About adjusting the contradiction between quality and performance

[biller]

First of all I have to admit that what the DCSW is doing is without parallel in history .It looks like that ED is not only expert in aero dynamics but also expert in graphic processing.

However I think that recently DCSW is (maybe slightly) not on the right way when it comes to the quality and performance.

 

 

 

The graphics and effects are not the primary concern of a simulator ;they are the primary concern of games like ACE combat or HAWX. If we must recite what is very important(related to the monitor) to a flight simulator,it should be the continuity,or FPS.

 

In early 2013, the DCSW had a very good optimization, many PC can handle it. As the time passing , with many new models with large amount of polygons being added to DCSW, the game gradually became a hard ware killer.

The 3GO cube aerobatic team said once that they cannot train well with DCSW/FC3 because when they tighten the formation the fps will become very low .But in FC2, there is no such thing. I think it is caused by too much polygon computing,even they set textures,sences...to low. Since I do not have a tool to check the number of the polygens of one SU-27, I have to compare that with the gys F-15C mod.

The magnitude of the number of the polygens of the ED's new SU-27 model should be at about 100,000, if 5 SU-27 fly togeher, the wingman's PC will have to process at least 400,000 polygens at the sametime while the computer still have to deal with other dynamic equations( when you introduce the AFM/PFM model the situation will become worth).

However with my experience, a model with just 10,000 polygens is enough to be used as a high quality model for a flight simulation.

But what the ED is doing is trying to make one singel model more and more complex, turning an air combat simlation to a singel high quality aircraft 3D model panorama show.

With the CPU killer AFM/PFM and very complex 3d model ,now even a pretty good PC cannot maintain enough FPS when it runs a medium lever combat mission.

Even a good sever will crash when it runs a mission with 1000 vehicle with the 3D rendering already shut down.

 

The falcon BMS4 does not have many complex vehicle 3D model but it is still

very successful.

When your target is 10km away, even a 500-1000 polygen model will be enough.

 

So my conclusion is that the necessity of high polygon model for a sim isn't very urgent.

For example, the new MIG-21bis mod uses a lot of polygens to form the external cockpit details, it is meaningless.

And check the models in ACE combat 6 and you will find the new F-15C's model in DCSW has already crushed that in ACE combat 6,or 7

The graphics of the planes,tanks or ships should be concerned when everything else is done and with all respect I don't think that ED did well in hardware resource managing. A lot of senior LOFC/DCS fans have been driven to the BMS4. And among 3 major DCS/LOFC fans club in China, two of them start to hibernate, one start to split. It is related to the soft ware itself.

[biller]

Like this..

en..

Where is my attachments...?

[Exorcet]

I do not like to see graphics as the primary focus of a game, however, DCS being a sim does make the graphics important. Often times you may trying to pick out small things in the landscape (MANPADS launch, another plane in a dogfight) and having real life detail in these situation is important.

 

When the F-86 came out, I went online to dogfight in it. In the initial release, the plane had one skin (for all countries) The only way I could ID aircraft was by reading the numbers posted on the side of the fuselage. That might be a bit exaggerated since different sides usually have different camo, but it makes the point that visuals can be important. When you have similar looking aircraft like the F-86 and MiG-15, differences between the 3D models become even more important.

 

Also, when it comes to performance, ED is addressing this with EDGE.

[nervousenergy]

While I agree that the multiplayer performance of 1.2.10 is one of its weakest links, I'm thinking you should hold off on criticizing the graphical direction ED has gone until we see EDGE 2 in motion, hopefully sometime in the next 3 months. I'm personally trying to avoid too much over-anticipation, but the main thrust of EDGE according to the devs and test teams working on it is optimization - delivering better visuals than 1.x while consuming fewer compute resources. Let's see how well they do before you break out the pitchforks, as right now we're running cutting-edge models on a 3 (4?) year old graphics engine. And even on that engine, I'm really not getting consistent performance drops just because several planes are in the air near me... it's actually quite a bit more random than that. I can have 4-5 planes swirling around me with flak bursting randomly (note that I only play WWII multi, not jets, though I'll be happy to fly Korean multi when we get a flyable Mig-15) and it's smooth as silk, then get 1 FPS for a few seconds while flying away from it. I'm hoping EDGE optimization will correct much of that.

