Moa

A poly can have three or more sides. I always triangulate the model (turn polys with more than three sides to multiple polys of three sides). This can increase the poly count of a model, even though the number of vertices stays constant. I prefer to create triangles as it saves the video card from doing it at runtime, and ensures that all polys are planar (a requirement for some rendering calculations). I wouldn't be too worried, "Premature optimisation is the root of all evil" as the great computer scientist Don Knuth (jokingly) says. Try the model out with all the bells and whistles showing. If you then find performance is unacceptable you can reduce detail or group parts together. What I was cautioning against was not so much the polycount but the number of parts. There is an overhead for posing (rotating, translating) each part as well as setting up the Vertex Buffer Object "VBO" (assuming the ED rendering engine has a VBO per-part). When the number of parts becomes large it will drop the frame rate because the card is busy with the overhead of retrieving and posing the VBO rather than rendering. That's what I was trying to caution about - try to have fewer bigger parts rather than very many small parts. That can mean aggregating the polygons from some parts together (especially if the parts don't move relative to the body axes of the vessel). Cheers.

Lovely! You can get away with a reasonably high polycount - provided that the number of groups of objects is not too high. For example, I licensed this model of CVN-76 from Turbosquid for a project I'm working on: http://www.turbosquid.com/FullPreview/Index.cfm/ID/491281 The polycount is maybe double what the Hornets I have are, but the part count for the carrier is around 2000 parts (lots of little lifeboats and fire hoses). Having a high part count is a bit problematic - takes my framerate from over 700 fps with just a SuperHornet and sky down to around 50 once the carrier is in. So, try to design with polys below 300k and around 200 parts, if you can help it. The 2000 parts on that carrier is getting too much. I hope that helps with your design. Great work man - she's looking bee-yooo-ti-ful.

Thanks for the data point CptSmiley. I just knew that the military-grade Australian F/A-18 and the large number of US F-16 sims were Linux based - I'm sure there are now many alternatives too. I'm curious as to the aircraft that are simulated in your establishment (not problem if you are unable to say, just ignore this question if this is the case). While I'm at it, many thanks for releasing the source code of your F-16 implementation - it is nicely written and well documented source and interesting to compare with my own in-progress [non-DCS] flight model.

The *serious* sims are running on Linux/Unix (see my reference). This is for numerous reasons (eg. Linux is faster, see my reference from Valve below; and Linux is more flexible etc). Anything else is a game. So no point getting pissy about your favourite *game* vs anyone else's favourite *game* - especially when you are being inappropriately and consistently off-topic in the wrong thread. You may not care about multiplatform products but the world is starting to disagree. It is well known that Mac, especially the MacBooks, are excellent and even give a better Windows experience better than PCs do (eg. when using Bootcamp). The number of operating systems in use is increasing not decreasing (the Windows 'monoculture' of the previous decade is over, get used to it). In fact, given that Android (which is the marketing name of Google's customized Java running on Linux) has around two million new activations a day it turns out that Linux is now the fastest growing consumer operating system (although the users don't know it), and very soon will be the most widely used. In fact, here is a recent discussion: http://articles.businessinsider.com/2012-04-27/tech/31419233_1_market-share-android-apps Now, the point is that new software is not for Windows, or iOS, or Android, or any one thing exclusively. It must be developed cross-platform for whatever devices a customer has. This is exactly what the X-Plane developer aimed for, and why he was able to personally make $3.5 US million in a few months because he could rapidly move to iOS (while still supporting his Mac and Windows and Linux users; collecting extra revenue while he maintained a single codebase): http://techhaze.com/2010/03/interview-with-x-plane-creator-austin-meyer/ That is a huge advantage for X-Plane over FSX/Prepar3d. X-Plane will always be available no matter what device you have (and I have Linux boxes, and a Mac laptop, and an Apple smartphone, an Android tablet and two Windows boxes - and having all these devices is becoming common, and you game with whatever device you have on hand at the time). Ok, I'm now off-topic. But please note that professional simulation is not usually on Windows, and the majority of gaming is not on Windows (it is on phones and console - I'm not interested these either, but I do accept that bulk of gaming worldwide is now on these), and even high end games are available on other platforms than Windows (eg. many games have first rate ports on the Mac; Valve is migrating all its games to Linux for its 'Steam Box' project - and finds that Linux is noticeably faster for gaming, see: http://blogs.valvesoftware.com/linux/faster-zombies/). The trend is away from Windows monoculture, even for gaming. The next decade in gaming is not going to be like the last and it is a mistake to think so. The only direction for FSX usage is down. The only people buying Prepar3d are ex-FSX people, since it is not being actively retail marketed to new consumers (as far as I can see). It is good that you are happy with FSX but stop trying to push it as a good choice for new people when it is now a legacy product that has no support from its developer (and the vendor's intended successor was a dismal failure). Prepar3d is supported, but the vendor does not really care about consumers (look at its price points and lack of retail marketing) like you or I. It seems you are 'mistaking the heat from the embers of a dying fire' for an active product. Even worse, you choose to push this view in a forum for a different sim. So please desist. The X-Plane users are not interested in personal opinions about FSX in this thread. We get the fact that you think FSX is better in some way. Fine. You've said your piece. Now please stop trolling this thread, thank you.

