PDA

View Full Version : 4.09 Flight Model Analysis



Daiichidoku
11-06-2010, 08:19 PM
interesting..... http://forums.ubi.com/images/smilies/winky.gif

http://www.flightsimtesting.com/

ElAurens
11-06-2010, 08:36 PM
The late P47D is more than a bit too slow.

Interesting, but not surprising.

AndyJWest
11-06-2010, 09:07 PM
The zINFOMOD mod estimates the performance of the aircraft based on the flight model parameters. These estimates are not guarantied to be 100% correct. An actual manual flight test should be preformed to confirm these estimated results.

Or to put it another way, these results don't actually tell you anything definite in absolute terms, though they may be useful for making general comparisons between aircraft modelled in IL-2 - as Oleg said years back about this, and about IL-2 Compare, which uses the a similar method to generate data.

This sort of 'evidence' has been used far too often to 'prove' that an aircraft underperforms (never overperforms though, I wonder why?), without actually doing the required real-(sim)world testing. I'm inclined to think that if people didn't keep quoting results derived this way as evidence, proper testing methods might have been developed by now.

ElAurens
11-06-2010, 10:20 PM
Tagert's testing methods have in the past indeed showed several aircraft that were overperforming.

This type of testing is very accurate in comparing aircraft in the sim to each other as it removes the human element, and all it's inherant flaws from the equation.

The results are repeatable, unlike human sim testing, which can have very different results for the same aircraft on multiple test "flights", as no one is a robot and able to fly a perfect test profile.

And these tests are indeed "flown" by the testing program, and not just numbers pulled from the FM data in the code. And they are done on a map that has correct standard day conditions, unike any stock map in the sim.

M_Gunz
11-07-2010, 02:05 AM
Originally posted by ElAurens:
Tagert's testing methods have in the past indeed showed several aircraft that were overperforming.

LOL! ROFL!

ytareh
11-07-2010, 02:14 AM
Wow this looks like Graph Heaven !But Im curious is it just the beloved IL2 Compare reformatted or is it new , providing alternative information?

JtD
11-07-2010, 02:44 AM
Mind you, the games "standard" map is Crimea, aircraft performance is supposed to reach real values on that map. Making flight tests on a modded map with other conditions makes it impossible to compare in game values to real life values.

So that:

The late P47D is more than a bit too slow.
is just as ridiculous as

And these tests are indeed "flown" by the testing program, and not just numbers pulled from the FM data in the code.
since they are very obviously not flown. It is, believe it or not, nothing more than il-2 compare data.

I also find it disturbing that someone finds it suitable to label himself NACA.

BaronUnderpants
11-07-2010, 06:52 AM
So real ac performance, flown by real pilots is compared to ac in a game flown by a robot (program) and then labeled as "uber" or "allied"? (couldnt help myselfe)

Sounds accurate enough. http://forums.ubi.com/images/smilies/shady.gif

Erkki_M
11-07-2010, 07:07 AM
Nothing personal, ElAurens, really, but:


Originally posted by ElAurens:
Tagert's testing methods..

Lol.

AndyJWest
11-07-2010, 07:28 AM
...these tests are indeed "flown" by the testing program, and not just numbers pulled from the FM data in the code.

Are they actually 'flown' though? This isn't what the reports seem to say:

SIMULATED WORLD DATA

All of the flight model parameters and raw performance data were obtained using the zINFOMOD. Note that for the most part the math of the 6DOF flight model does not change from one aircraft to another. The specific aircraft parameters used by the flight model math is what changes from one aircraft to another and makes one aircraft different from another.

To make an analogy the flight model along with the flight model parameters comprise the ‘black box’; the user’s inputs along with the planes current state in the 3D world are the ‘inputs’ to the ‘black box’; the performance values shown in the following graphs are the ‘outputs’ of the ‘black box’. The difference here being the user’s is replaced by zINFOMOD. The zINFOMOD internally puts the aircraft through a set of standard tests and outputs the raw performance results to several log files. The raw performance results and derived performance results are presented in the following graphs and tables.

The zINFOMOD mod estimates the performance of the aircraft based on the flight model parameters. These estimates are not guarantied to be 100% correct. An actual manual flight test should be preformed to confirm these estimated results. Also note that some of the min and max values listed in the table of flight model parameters may not match the min max values of the graph. This is in part due to the fact that some of these parameters can interact which in turn force the flight model maker to tweak a flight model parameter to get the desired ‘black box’ output.

Now, I've not used zINFOMOD, and don't know exactly how it works, but I've always understood that it extracts FM data, and uses this to estimate performance. This seems to be what the above suggests.

