FG: VOXELS VS POLYGONS ?

[UR67393]

While viewing videos with beautiful latest Outerra and Proland I came to my convinced opinion: FlightGear flight simulator must have free 3D voxel engine for generating accurate, beautiful terrain. And will be better if FG will have prepared free voxel terrain of the whole earth- are You agree? .                                                                                                     With the appear of Advanced weather, Rembrand I believe that the emergence of virtual FG 3D voxel terrain is necessary. The polygon and voxel graphics have their better parts and disadvantages. Using better parts of voxel and polygonal graphics FG believe will lead to a better view and hope to large values of the FPS. FG will become a newest and super modern of 3D graphic engine, isn`t it?                                                                                                                                                                                 Is there used to be such an idea about FW 3D voxel engine integration with prepared FW 3D voxel terrain? If this idea was what are the results? Does it make sense to start a discussion about the integration of FW 3D voxel graphic engine in FG with prepared FW 3D voxel terrain?                                                                                                                                    What is opinions of experts about 3D voxel graphic engine with(poly engine of course) in FG? You are welcome- Your opinion is interesting. Thank you in advance.

[mgh]

FlightGear is open source so you could write a voxel engine to interface with OpenGL.

[jcomm]

And the new version is about to be released, mid August, just a few days from now :-)

Not voxel-ready, but I am willing to try it for sure!

[UR67393]

Hello dear users! Thank You for Your answers. My hope at user`s answers began even to fade. My sincerely congratulations at new release of FG to all developers and all FG`s users. FG becoming better and better, but virtual ground in FG very need at improves.                                                                                                                                                                                                                                                                     To mgh- thank You at Your suggestion, but to my sadly I`m not programmer, and I do not know any pc language, really. I can`t write a voxel engine for FG...                                                                                                                                                                                                       To jcomm- are You try to write of voxel engine? =) Am I understand correctly? If it correct it may be at any FW 3D graphical voxel engine on network? And may be such FW 3D voxel graphical engine can any remaked at fs features? It is not such long time for create.              

[stuart_d_buchanan]

Hi UR67393,

 

I think the probability of FG changing to support voxels in the near future is very close to 0, and I say that as one of the FG developers responsible for some of the graphics.  That said, if you wanted to work on it yourself, you are very welcome, and there has been significant work recently to separate the rendering engine from the underlying simulation..

 

The entire scenery toolchain is based on polygon data and would need to be re-written, plus FG uses OpenSceneGraph as it's OpenGL interface, and that would have to be modified to handle voxels. 

 

-Stuart

[UR67393]

Hi UR67393,

 

I think the probability of FG changing to support voxels in the near future is very close to 0, and I say that as one of the FG developers responsible for some of the graphics.  That said, if you wanted to work on it yourself, you are very welcome, and there has been significant work recently to separate the rendering engine from the underlying simulation..

 

The entire scenery toolchain is based on polygon data and would need to be re-written, plus FG uses OpenSceneGraph as it's OpenGL interface, and that would have to be modified to handle voxels. 

 

-Stuart

 

Hi Stuart.

Thank You. I`m not programmer and I can not to write voxel engine for FG... Stuart, pardon me, but prompt me please, the FG developers had the idea of ​​integration with the developers of Proland ( http://proland.inrialpes.fr/ )? You may know this site, I suppose) Proland have GNU/GPL licence  But may be ​​integration not possible due to technical reason, I suppose too...

 

 

 

And else, I only notice next- if we increase the amount of polygons I suppose that for  FG will need at pc with greater and greater power, as for XP10 now, isn`t it? And in the future it may be to grow at big problem... Need use all advantages of voxel and polygon graphics. It main priority of further development- and may be with Proland? Or without?  

Thank You for attention.

[stuart_d_buchanan]

Yes, I think we've looked at Proland in the past.  I don't think there's anything the Proland offers that we couldn't develop within our existing framework.  Fundamentally it all boils down to data being written as OpenGL primitives and rendered using shaders. 

 

Regarding polygons - at the moment we don't think that we're limited by pure polygon/vertex count, provided those vertices are sensibly arranged.  Modern graphics cards are very good at churning through huge numbers of polygons.  What tends to be more of an issue is when the polygons are organized inefficiently, or rely too much on complex shaders.  A case in point is the trees and random buildings, which generate millions of polygons, yet render quite rapidly, whereas having a much smaller number of discrete models slows things down significantly.

