• Which the release of FS2020 we see an explosition of activity on the forun and of course we are very happy to see this. But having all questions about FS2020 in one forum becomes a bit messy. So therefore we would like to ask you all to use the following guidelines when posting your questions:

    • Tag FS2020 specific questions with the MSFS2020 tag.
    • Questions about making 3D assets can be posted in the 3D asset design forum. Either post them in the subforum of the modelling tool you use or in the general forum if they are general.
    • Questions about aircraft design can be posted in the Aircraft design forum
    • Questions about airport design can be posted in the FS2020 airport design forum. Once airport development tools have been updated for FS2020 you can post tool speciifc questions in the subforums of those tools as well of course.
    • Questions about terrain design can be posted in the FS2020 terrain design forum.
    • Questions about SimConnect can be posted in the SimConnect forum.

    Any other question that is not specific to an aspect of development or tool can be posted in the General chat forum.

    By following these guidelines we make sure that the forums remain easy to read for everybody and also that the right people can find your post to answer it.

hardware setup for extremely high polygon count objects scenery

Messages
64
Country
australia
hi guys,
im planning on building large numbers of scenery objects (specifically buildings) each with high polygon count.

the issue is of course what hardware configurations would allow this kind of scenery complexity whilst maintaining at least 30fps or higher?

im assuming that perhaps a multichannel situation would be needed, however can groups of similar scenery objects that are very complex be spread between several channels?

if only 1 channel assigned to scenery, would perhaps the channel assigned to scenery be better if it were a multi-processor gaming server system (such as the evga series of gaming servers)?

example 1,
imagine that there were 1,000 homes visible within a 2 km radius around sydney airport and each of these houses contained 40-60 polygons...and ideally i would like this scenery to extend out to at least 5km from airport.

or example 2,

i wish to model both the interior and exterior of airport terminals, buildings and hangars...including window and door frames, stairs, lounges, dining suites, televisions etc.
 
Hi,

What matters most for the performance is the amount of drawcalls. So if you use the same texture on many of the houses you will be surprised how well the performance can be. Around 50 triangles for a house does not sound overly complex, I don't expect too much trouble from that.
 
The software will be the limiting factor in terms of capacity, assuming you are using the fastest stuff available.

I don't know anyone who is running FSX/P3D on a 'multichannel' setup, or indeed how that works, but within the software:

A 65535 triangle limit per drawcall means that you would be able to cover your 5 km with approx 3-4 drawcalls, but not less than this (unless you make the houses less complex)

In FSX the YSSY to YBBN corridor is somewhat of a performance 'black-hole' with default scenery only.

To afford the airport and ground surfaces justice, your battle will most likely be with memory utilization within the software itself.

Whilst the idea of.modelling every doorframe and TV is appealing, the realities are somewhat different, when you factor in how the sim works.

On approach to runway 16L your Qantas VA pilot will be annoyed by any stuttering, while FSX dutifully renders the 100 TV's in the International terminal, that he can't see!

Ah, just LOD them out, and then performance is .... worse! Damn.

My point is: sure include cool stuff that you like, but concentrate on stuff that pilots will see.

YSSY is an ambitious project enough. Just ask the guys who made YBBN!

Make a start, and your stamina will tell you how much detail you want to put in.
 
thanks Arno I hadn't thought about that...very good idea!

YSSY is an ambitious project enough. Just ask the guys who made YBBN!

Make a start, and your stamina will tell you how much detail you want to put in.

Hcornea,
that's the main reason why I wanted to use an architectural software for modeling any airport... it is significantly faster than trying to freestyle model complex buildings using 3dsmax or blender.(btw, I wasn't thinking of doing Sydney Intl, that was just an example)

To give you an idea of how fast I can model houses...

on my other computer I have an image of a 3d house (scaled up to enormous size) that I have flown a mooney bravo inside of. The house interior is quite well modeled, contains a few items of furniture, a kitchen, and tv set...with timber flooring. the entire process of modeling the house and placing it into prepar3d took me less than 1 hour...in fact now that I have worked out how to do it efficiently I can now easily do the same thing in half that time.
The beauty of my software package is... in the future, when systems are powerful enough...

1. I wont have to remodel the entire structure in order to generate a fully immersive 3d environment both interior and the exterior. My doors and windows can be animated in blender in just a few minutes without and structural remodeling...they already contain everything required to look like the real thing when animated.