In any case, the choice when making tests is not necessarily just between zINFOMOD and a human pilot. There are ways to control the real-(sim)world aircraft using DeviceLink to eliminate the human factors, as was shown by the LesniHU application some time ago. Using this, or similar applications, should give consistent 'real' results, avoiding external 'black box' estimates.

ElAurens
11-07-2010, 07:30 AM
Not taken personally at all.

I doubt many of you have actually spoken to Tagert, like I have, he knows his stuff, wether you believe it or not.

The problem with real human "testing" in flight sims is that if you took 20 proficient sim "pilots" and had them test one aircraft, there would be 20 different sets of numbers that in all likelyhood would have enough variance to invalidate the whole proceedure.

Why? They are not all "flying" the same aircraft. They all have different sets of PC interface equipment, different input settings, etc... so in effect they are all flying a "different" aircraft.

This is unlike real world testing where you had 20 military pilots all test the same aircraft. Their numbers would be much much closer, and an average of them would yield an acceptable result.

There is simply too much transfferal of the real world to the sim in some of your thoughts.

These are just pixel representations of aircraft in a matematically generated "world".

Sometimes a lot of you guys forget this.

AndyJWest
11-07-2010, 10:37 AM
ElAurens, from looking at the timestamps, you may well not have seen my response to your earlier points (above). I'd be interested to see one, as this may help resolve the issue.

JtD
11-07-2010, 11:05 AM
Originally posted by ElAurens:
The problem with real human "testing" in flight sims is that if you took 20 proficient sim "pilots" and had them test one aircraft, there would be 20 different sets of numbers that in all likelyhood would have enough variance to invalidate the whole proceedure.

...

This is unlike real world testing where you had 20 military pilots all test the same aircraft. Their numbers would be much much closer, and an average of them would yield an acceptable result.

I disagree with that very much, as it is much easier to reproduce testing in game than it is in the real world, simply, because the testing conditions in game can be made the same, while they never are in real life.

I have double checked more than enough flight test results given for il-2 and can very easily get very good matches. For instance, in terms of speed, this will mean as good as 1 km/h, which is something you'll never get in real life without a **** load of corrections.

You will, of course, have to follow the same procedure.

ElAurens
11-07-2010, 01:23 PM
Gents, here is a comparison between manually obtained data and the zINFOMOD obtained data, as you will see they are virtually spot on with each other.




Manual P38H Rate of Climb
http://www.flightsimtesting.co...43/BF_ROC_VS_ALT.png

zINFOMOD P38H Rate of Climb
http://www.flightsimtesting.co...3/ZIM_ROC_VS_ALT.png

And I know for a fact that when the zINFOMOD data is questionable that Tagert will obtain data with Lesinu's program and manual tests for verification.

zINFOMOD's data is reliable and it saves vast amounts of time so that more aircraft can be tested in shorter time frames.

For those that build mod aircraft models the benefits should be obvious.

I know that Tagert's um, brash online personality rubs many the wrong way, but he is very sincere about his work to help improve this simulation.

JtD
11-07-2010, 02:04 PM
Originally posted by ElAurens:
Gents, here is a comparison between manually obtained data and the zINFOMOD obtained data, as you will see they are virtually spot on with each other.

Manual P38H Rate of Climb
http://www.flightsimtesting.co...43/BF_ROC_VS_ALT.png

zINFOMOD P38H Rate of Climb
http://www.flightsimtesting.co...3/ZIM_ROC_VS_ALT.png

Links don't work.


And I know for a fact that when the zINFOMOD data is questionable that Tagert will obtain data with Lesinu's program and manual tests for verification.

Where's the verification for the 25lb Spitfire?

M_Gunz
11-07-2010, 02:30 PM
Originally posted by AndyJWest:
In any case, the choice when making tests is not necessarily just between zINFOMOD and a human pilot. There are ways to control the real-(sim)world aircraft using DeviceLink to eliminate the human factors, as was shown by the LesniHU application some time ago. Using this, or similar applications, should give consistent 'real' results, avoiding external 'black box' estimates.

It may eliminate human factors but it makes its own. Best to call it consistent as it will do the same thing same way, best or not for the plane chosen and the plane chosen will fly however that shtick is done. You make your own black box with that, as the stick jerk transition in "Engineer: Grant Senn's" little "it's AP, it must be right" game. Sorry but some planes need more care taken in certain maneuvers than others and it slants whole "tests" to do otherwise.

steiner562
11-07-2010, 02:30 PM
Pretty cool info daii thks for posting.