With regard to the number of computers required to drive it. I read somewhere that the Australian DSTO developed a full-fidelity F/A-18 simulator and 33 Linux computers/cores were required to run it. I guess plenty of horsepower goes to radars and signals and stuff. Reference: http://www.itnews.com.au/News/154262,defence-spends-17m-on-ultimate-linux-flight-simulator.aspx Edit: You can't fault the Aussies for their sense of humor - found while searching for the reference:

Enough with the FSX, please! Use another thread for that. I know you just want to be informative but please post that to an FSX thread. I'm sure all of the X-Plane users in this thread are well aware that the *discontinued* FSX exists and most are aware that Prepar3d has taken the mantle. So no need to keep hammering here, k? Speaking of "platforms". X-Plane runs on my Mac, runs on Windows and runs on Linux too. Even some of the bundled aircraft in X-Plane is great (eg. the AV-8B is great fun, especially with the excellent X-Plane flight model supporting proper thrust vectoring). There are heaps of first rate aircraft for X-Plane (and X-Plane is exceedingly moddable). If you don't know of them then that's ok (Google is your friend, eh?) but please don't mistake that for them not existing. Many of the sim teams that do great FSX work are now doing great X-Plane work, eg Carendo, REX etc While I'm at it here is some *free* high-quality New Zealand scenery for X-Plane 10 (all credit to Andras Fabian for his hard work and generosity at giving it away): http://www.alpilotx.net/2012/10/26/x-plane-10-new-zealand-pro-preview/ Gallery (comparing default X-Plane 10 scenery and X-Plane 10 New Zealand Pro scenery): https://picasaweb.google.com/101666907909842492197/XPlane10NewZealandProRELEASE

Well, if you look at the licensing terms of the software, anything created with ED's tools belongs to ED, not the creator (unless the license has been changed in the last few months - I don't have a copy onhand to re-check). They would have to make an exception for models created in EDM, but without that exception they are likely to belong to ED. Probably if you enter a license agreement with ED you may get to keep copyright over the things you create. I'd suggest contacting them directly (JimMacK or Wags) to ask, rather than have us speculate.

Usually to maximize profit you charge more than the average person will pay. This nets the 'early adopters'. As time goes on you drop the price. This gets those more and more reluctant to pay, but they see that they are getting '$150 value for $20'. Apple does this (its enthusiasts and early adopters are screwed hard), Microsoft does this, ED does this (price drops to push sales come eventually). I can't see why it would make sense for modders to start the bidding at a low price. I've done a few freeware mods myself. It's nice to give to the community but then some people within it feel you should be grateful that they are using your mod. You are also expected to provide all sorts of support, even for non-mod related things, for free. This sux and is more work than fun. What seems to be the theme is that people don't want to pay more than $20 (or thereabouts) for a 3D model. Fortunately modders are not only motivated by money, as for that amount very few mods would ever get made. Paying money forces those that start a project to finish it (if they've taken pre-orders). Look at how many great projects have screenshots but never come to fruition. The money would help get some of them finished and we've have a lot more high-quality models than we do at present. ED are working on improving the models to match the capabilities of modern GPU (that can handle a lot of polys and all sorts of shader effects), but there are plenty of aircraft they're unlikely to get around to (eg. anything by Saab, Aermacci, the Yugoslav aircraft, light trainers etc). The community is the likely source of these, but even then it won't get done without a better carrot than a few hundred sales at $20. You only need one fifth the sales at $100 to make the same money, and then you progressively drop the price to get more sales. *That* is basic good business :) Edit: The MiG-21 module for $20 is fantastic value. That amount is a token of the effort and would never sustain a full-time commercial developer. The reason it is so cheap is that beczl would have done the mod anyway since he clearly has a passion for the aircraft. He's just making some 'pocket-money' on the side, and very good luck to him - he deserves it (in the past he has made many great models and given them to the community for free). But please don't confuse the $20 as being a realistic amount for companies that want to produce things commercially. Even though these hypothetical companies have smaller scope than ED, what they do often requires more work, since they have to discover and work around lots of things that ED can trivially change (since ED have the source code and the capability to make arbitrary changes as they see fit). Be very glad that you do great things for free and for small amounts like $20.