 

If you ran 2013 models without much issue but are having problems with newer ones on the same hardware, then I'm afraid the issue is simple: DCS isn't going to stand still in the PC hardware arena. If you want to run the latest models with the latest terrain, it's likely even EDGE 2 will require more horsepower than a 2-3 year old PC can provide.

 

What kind of machines are your Chinese compatriots running?

[AtaliaA1]

I don't think anyone would agree that a DX9 Software Title is an "Experts" attempt at Graphics Processing, as you say.

 

But an attempt at giving users the "Flight Simulator" we all clamored for. Unfortunately at the time of development DX9 was the Bees Knees (good thing). As the development of this "Great Title" progressed so to did the skills and utilities used for software development. Unable to do 2 things at once (Financial Constraints) starting over in the middle of development wasn't an option. i.e. adding multi-core processing/updating to DX functionality. As their hands were full developing a title that inherently comes with bugs needing to be worked out. (Keep in mind the size of the development team).

 

It has already been discussed over and over the unrealistic comparisons of this title and "Games" done many yrs. ago resulting in an agreement that this software doesn't compare to games as the level of simultaneous systems out way the "Shooter" genre in complexity. It is unfortunate that you missed out on those long chats, as they would have added a more thorough knowledge base to your discourse.

 

With this in mind would you suggest that all development cease as to wait for better graphics processing abilities of DX9? or would it better move along our genre to continue development (Even adding "Polygons"). an addition to the title that allows the continued "Forward Thinking" that will soon be a thing of the past, as the new engine comes online. in order to epees a handful of "Blue Angel" want a bees. I think not Sir. Sacrifices must be made in a small financial market as we enjoy. This is, you will admit? a Non-Block Buster Title even genre. So to compare the two is a bit sophomoric, don't you think?

Edited October 4, 2014 by AtaliaA1

[biller]

While I agree that the multiplayer performance of 1.2.10 is one of its weakest links, I'm thinking you should hold off on criticizing the graphical direction ED has gone until we see EDGE 2 in motion, hopefully sometime in the next 3 months. I'm personally trying to avoid too much over-anticipation, but the main thrust of EDGE according to the devs and test teams working on it is optimization - delivering better visuals than 1.x while consuming fewer compute resources. Let's see how well they do before you break out the pitchforks, as right now we're running cutting-edge models on a 3 (4?) year old graphics engine. And even on that engine, I'm really not getting consistent performance drops just because several planes are in the air near me... it's actually quite a bit more random than that. I can have 4-5 planes swirling around me with flak bursting randomly (note that I only play WWII multi, not jets, though I'll be happy to fly Korean multi when we get a flyable Mig-15) and it's smooth as silk, then get 1 FPS for a few seconds while flying away from it. I'm hoping EDGE optimization will correct much of that.

 

If you ran 2013 models without much issue but are having problems with newer ones on the same hardware, then I'm afraid the issue is simple: DCS isn't going to stand still in the PC hardware arena. If you want to run the latest models with the latest terrain, it's likely even EDGE 2 will require more horsepower than a 2-3 year old PC can provide.

 

What kind of machines are your Chinese compatriots running?

 

 

Most people use an i5 or i7.And they make big investment on graphic card...Like GTX 660/660TI....GTX-670...........

RAM is within 6G and 16G

 

Actually the machines of the flight sim fans here are at the same level wth those in other countries, largely because many of the design corporation have a foundry in Taiwan and Guangdong.

 

Oh I found that shot.

I don't think such a highly detailed external cockpit model is necessary

 

The question is ,e.g ,a 2013 pc can run the 2014 model fluently when there is 40 Mig-21s,a 2014 pc can run the 2014 model fluently when there is 80 Mig-21s in the view.