An aircraft hanger is a breeze to model. An aircraft can be towed inside it through fully animated doors...and the interior is not merely a photo on a flat plane...it contains real structural framing

2. My 3d architectural software is able to export directly to .3ds...allowing for direct import into ModelConverterX without the need for either Blender or 3ds Max. (obviously, there are times when a little tweaking in one of those packages is necessary)

I can think of one situation where Model converter X seems to be giving me grief right now actually...It refuses to increase the elevation of the object in the simulator irrespective of what number I put in the elevation dbx during the "convert and place process"! For that, I feel there will be times when I must go into Blender to fix it...although I haven't experimented with changing elevation of the building in my own program then sending it to Modelconverter X...perhaps that may work

Will the number of drawcalls available in current 32 bit versions of prepar3d/fsx be increased in the 64 bit version of Prepar3d when it gets released in the future?
 
Hi,

The amount of drawcall available has nothing to do with 32 or 64 bit. That part of the performance is determined by your GPU. For every new material the graphics card has to change its rendering state. That gives overhead and therefore reduces the performance. So by minimizing the amount of drawcalls you reduce this overhead and speed up rendering.

About details, in general it is much better to use textures to show detail on your object, instead of using polygons for them. You can do a lot with well made textures and still have good performance.
 
The main reason that structures in FSX / P3D do not have all the details you mention is not simply to do with 'time to model'.

It is mainly a performance-based decision. The autogen engine in FSX is MUCH more efficient at organising what needs to be rendered.

That said, it takes a 1-2 seconds to annotate a house onto photoreal scenery. My recent scenery has something like 70,000 of them.

At 1000 houses in a 2km radius, you will have 6250 in a 5km radius, which will equate to 3125 hours of house modelling, at your current workflow pace (not allowing for experience and increasing efficiency) If you become twice as quick, then you will 1562 hours (just for houses).

You can similarly add architectural details to the internals of hangars, again not requiring a great deal of modelling time. It has been done, but most designer choose to preserve the sim's resources in ways that are useful.

In short, hardware alone will not solve all your problems. FSX / P3D runs with a max of 4Gb of memory. The fastest computers will start to choke if you feed them enough detail, so you have to choose what you put in!
 
point taken...however...my experience in Industrial Technology and Design Education comes from a university degree and 10 years experience in business as a result of that degree...not from illusionary ideology. im more interested in future development than I am in being stuck in "the here an now" with current limiting factors! Arno sums it up well in his signature

"If the world should blow itself up, the last audible voice would be that of an expert saying it can't be done"

What happens when simulator systems and software are capable of running more complex models...I see you having to spend thousands of hours redoing all of your current building models! Whereas I can simply turn off the layers containing the internals of my buildings so they are not imported into ModelConverterX (or blender for that matter)...reducing the complexity considerably!

I spend a considerable amount of time in high end online gaming (have to blame some year 10 students here in Sydney in my former teaching days for introducing me to that habit)...I would suggest that its time the flight sim community started looking very closely at the kind of imaging, modeling and animating used in that area...if we don't attempt to follow suit...well...we are hanging off the end of a burning stick!

My belief is people, including QANTAS pilots, want to be able to fully immerse themselves into a simulator...one capable of simulating the entire environment around them...not just what is visible on final approach! If what you say about final approach is true...why do we spend so much time with things such as commercial aircraft cabin interiors including lighting, music, LED screens on the backs of seats, and aircraft exterior door animations? (none of which are visible from the cockpit of an A380 during final approach)

What about the enthusiast, or military client who wishes to be able to integrate flight, automotive, personnel, and marine simulation experiences all in the one package. Imagine being able to hop out of your aircraft at Redcliffe airport...into a car, drive to Brisbane...play a little Counterstrike style shooting game at an internet café...head over the road to a go cart race track...etc.

Or perhaps the movie avatar excites one enough to wish to simulate that entire environmental experience in the simulator???

I believe the future of simulators like Prepar3d intends to focus more and more on simulating an entire real world experience!

I would love to know how many hours of simulations Seal team 6 put into the preparation of the Bin Laden mission! I am certain that simulation experience did not just involve the aircrew/aircraft simulation!!! (no offence to my A380 QANTAS pilot mate) I do not believe the best method of training and preparing those guys for the task ahead was to separate the simulations on different packages and have the aircrews, seal team, an other associated personel train in a disconnected fashion.

not having a go at anyone btw...just food for thought. Essentially, Im interested in learning what the limitations are currently, how to side step around some of them...and to plan ahead a little that's all.:)
 
I'm still at a loss, Adam.

What sim platform are you going to be using? An existing platform, such as P3D or FSX, or a future platform or one that you are developing yourself?
The original question was with regard to hardware to run a flight simulation.

The hardware will be relevant to the program you are running on it, and the O/S environment in which it runs. The limitations, quirks etc will all be related to this, and how the code was written take advantage of the technology available at the time [sic FSX].
If you are referring to a hypothetical simulator, then there is no real answer to your question.