 

One area that I'd like to look at if I get the chance is transformations we can do within the graphics structures to make it more efficient at loading time.  However, that's quite a big project, and like more FG developer's I've got plenty of other things on my plate!

 

-Stuart

[MikeT707]

Yes, I think we've looked at Proland in the past.  I don't think there's anything the Proland offers that we couldn't develop within our existing framework.  Fundamentally it all boils down to data being written as OpenGL primitives and rendered using shaders. 

 

Regarding polygons - at the moment we don't think that we're limited by pure polygon/vertex count, provided those vertices are sensibly arranged.  Modern graphics cards are very good at churning through huge numbers of polygons.  What tends to be more of an issue is when the polygons are organized inefficiently, or rely too much on complex shaders.  A case in point is the trees and random buildings, which generate millions of polygons, yet render quite rapidly, whereas having a much smaller number of discrete models slows things down significantly.

 

One area that I'd like to look at if I get the chance is transformations we can do within the graphics structures to make it more efficient at loading time.  However, that's quite a big project, and like more FG developer's I've got plenty of other things on my plate!

 

-Stuart

Do you see FG getting to a point in the next couple of years of graphics being as polished as those that display within Proland or Outerra?

[UR67393]

Yes, I think we've looked at Proland in the past. I don't think there's anything the Proland offers that we couldn't develop within our existing framework. Fundamentally it all boils down to data being written as OpenGL primitives and rendered using shaders.

 

Regarding polygons - at the moment we don't think that we're limited by pure polygon/vertex count, provided those vertices are sensibly arranged. Modern graphics cards are very good at churning through huge numbers of polygons. What tends to be more of an issue is when the polygons are organized inefficiently, or rely too much on complex shaders. A case in point is the trees and random buildings, which generate millions of polygons, yet render quite rapidly, whereas having a much smaller number of discrete models slows things down significantly.

 

One area that I'd like to look at if I get the chance is transformations we can do within the graphics structures to make it more efficient at loading time. However, that's quite a big project, and like more FG developer's I've got plenty of other things on my plate!

 

-Stuart

 

 

 

Thank You Stuart.

I understand You, I have not at flight sim enough time too. All my questions and notices not commands to action. It only questions and notices- before and after I understand that the integration of FG and Proland labor intensive and timely process.

 

 

My position- unreasonable and impractical to expect in the future for the most part on the power of video adapters. Of course now yes, but on future no- if I'm not mistaken Proland uses fractal graphics technology. I sure FG developers know about it )

I only want to add else, let`s remember next: FW poly FG began to develop as an alternative to commercial polygonal simulators , and this is a success , because FG have open source, new lighting, luxurious sunsets , interesting weather engine, and of course the famous JSBsim and more.

Then there was a closed -poly Projects FSX and for new polygon XР10 requires the PC is much more powerful .

The question appears- will the FG as alternative for newest commercial voxel 3D graphics engines, such as Outerra for example?

I believe that the time polygonal commercial sims and commercial 3D graphics engines start to finish , and if FG will be integrated with such Proland while the existence of commercial voxel engines will start to run out, I am sure that FG developers understand it

If I'm not mistaken Proland uses fractal graphics technology , i.e. I assume that this technology is similar to the fractal technology as in Outerra. It is particularly important that the FG and Proland are open source, and it is very important for developers because addons can do much better. Perhaps open project will allow to us in future make integration with another simmers and their developer(addons)- railway auto water and may be another. For VFR flights as example we very need at correct railways and railway station, need at roads(it all name as line landmarks) and etc. We are together will fill up virtual space our addons which helpfull for all. It very important for all I think... Stuart, if the idea of integration and FG and Proland approved by You- it may be necessary to address the developers of other free add-ons for simulation-car train, and so and so? Stuart, if the idea of integration FG and Proland approved by You- may be necessary to ask of developers of other free add-ons simulation(ato railworks and etc) their opinions about integration their works in Proland? What do You think and another? I think it very need and important for coordination of works on general joint integration. I'll try tо ask them, but I do not know the address of the sites, and may be somebody tell me what are the free auto railway water transport fw(!) simulators? I'm just a user of fs. Thank You for in advance for Your advices.

