (got this last point confirmed by reading letters sent by Dorme to family/friends).

Dorme, who was good enough not to be frequently hit, has finally been shot down by the end of May 1917, with his SPAD outnumbered (by members of Jasta 9 and 21), south-east of Reims. Like Guynemer, Dorme was able to efficiently use his SPAD for circling fights: he was reported by one of his victors (ace Henrich Kroll), to have given a fierce fight and when you realize that he and Deullin fought against 4 to 6 enemies, among whom 2 were aces (Kroll & von Schleich), and that as seen (but never confirmed) by Deullin, he managed to shot one German down in flames (possibly Ltn A. Paulus Jasta 9)… you could easily understand how good he was.

Dorme got 23 confirmed victories (seen by 3 independent witnesses, squadies excluded, as per Cdt Brocard's rule of confirmation), along with 40 probables behind enemy lines, and took part in 120 fights during his 623 hours spent in the air. He really was a great ace, perhaps one of, if not the most underrated, but he did the main part of his career flying Nieuports, and he hated SPADs (he has been one of the last Storks to change for a SPAD)… different planes, different techniques. :)

So, to conclude briefly, although it was possible to use a SPAD for doing acrobatics during a fight, even the best ones at this game finally died using this method only.
  • 0

#57 gavagai

gavagai
  • Posts: 15542

Posted 18 December 2013 - 18:10

xjouve, that was Cecil Lewis in an SE5a, not a Sopwith Camel.
  • 0

#58 xjouve

xjouve
  • Posts: 70

Posted 18 December 2013 - 19:10

xjouve, that was Cecil Lewis in an SE5a, not a Sopwith Camel.

Ok ;) I wasn't sure; I've checked a thread on theaerodrome.com. You might very well be right.
  • 0

#59 gavagai

gavagai
  • Posts: 15542

Posted 18 December 2013 - 19:12

Well, the French fighter that could supposedly out-maneuver a Camel was the Nieuport 28, not the Spad. :D
  • 0

#60 xjouve

xjouve
  • Posts: 70

Posted 19 December 2013 - 11:00

Supposedly is the word. :D

Anyway, the real question is to what extent pilot's skills could turn the SPAD, supposedly a poor dogfithing machine, into a respectable dogfighting machine?

Here is a quote from Lewis' Sagittarius Rising "In his hands the SPAD was a marvel of flexibility."
  • 0


0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users