john_X's price of $100 euros is staggeringly *cheap* for the amount of work that goes in to accurate models. Please go and have a look at Turbosquid (http://www.turbosquid.com/index.cfm) and then come back and we can discuss further. The high-quality aircraft models there run between $US 300 and $US 1399. You can get some bargins for less, but the good stuff starts at the amount john_x quotes. I've been licensing models from Turbosquid and the good models are much more expensive than perhaps you imagine - although still cheaper than it probably costs ED to develop exclusive high-quality models from scratch (some of the cost of which they might recover by licensing to others, which is some of the point of this thread, no?). The economics of Turbosquid are this. It takes a long time (and time is money) to develop a model. However, if many users buy the model then the cost of development + profit (eg. required to make the next model), is amortized over several (hopefully many) paying customers. @john_X: have you ever considered licensing your models though Turbosquid ? (if you already do, then I didn't notice, sorry). If you don't then perhaps you could consider it - hopefully it would get you more sales/money for your hard work. Some aircraft are cheap from there. I got the following F-16C Block 52 model for $60 (which I then had to hack up and animate in the way I wanted; then run through OpenGL shaders I'm developing - this is rendered by me through my own code, not a 3D package): So I hope you now understand that the 100-150 Euro price for a model is very normal for a non-exclusive license, in fact, is is great value. Edit: I should mention that the Turbosquid license permits re-distribution of the model in a product (provided the format used is proprietary; presumably so further sales of the original author are not affected). You are not paying for a model license just for your own use, hence the cost is higher than a game module where you are only a user. I'd happily pay $US 400 for a non-exclusive license for new cockpits (meaning you can sell them to everyone else too). I'm looking for MiG-29C and K; F-16C and F/A-18E&F (would like to buy otherwise I have to do them myself, and I prefer to write code). They're only interesting to me if I also got a license to re-distribute in my OpenGL code (as Turbosquid does).

screenshot or it didn't happen!

For those that are interested in X-Plane, a MiG-29 has just been released (with full 3D cockpit): http://store01.prostores.com/servlet/x-planestore/Detail?no=415 Video: http://www.youtube.com/watch?v=ujCldqdTuOQ Lots of screenshots here: https://plus.google.com/photos/118179421914452480091/albums/5818891885822014449

Awesome work Tyger, Mustang and the crew of the 74th. Sharing this information is much appreciated. Thanks for your dedication and perseverance to isolate this issue. S!

While modern aircraft still carry flares the more advanced systems carry sensors and lasers to automatically detect IR missile launches and blind the sensitive seekers (the pilot doesn't have to react). This is equivalent to having infinite continuous flares in this game. MANPAD IR missiles are not the threat they once were for aircraft suitably equipped (modern combat aircraft and even some airliners).

I think it depends. In the Fulda Gap scenario there is no way you can kill all defences. If you go high you will die for sure, so the only way to survive is to be low. In the Gulf War most of the larger defences were killed by Wild Weasels and Stealth Fighters. Going in low didn't make sense but it was still attempted. Same thing in Libya. Defences were limited enough they could be destroyed and then aircraft could operate high. In Afghanistan and Iraq there were no defences left so aircraft could operate high. Whether aircraft must operate high or low in the next conflict they are used depends on the density of defences. In most cases they can be used high. Against an opponent with so many systems they cannot be wiped out in reasonable time (eg. attack the US, China or Russia) then aircraft will be operating low. It is very dangerous low (below 2000 feet) in this situation, but that better than suicide at higher altitudes. This is why the B1 was designed for low-level penetration and the B-52 retrofitted for low level, it was not possible to survive at high levels against Warsaw Pact defences.