 

 

By the way, as You all know, on polygonal fs big problem is animation of vegetations and water. Why should animation of vegetation and water? Answer- by variations of vegetation and the raging of the water a pilots set power and direction of the wind at the ground for take- off and landing. By the way in Outerra have small variations of vegetation. And for FG need too, but with integration of voxel Proland, I sure it very important for all I think too... This is a new stage in the display for the new fs.

I notice else- it only questions and notice, before and after and they addresses to those users who are small know Proland and FG too. But my questions and notices, and answers of developers and users already need- it's time for this, isn`t it?

Next to experts and users, I have nothing to add. Thank You and all for Your attention.

[stuart_d_buchanan]

Hi Guys,

 

Mike T707 - No idea when/if we'll get to Outerra/Proland levels of polish. At one level, we're fairly close with procedural texturing and various shaders.  If you look at some of the screenshots that Thorsten Renk has posted on the flightgear.org website showing off atmospheric scattering and his procedural texturing work they are quite comparable.  However, our autogen and vegetation rendering isn't as advanced, and I'm not sure how to do it without hitting frame-rates massively.

 

It's also worth pointing out that the screenshots from both Outerra and Proland have a fairly narrow scope - they are generally showing temperate forested mountains and have been tuned for rendering those specific scenes.  For FG we need to be able to render a much wider geographical area (deserts, farmland, tundra), plus urban areas.  We also seek to do so in such a way that someone without programming knowledge can fairly easily create an environment that matches their local area (we've got different terrain for the Caribbean for example)

 

Finally, we're very short of graphics programmers.

 

UR67393 -  Your comments about polygon-based simulators coming to the end of their life is very interesting.  As I said previously, we have put some effort into separating out the FG core simulator from the graphics - there is a component call fgviewer which implements this.  It should be possible to integrate with Proland, but it all boils down to having the development resource to do so.  It took about 3 years for us to completely convert from plib to OSG, which was simply a conversion from one polygon based system to another.

 

You have enthused me to take a further look at proland .

 

-Stuart

[UR67393]

Hi Guys,

 

Mike T707 - No idea when/if we'll get to Outerra/Proland levels of polish. At one level, we're fairly close with procedural texturing and various shaders.  If you look at some of the screenshots that Thorsten Renk has posted on the flightgear.org website showing off atmospheric scattering and his procedural texturing work they are quite comparable.  However, our autogen and vegetation rendering isn't as advanced, and I'm not sure how to do it without hitting frame-rates massively.

 

It's also worth pointing out that the screenshots from both Outerra and Proland have a fairly narrow scope - they are generally showing temperate forested mountains and have been tuned for rendering those specific scenes.  For FG we need to be able to render a much wider geographical area (deserts, farmland, tundra), plus urban areas.  We also seek to do so in such a way that someone without programming knowledge can fairly easily create an environment that matches their local area (we've got different terrain for the Caribbean for example)

 

Finally, we're very short of graphics programmers.

 

UR67393 -  Your comments about polygon-based simulators coming to the end of their life is very interesting.  As I said previously, we have put some effort into separating out the FG core simulator from the graphics - there is a component call fgviewer which implements this.  It should be possible to integrate with Proland, but it all boils down to having the development resource to do so.  It took about 3 years for us to completely convert from plib to OSG, which was simply a conversion from one polygon based system to another.

 

You have enthused me to take a further look at proland .

 

-Stuart

 

Hi all!

Thank You Stuart =).  FG load me of enthusiasm too - sky clouds weather, rembrandt ). Beauty Proland and FG will becoming much better together, I sure, isn`t it? =)..   Of course I understand a little of size work which need. But I only notice- we need think about integration, with another transport simmers together, with Proland now, because integration after may be reach much more time. I.e. need mandatory coordination of joint action, it  help us at development resource for integration and optimisation. I will try to invite to this topic for the dialogue of developers and users of other transport simulators, I need time for announcement on other sim transport sites... I`ll try report about opinion another transport simmers if they not answered in this topic. We need to use better parts of voxel and polygonal graphic in future, note please- better parts. Thank You for attention. Best Regards!

 

PS:  For name of integration project after Proland need add name simulator. i.e. it will get name as Proland sim, or simply PS