M_Gunz
11-07-2010, 02:33 PM
Originally posted by ElAurens:
I doubt many of you have actually spoken to Tagert, like I have, he knows his stuff, wether you believe it or not.

Yeah. He knows more than YOU and can BS YOU all day.

In fact, what he knows isn't the problem, it's the BS he plays that's the problem and if he knows better then it's worse.

steiner562
11-07-2010, 02:35 PM
Originally posted by M_Gunz:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by ElAurens:
I doubt many of you have actually spoken to Tagert, like I have, he knows his stuff, wether you believe it or not.

Yeah. He knows more than YOU and can BS YOU all day.

In fact, what he knows isn't the problem, it's the BS he plays that's the problem and if he knows better then it's worse. </div></BLOCKQUOTE>


Alright mgunz enough, dont push your trollin too far.

Daiichidoku
11-07-2010, 02:50 PM
Originally posted by steiner562:
Pretty cool info daii thks for posting.

np, only took me a moment to do, the author did all the work of course; certainly is cool

"T-lite"; just the facts, ma'am, without the poor nancy, savvy? :P

ElAurens
11-07-2010, 06:01 PM
Originally posted by JtD:
Links don't work.


Sorry about the bad links. Go to the index page in Dai's first post and both the P38 H manual and zINFOMOD tests are there for your reading enjoyment.

And what does the 25 lb. Spit have to do with this conversation?

Why must you always make it a Red vs. Blue thing?

thefruitbat
11-07-2010, 06:15 PM
i've never understood the red/blue thing.

i expect thee to be a mass revival of this trend sometime about febuary, lol.

JtD
11-07-2010, 10:23 PM
Originally posted by ElAurens:

And what does the 25 lb. Spit have to do with this conversation?

Why must you always make it a Red vs. Blue thing?

You lost me here, why are YOU making it a Red vs. Blue thing?

I want to see the verification for the 25lb Spitfire, because if there is none you've either not been telling the truth or Mr.NACA doesn't really ask a lot of questions.

M_Gunz
11-08-2010, 02:55 AM
Just looked at the Spit 25 'data' and there is only zInfomod there. Why should Mr. NACA check? Mr. NACA doesn't double check his own work or he'd see that he's screwed up Altitude vs Mach by using IAS to determine Mach! Yeah, TAS increases with alt up to about 18,000 ft and Mach 1 decreases with alt but the highest Mach shows at sea level! I'm not making it up but check it before it gets fixed, LOL! The thing is that not even obvious errors were checked for, the method is "Mr. NACA Approved"(TM) whether it's any good or not.

For some odd reason I don't think I have to guess which planes are most overmodeled. It's the 109's and the ubering is the climb.

But it's not about Red vs Blue, oh no no no. It's about USA vs Germany and always has been.

ElAurens
11-08-2010, 07:33 PM
So, this all boils down to your hatred of one airframe in the game, and no doubt your stats on Warclouds eh? That about cover it?

You guys are losers, with a capital L.

Honestly I don't know why anyone bothers with either of you anymore.

I'd say "have a nice day", but it would be a lie.

BillSwagger
11-08-2010, 09:32 PM
I think what you have to realize is that Il2 is a video game. Its made to enjoy and marketed to an audience. Depictions of aircraft might not be historically accurate if it means making the game funner.
There are planes that over perform on both blue and red, as there are planes on both sides that also get the short end.
It would be great to see more balance in a game and not always have to use the same heavy hitters. It doesn't matter much to me, if a map is too lop sided then i just don't play. I either look for historical maps with more balance or find servers that offer the same planes on both sides. I think some servers recognize that and much of the time the maps will also invite a lot of players. Overall, though, there are less and less people playing and it shows.

Bill

JtD
11-08-2010, 10:40 PM
Originally posted by ElAurens:
So, this all boils down to your hatred of one airframe in the game, and no doubt your stats on Warclouds eh? That about cover it?

You guys are losers, with a capital L.

Honestly I don't know why anyone bothers with either of you anymore.

I'd say "have a nice day", but it would be a lie.

That's pathetic, really. Throwing insults to cover up lying.
And I know for a fact that when the zINFOMOD data is questionable that Tagert will obtain data with Lesinu's program and manual tests for verification. He doesn't do **** about questionable data.


Originally posted by M_Gunz:
Why should Mr. NACA check?
Because in game it does not behave like il-2 compare shows.

M_Gunz
11-08-2010, 11:47 PM
Originally posted by ElAurens:
So, this all boils down to your hatred of one airframe in the game, and no doubt your stats on Warclouds eh? That about cover it?