Without a platform all you have is a hugely complex set of 3D geometry.

Incidentally ... airports and cities alter as does technology, so the highly detailed scenery of 2009 may be out of date in 2013 (after a few runway upgrades, a new terminal and a housing estate) Then you need to re-do it. My personal experience, the geometry outdates often quicker than the technology.

You'll no doubt have noted i have published work through ORBX, and we tend to put more detail than most developers into FSX and P3D scenery. On that basis, you'll appreciate I am speaking from some degree of experience about pushing the boundaries of what the sim can do, and also of customer expectations.

Remember, if you have your eyes firmly fixed on the future, you never get anything done now.

I am looking forward to see your sim offerings in the showroom.
 
Last edited:
im more interested in future development than I am in being stuck in "the here an now" with current limiting factors! Arno sums it up well in his signature

This could change weekly. The future is ALWAYS coming, thus every week you could do something to upgrade.
Strictly for hardware, you need to decide what you want as far as performance. This all depends on, as Hcornea stated, what OS and software you plan to run on the machine. Decide in the now, not so much on whats going to be available in the future. Doing it based on what the future holds can cause you some headaches and leave you with an unfinished project, always 'waiting for the future' software or hardware to come out.

Whatever comes out in 6 months from now, software wise, will still run on todays machines just fine. People wont buy a new pc just to install 1 program that couldnt run on their 5 month old computer, therefore most software designers design it for the future, but usually base it on the hardware available today, even if all functionality is unable to be attained, its still able to run. This is also where Service Packs come from.
 
Hi,

I don't think fsx will ever show interiors of houses and all such details. You simply don't need them from flight perspective (except maybe the terminal where you park).

Even for collective military training where you would like to link aircraft and infantry in the same environment, the different simulators do not have the same amount of detail for the environment. The infantry simulator will indeed have buildings you can enter, with interior. But the aircraft simulator will show a more simple version of the same house. The different systems have different requirements, since the aircraft simulator probably needs to show hundreds of kilometers of terrain, while the infantry only needs a few kilometers.

And eye candy like detailed hangars etc looks very nice, but in general it hardly adds any training value. For a civil flight simulator the focus will remain on landing and handling emergencies.

So who knows in the future there will be a platform that can handle more detail but then it is very likely it uses a different format and different modeling tools. So I wouldn't add items that are not expected to be needed on short notice. Because if you want to use it in a different platform you are likely to need to alter other things as well. So often it is best to start again with the requirements of the new platform in mind.
 
I agree that fsx probably isnt the platform for such ideas, although i feel that perhaps the inclusion of ATC as a viewpoint does provide some insight into what the future thoughts of FSX developers may have been at the time. However prepar3d on the other hand would seem to have inferences of such grand ideas in that it has included a submersible, x-plane is making allowances for offworld simulations...there are images of motorcycle and automotive vehicles in the sim too.

I have no intentions of trying to develop a new platform of my own, goodness that is way beyond the resources of someone like me!

I guess i merely am probing ideas for striding around the framerates issues that keep rearing their ugly head all the time. I am certain that if we could simply setup hardware in a manner that efficiently handles these restrictions perhaps the immersive experience of a full flight, automotive, marine, or avatar may not be as outrageous as it seems right now.
 
You mentioned Seal team training...

which is, in context of a civil airplane simulator, functioning in a truly worldwide scale, a bit of like comparing apples to oranges... or like one dr. Sheldon Cooper once put it, comparing apples to weapon-grade plutonium.

I most likelihood, the teams involved trained in a simulator similar to VBS, possibly in VBS itself.

VBS, is a simulator built upon the foundation of the original Operation Flashpoint, later followed by Armed Assault and ArmA II.

As you would know, these allow for infantry, armor, airborne etc. cooperation in a very realistic rendering of a small part of world, sized in single-digit square kilometrage in OFP to triple-digit square kilometrage in ArmA II.

This is, as such, completely incomparable with a civil simulator of FSX/P3D type, handling a single aspect of the simulation, but world-wide.

The pilots that took part in that operation did not need to fly their helicopter realistically, in any part of the plausible world, they needed to learn to orient themselves in a very specific, very small environment, and cooperate with infantry. In this environment, nothing that makes FS what it is, is needed. No realistic dynamics, no worldwide coverage, no failures, no nothing. What it DOES need is completely irrelevant to FS type product, extremely detailed environment, including interiors, provisions for different vehicle types... A flight simulator will never need this level of detail, so you are just loosing time if you make it with visions of future use.
 
Back
Top