Thanks gents. In case you have not come across it, there is a wiki for ED products. It needs some more input from us :) http://en.wiki.eagle.ru/wiki/Main_Page

As someone who has done a lot of modding for FC2 and DCS I feel I have something to add to this debate. It will be great if all these modules come out. However, something I have noticed is that the APIs in the past have not been very stable. For example, after doing extensive work to collect outputs for the purpose of pilot statistics from A-10C a point-change release was put out by ED and that broke all the access methods I'd used. In fact I now had to do a huge amount of gymnastics to collect data and some info worked and some didn't (thank goodness Speed had worked through a lot of this - work that later became Combined Arms). Hopefully since the APIs are now 'official' they will be kept stable. If they are not then the 3rd party devs are going to be in a world of pain (a special region underneath DCS World). Stable APIs have not been the case in the past as FC/FC2/DCS has had to morph to accommodate new features. This is understandable but extremely disruptive to third party devs. So much so I basically gave up modding for these products (and I had moving map software working for FC2 and would have been great for FC3 aircraft, but it wasn't worth the hassle for me to port the Lua [although the bulk of the clevers was in Java, so I could run it on my Mac]). So, some skepticism is warranted based on historical experience. What would be excellent is if: * ED was able to keep the APIs stable. No changing the event outputs due to enhanced mission triggers, or different aircraft, or JTACs/vehicles etc. * ED provided great documentation on the public APIs - including notes on what values will break the sim (it is so easy to supply values that cause the sim to crash). I believe this is something that CptSmiley has been patiently asking for. It is needed by all 3rd party devs if they are to get 'DCS' quality behaviour of their mods. * 3rd party guys made sure they isolated their calls to the ED API to a small subset of their code (this is called, "abstraction and encapsulation" of the ED implementation). That way if the ED API changes they won't have to do a huge amount of re-work (I did this in my own code but the changes were too extensive at that time). Fingers crossed lots of these aircraft etc makes it past the finished line.

Well, one day you come across an target zone with more threats than you have missiles for. You can't kill all targets. In a multi-player tournament quite often you'll only have a few minutes over the target area where you top cover is overhead. You have to kill your *targets* (boring stuff like trucks and tanks) and don't have the time nor stores to wipe every last SHORAD off the map. That's why it is good to mix it up and practice the quick striking team tactics, from time to time :)

I find the following tactic very useful when you have a human wingman. Adopt trail formation at the Initial Point (about 10 nm from the target area). The lead aircraft aircraft goes in a "shooter" and the trailing goes "cover". The trailing aircraft has two objectives: 1) keep sight on the shooter, and 2) look for AAA or missile launches on the shooter - and call out when he sees them (and probably an imperative, "Break left! flare! missile 3 o clock low") If you are not hearing warnings from your cover then the missile is not aimed at you. This especially works when other players are not on comms or not calling "Rifle" for their shots. Once the shooter is out of missiles you both extend back to the IP, swap over roles and the previous cover is now free to shoot with someone spotting for launches near him. This is the ground attack equivalent of air-to-air attacker and wingman (where the wingman has the exact same roles: keep visual on lead and watch for threats). With more practice you can swap between shooter and cover roles quite quickly as the situation arises. The important thing is that if both of you are concentrated solely on attacking then you are vulnerable to lots of threats. Make one guy offensive and one in a defensive posture and you will both kill more targets overall and both make it back. This is why Western combat aircraft *always* plan to work in units of pairs or multiples of pairs for such "search and destroy missions". Pre-planned target missions appear to be run differently, with all attacks occurring simultaneously from multiple directions (make "one pass, and haul ass" outta there). Plus it is a lot of fun to work together as as team :)