You guys are losers, with a capital L.

Honestly I don't know why anyone bothers with either of you anymore.

I'd say "have a nice day", but it would be a lie.

I see how you can be buddies with Tagert.

M_Gunz
11-09-2010, 12:11 AM
Originally posted by JtD:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by M_Gunz:
Why should Mr. NACA check?
Because in game it does not behave like il-2 compare shows. </div></BLOCKQUOTE>

Between some of the core algorithms and the actual running of the whole package there's a difference for sure. I think Oleg had a cryptic thing to say along those lines just a time or three. And very few people who can say what other parts make the difference, the code isn't exactly public domain and isn't all in the Java either.

triad773
11-09-2010, 12:53 AM
[[BIG] POPCORN] 15 pager sure

IBFTL

http://forums.ubi.com/images/smilies/10.gif

steiner562
11-09-2010, 12:54 PM
Originally posted by JtD:
Links don't work.



These ones should work.

P38H ROC per zINFOMOD
http://www.flightsimtesting.co...3/ZIM_ROC_VS_ALT.png (http://www.flightsimtesting.com/my/analysis/il2/zinfomod/4.09m/fm/HSFX_4.1/P-38H-05-LO_1943/ZIM_ROC_VS_ALT.png)


P38H ROC per manual test flight using LesniHU auto pilot
http://www.flightsimtesting.co...43/BF_ROC_VS_ALT.png (http://www.flightsimtesting.com/my/analysis/il2/manual/4.09b1m/roc/UI_1.2/P-38H-05-LO_1943/BF_ROC_VS_ALT.png)

JtD
11-09-2010, 01:09 PM
Yeah, thanks.

Found my way there already. http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif

Aviar
11-09-2010, 03:17 PM
Originally posted by M_Gunz:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by JtD:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by M_Gunz:
Why should Mr. NACA check?
Because in game it does not behave like il-2 compare shows. </div></BLOCKQUOTE>

Between some of the core algorithms and the actual running of the whole package there's a difference for sure. I think Oleg had a cryptic thing to say along those lines just a time or three. </div></BLOCKQUOTE>

I'm not taking any side here, but maybe you were talking about this Oleg quote:


"I would be glad if you all will not use Il-2 compare for the anounce of real sim performance for the planes.

Il2 compare isn't a source for comparison of real things that are in sim.
That reads data for tunings but in the sim this data correllate with hundreds more parameters and final result is other.

So the differences in speeds will be great sometime. Say on high altitude the speeds by Il-2 compare program and real speed in the sim could be different from 30-70 km/h comparing to the curver that draw the program Il-2 compare.

I asked Wastel to try himself in add-on beta to get the same speed and climb for P-51D as he posted. I'm
interested for the result from him .


I already posted many times that Il-2 compare is good just for the novices to take the first image of the plane that he likes to select. Taht all. And this program couldn't be recognised as serious sourcre or refernce of prcise tunings in a sim"


Oleg Maddox
1C:Maddox Games

Here is the link:

http://forums.ubi.com/messages...oxreadyroom&id=zucve (http://forums.ubi.com/messages/message_view-topic.asp?name=Olegmaddoxreadyroom&id=zucve)

**I see the link is now dead, so don't blame me, blame the forum.


Aviar

ElAurens
11-09-2010, 03:29 PM
Thanks Aviar, I've been searching for that post of Olegs for a while.

AndyJWest
11-09-2010, 04:10 PM
Thanks, Aviar. I'd say that unless the pseudo-NACA can provide evidence that his results were all checked against real-(sim)world figures, Oleg's comments are quite sufficient to disregard any detailed claims made using them.

That takes us back to the question of how to do real-(sim)world testing. Personally, I'd be inclined to say that as a minimum, any contested results need to be backed up with .ntrk files - that way we can at least analyse them using DeviceLink. Map, conditions difficulty settings etc all need to be given too for results to be meaningful. Unfortunately, this requires a great deal of work, but there aren't any shortcuts to doing it properly.

M_Gunz
11-09-2010, 05:22 PM
Pseudo NACA had a total love thing going with the difference between IL2Compare's 109K-4 Climb 'data' and a historic chart that Kurfurst produced. In a small area it didn't match by 30%, not that while climbing you would stay there long at all, but that's not the point when you're on an agenda and want to play Glenn Beck over it. I doubt that anything Oleg ever wrote would knock him off his obsession about 109 climbs. Of course if one question about the Spit 25 makes JtD and myself Losers then what does that and now now two complete _fundamentally flawed_ pseudo-productions complete with 'realistic' trimmings say about "Mr. NACA"?