But when a mission editor directly put 80 Mig-21s and 80 F-4E in a mission , can the newest PC deal with it?

Updating the hardware maybe can improve the performance 2 times or 3 times,however a single change in the numbers of the units can easily cause the updating meaningless because the number of the units can be increased by 5 times or 6 times

Edited October 4, 2014 by biller

[biller]

I don't think anyone would agree that a DX9 Software Title is an "Experts" attempt at Graphics Processing, as you say.

 

But an attempt at giving users the "Flight Simulator" we all clamored for. Unfortunately at the time of development DX9 was the Bees Knees (good thing). As the development of this "Great Title" progressed so to did the skills and utilities used for software development. Unable to do 2 things at once (Financial Constraints) starting over in the middle of development wasn't an option. i.e. adding multi-core processing/updating to DX functionality. As their hands were full developing a title that inherently comes with bugs needing to be worked out. (Keep in mind the size of the development team).

 

It has already been discussed over and over the unrealistic comparisons of this title and "Games" done many yrs. ago resulting in an agreement that this software doesn't compare to games as the level of simultaneous systems out way the "Shooter" genre in complexity. It is unfortunate that you missed out on those long chats, as they would have added a more thorough knowledge base to your discourse.

 

With this in mind would you suggest that all development cease as to wait for better graphics processing abilities of DX9? or would it better move along our genre to continue development (Even adding "Polygons"). an addition to the title that allows the continued "Forward Thinking" that will soon be a thing of the past, as the new engine comes online. in order to epees a handful of "Blue Angel" want a bees. I think not Sir. Sacrifices must be made in a small financial market as we enjoy. This is, you will admit? a Non-Block Buster Title even genre. So to compare the two is a bit sophomoric, don't you think?

 

I have to say that my intention is not to let the current development to be stopped, my suggestion is that when building the model in the 3DSMAX, try to distribute the polygons wisely ,for example the MIG-21 bis,the oxygen mask obviously is a luxurious detail.

 

While I was studing the 3dsmax myself, a senior mod maker told me that a plane model with 100,000~300,000 polygons is already enough for a movie

The current CPU and GPUs' manufacturing technology has decided the scope of its graphic processing ability( xxxxxxxx~xxxxxxx polygons/second).

And our pc is not a supercomputer

 

If you made one single model too meticulous, what would happen when you have 100 of that in the sky?

Frankly speaking ,ED has already done well in graphics, its SU-27 model (EDM file),its normal mapping and its rendering.And I also remember that Wagner said on SIM HQ that DCS:A-10C support (maybe not fully) the DX11.

Of course, I'm also looking forward to the new engine but ideals and reality sometimes don't mix very well.

Edited October 4, 2014 by biller

[Archer7]

Having a hard time? Settins to minimum... solved.

Graphics are important for immersion.

Edited October 4, 2014 by Archer7

[biller]

I do not like to see graphics as the primary focus of a game, however, DCS being a sim does make the graphics important. Often times you may trying to pick out small things in the landscape (MANPADS launch, another plane in a dogfight) and having real life detail in these situation is important.

 

When the F-86 came out, I went online to dogfight in it. In the initial release, the plane had one skin (for all countries) The only way I could ID aircraft was by reading the numbers posted on the side of the fuselage. That might be a bit exaggerated since different sides usually have different camo, but it makes the point that visuals can be important. When you have similar looking aircraft like the F-86 and MiG-15, differences between the 3D models become even more important.

 

Also, when it comes to performance, ED is addressing this with EDGE.

 

I agree that visual ID and details in graphic are important.

However the methods to benefit the visual identification in a sim doesn't means it should be based on hard ware resources using.

In BMS4 although the graphic is absolutly inferior to DCS, we can still distinguish between a MIG-21 and a J-8

 

I mean that ED(Or its third partner) tend to make very exaggerated settings

http://forums.eagle.ru/showthread.php?t=51339

not only in data but also in its 3d models.

e.g they set the maximum visual distance for military operation 80kilometers........