One thing about Russian designed aircraft is they are still dependent on high-performance chips only obtained from the US. Here's an article for you guys to chew on: http://www.nytimes.com/2012/10/05/world/europe/russian-agents-smuggled-us-microchips-fbi-says.html The supposed Soviet "wonderweapons" (aircraft and missiles) look great on paper, but in fact their real-world performance is less than the specs would suggest. That is not to say they are paperweights. They are still dangerous, but less than many people think. Anyway, besides the relatively limited funds in the Russian economy for building new stuff an even bigger problem is that critical parts for the Russian arms industry cannot be obtained except in very limited smuggled quantities. Actually, this had been a problem for a long time. During the Cuban Missile Crisis the Soviets had many ICBMs that were missing critical guidance parts that had to be smuggled from the US; the Soviets were far weaker than they appeared by counting missile bodies (something the Soviet leadership were well aware of, which is one reason they knew they had to back down). Based only on aircraft maneuvering performance the Flanker has a good chance of beating an Eagle. Once you add in realistic battle factors (USAF pilots that routinely train to beat 2:1 or greater odds against them), better C3I used by the US, more reliable aircraft and missiles, better maintenance, vastly better Western ECM/CPU power and software+electronics, and tactics designed to try keep the fight BVR then you'd have to bet on the Eagles.

There is no realistic jet combat alternative to DCS. 'Customer retention' is pretty much a non-issue in the current captive market. So let us all get real. We're just gonna have to wait until ED are good and ready. No point in getting into a lather.

The Su-25T has the Merkury pod for night operations, whose output is shown on your Skhval. You don't actually need NVG for this aircraft.

It is definitely the GPU that is the issue. 1 GB of VideoRAM is sufficient, but more is better. Other things you can check: * Anti-Aliasing (AA) should be on in the driver (ATI Catalyst Control Panel) or in game (applies to DCS World, not to LockOn) but never both or your game will run like a slug. * similarly for An-isotropic Filtering (AF). Switch on in the driver or in the game but not both. The rules are: * AA and AF should be set in DCS but off in the ATI driver (Catalyst Control). * AA and AF should be set on in the ATI driver when using LockOn (since there is no in-game control for these). Unlike a previous poster's suggestions there is nothing wrong with ATI. In fact, it generally gives you much better 'bang for the buck' vs NVidia. As an owner of both brands of GPU I'd also say that the visual quality of ATI in ED products generally has less artifacts. However, NVidia beats ATI hands down with regard to stability of its drivers - with ATI it is always a lottery whether the latest driver will break something (eg. like TrackIR support in 12.8 ). Once you find a stable ATI driver version it is good to stick with it. Anyway, please check that you are not doing AA and AF in two places. Doing so uses a lot of your precious VideoRAM and means that textures get swapped between VRAM and normal RAM, and things run pretty slow when that is happening.

Little-Dog did a mod called EFA (Extra Flyable Aircraft). All aircraft are flyable with this mod, but use one of the existing cockpits. eg, F-16C, A-10A, MiG-29A/C/G, Su-27/33 etc. See: http://forums.eagle.ru/showthread.php?t=52783

I think you've misread my first post (or perhaps, it was mis-written?). We agree that pilots are trained to apply G progressively to help with avoid G-Loc. My point was that there was some irreducible lag due to the flight control system and actuator slew rate. I was heading off the argument that there is already lag in the system - but that lag is not enough to reduce the G onset effects - pilots must consciously soften their inputs as they go to high Gs (unlike us in simulations that can yank much harder and get away with it - which some people may not realise is not realistic). So I think we are actually in agreement :)

Apologies to those who I saw the value I wrote of 4.95 ms. This is incorrect, the correct figure is 10 times higher (0.0495 s == 49.5 ms) and in addition the surfaces have rate limited responses (partially due to software, partially due to physical inertia, I guess). Here's the relevant section (Appendix A of NASA Technical Paper 1538 ): So, everyone was write that 5 ms was negligible. 50 ms becomes noticeable (the average human can notice changes in video rates 25 Hz = 40 ms). Add on to the half-second or so control response rate from zero to full deflection and there is a bit of lag in the system. This is the fastest you can alter controls. My understanding is that pilots are trained to progressively apply G over several seconds to decrease the likelihood of G-Loc. Again, my apologies for writing the incorrect figure. The conclusions you made were correct for the data I wrote. I hope with the new figure you'll reconsider my proposition that there is human-noticeable lag in fly-by-wire systems (at least the older ones) coupled with mechanical rate limits. Even with that minimum lag additional commanded delays need to be made to reduce the chance of G-Loc.