AndyJWest
11-09-2010, 08:22 PM
Just out of curiosity, M_G, what is the particular issue with the Spit 25lbs? I know it has a reputation as an overperforming 'noob' plane, but what exactly is the argument about? If I can persuade my PC to run long enough without crashing http://forums.ubi.com/images/smilies/compsmash.gif, I might even be able to do some tests myself.

M_Gunz
11-09-2010, 09:07 PM
I had to ask JtD so I dunno why ask me but he says there is a difference between IL2Compare Spit 25 and the game Spit 25 enough to serve as a check against IL2Compare's accuracy.

I doubt that the PLANE was chosen by JtD for any other reason and certainly not what 'side' it's on. It's about checking your work instead of finding a way to push an agenda and dismissing objections on 'poor nancy' grounds.

JtD
11-09-2010, 10:30 PM
The Spit 25lb doesn't change supercharger as it should, so the performance on one altitude band around 3500m is much worse than il2 compare shows.

While the overall significance to the pilot is low, it is an easy way to seeing how well performance tests are done. It's clear that these ones aren't better than il2 compare. Same data, presented differently. Properly done, this would have been found out and at least been commented on.

BillSwagger
11-09-2010, 11:10 PM
What about the fact you can fly around at max boost for tens of minutes?
That surely has to cause a dis-appropriation between game and historical performance.

AndyJWest
11-10-2010, 07:36 AM
If I get the chance, I'll see what results I get using my AP - should be easy enough to test, though I'm not sure exactly what fuel load etc IL-2 compare figures are measured at. Presumably the Crimea map?

BillSwagger, the discussion isn't about Spit 25lbs in IL-2 vs actual historical performance, it's about Spit 25lbs in IL-2 vs IL-2 Compare (and presumably pseudoNACA's) figures.

JtD
11-10-2010, 07:47 AM
Mr.NACA apparently did his testing on a so call "FM test map". It has atmosphere properties different from Crimea, and will lead to lower speeds. But since he's doing all his testing on it and only wants to use it for in game relations, there's no problem with that.

Il2 compare is measured at whatever number is entered into the calculation. Usually this is the weight of the "standard" configuration with 100% fuel. Could be different with the hack Mr.NACA is using.

AndyJWest
11-10-2010, 08:14 AM
If it just is a question of at what altitude the supercharger changes stage, it should be easy enough to figure out where the problem is occurring - it should be possible to match speeds above and below the discrepancy to estimate the map 'altitude' error. There are figures somewhere on this forum (the 'ground effect' thread I think) that show what this should be.

I could in theory do trials using the test map, but running my AP along with mods is pushing my ancient PC (only 1 Gb RAM) beyond practical limits, and in any case I'd rather do testing under stock conditions.

DKoor
11-10-2010, 10:44 AM
Originally posted by AndyJWest:
Just out of curiosity, M_G, what is the particular issue with the Spit 25lbs? I know it has a reputation as an overperforming 'noob' plane, but what exactly is the argument about? If I can persuade my PC to run long enough without crashing http://forums.ubi.com/images/smilies/compsmash.gif, I might even be able to do some tests myself. Try to make the 25 dying from overheat. You will fail (on any alt) unless you really go intentionally for it.

I'm surprised you didn't know that after all of the talk about it here.

MiG-3AM also does not overheat (and few others) while running on deck at full power (you can run it outta gas full throttle), but no one cares about MiG apparently.
It was Spitfire that draws most attn (ok, ok, Warclouds planeset really).

Some rather odd explanation for it was the radiator (never fully closes or something like that)... but in reality I'd like to hear from experienced pilots whether they were or weren't EVER flown an WW2 airplane that can go like START (IGNITION) - FULL POWER+WEP UNTIL OUTTA GAS @ level - LAND.

Don't think so.

AndyJWest
11-10-2010, 11:04 AM
Overheating issues don't really show up in IL-2 Compare, which is another good reason to not take too much notice of it for absolute performance figures.

I'd agree that the Spit 25 is hard to kill by overheating, but does it lose power if run too hot for an extended period - I thought it did? I can probably test this easily enough.

I don't think there is any real question that IL-2 engine modelling is too simplistic, but we're largely stuck with that. Hopefully SoW:BoB will do this better.

BaronUnderpants
11-10-2010, 03:33 PM
I remember testing the Spit myselfe. Ran full out with wep, at the end the topspeed dropped due to overheat to something like 470 km/h untill the engine finally stopped running out of fuel.