 

Visua identification does not necessarily means more and more polygons. Will a 3000 polygon part be more recognizable than a 1500 polygon part when the plane itself is a small dot on the screen? My concept is that every polygon count. What's more many identification can be realized by the combination of texture and model

 

You may think this idea is funny,but when you've got 1000 missiles in the air ...You may feel that this missile doesn't worth lots of polygons.

[qq515616480]

I think so

My notebook configuration:

I5 M460 / GT 420M / 6G / SSD /1366*768

Acceptable quality, only FPS 20

[KLR Rico]

As the time passing , with many new models with large amount of polygons being added to DCSW, the game gradually became a hard ware killer.

 

 

It seems that the point of the OP is "My PC ran FC2 just fine, but it struggles with DCSW 2.8+"

 

The game's graphical quality improved over time... and so has the hardware to run it. Time marches on, visual quality improves, hardware becomes outdated; upgrades are inevitable.

[c84]

perhaps i am wrong, but i think second problem is big mess in lod settings. For example, lets have a look on 2 helos

 

ka-27

model = {

lods = {

{"ka-27.edm",100.000000},

{"ka-27-lod1.edm",200.000000},

{"ka-27-lod2.edm",1000.000000},

{"ka-27-lod3.edm",50000.000000},

},

collision_shell = "ka-27-collision.edm",

 

ah-64d

model = {

lods = {

{"ah-64d.edm",50.000000},

{"ah-64d_lod1.edm",100.000000},

{"ah-64d_lod2.edm",200.000000},

{"ah-64d_lod3.edm",300.000000},

{"ah-64d_lod4.edm",400.000000},

{"ah-64d_lod5.edm",5000.000000},

{"ld_tec.edm",10000.000000},

},

collision_shell = "ah-64d-collision.edm",

 

So, (if understand correctly, how this stuff works), ka-27 have nice little low res model up to 50km, ah-64d have green box (btw invisible from botton) from 5km to 10km and then nothing.

Edited October 5, 2014 by c84

[AtaliaA1]

for example the MIG-21 bis,the oxygen mask obviously is a luxurious detail.

 

An Item we flight simmers want included.

 

 

The current CPU and GPUs' manufacturing technology has decided the scope of its graphic processing ability( xxxxxxxx~xxxxxxx polygons/second).

And our pc is not a supercomputer

 

What are the actual numbers? Is it negative to add polygon counts that equal Movie levels? I want that in my Graphics the Higher the better.

 

 

If you made one single model too meticulous, what would happen when you have 100 of that in the sky?

 

I get it. but I want it that high visuals are in demand by all simmers.

 

 

Of course, I'm also looking forward to the new engine but ideals and reality sometimes don't mix very well.

 

That is true. but the work around is in use now. so it will work. and is working the testers have it as we speak.

 

 

So this is what the users demand in a sim.

[AtaliaA1]

I agree that visual ID and details in graphic are important.

However the methods to benefit the visual identification in a sim doesn't means it should be based on hard ware resources using.

In BMS4 although the graphic is absolutly inferior to DCS, we can still distinguish between a MIG-21 and a J-8

 

I mean that ED(Or its third partner) tend to make very exaggerated settings

http://forums.eagle.ru/showthread.php?t=51339

not only in data but also in its 3d models.

e.g they set the maximum visual distance for military operation 80kilometers........

 

Visua identification does not necessarily means more and more polygons. Will a 3000 polygon part be more recognizable than a 1500 polygon part when the plane itself is a small dot on the screen? My concept is that every polygon count. What's more many identification can be realized by the combination of texture and model

 

[ATTACH]105447[/ATTACH]

You may think this idea is funny,but when you've got 1000 missiles in the air ...You may feel that this missile doesn't worth lots of polygons.

 

 

 

The latest update to the Mig21 answered the call for lower graphics settings to increase FPS.

 

 

Maybe they heard you. Even if not it still is a Good Topic of discussion, Kudos to you My Friend.