Open Rails development priorities

[muskokaandtahoe]

I see that I didn't explain fully in my first post... For what I understand, it's not really the terrain polys that are an issue-- there are a variety of ways to LOD individual polys such that even with a broad landscape in the display the workload is manageable -- but the objects in the tiles are a whole 'nuther story. Add more tiles to the display window and you'll very likely be adding more objects. Perhaps lots of objects. With lots of textures. Perhaps the tiles to the left and right of the track path are not so bad... but the ones ahead and behind could be rather full and to whatever extent they are is a lot of extra work that will have an effect on performance.

[OTTODAD]

As my screenshots show, MSTS has no problems displaying Distant Mountains for as far as the eye can see.

What it can display which is on them may vary.

I thought that MSTS already pre-loads the 8 tiles surrounding the tile the player train is in ?

O t t o

[Arista]

Can't you just ignore the objects on distant tiles, and load/display only the terrain? Though I have no idea how terrain data is organized in MSTS, it is usually quite easy to effectively LOD terrain by gradually reducing the mesh resolution (assuming a regular terrain mesh) - i.e. by using only every other data point, than every fourth, etc. That's how MS Flight Sim handles visibilities in the hundreds of miles, for example.

Judith

[conrail1]

That's how MS Flight Sim handles visibilities in the hundreds of miles, for example.

Judith

I would do some serious research on the subject before using just about anything that MS Flight Sim does as an example. It's well known in the flight sim community just how hideous the FSX game engine is.

[muskokaandtahoe]

Can't you just ignore the objects on distant tiles, and load/display only the terrain?
Judith

Sure... the programmers can do pretty much what they want/can do. But that doesn't change the fact that with many levels of route details, PC's of greatly differing capabilities, and as many opinions as there are people (perhaps more), it's going to be a challenge to find a solution -- or solutions -- that work well.

[Arista]

It's well known in the flight sim community just how hideous the FSX game engine is.

The point being? FSX (or any other version, for that matter) is still capable of displaying terrain for hundreds of miles, while OTRS isn't, using a simple resolution reduction scheme that is independent of the graphics engine used. And while some aspects of the FSX graphics engine could be better, of course, in general it is quite good for what it does actually.

Judith

Edit: Quote added

[conrail1]

The point being? FSX (or any other version, for that matter) is still capable of displaying terrain for hundreds of miles, while OTRS isn't

So does the the game engine (Outerra) I used as an example and it does it a lot more efficiently which is the point of using an up to date, capable game engine.

And while some aspects of the FSX graphics engine could be better, of course, in general it is quite good for what it does actually.

My experience with FSX goes back to when it was still in beta, there are very few things about the game engine that I can say are “quite good”.

[GaryG]

I know I would prefer accurate train handling and operations over distant mountains. For now, lets add some reddish-brown fog to imitate what our skies were like this summer because we could hardly see our near mountains. Because there are different teams working on ORTS, we will probably eventually get all our wishes satisfied but let's be patient.

A comment about the long train pauses, might it be possible for the sounds to be handled as a separate CPU thread? By the time ORTS is "ready for prime time" in a year or two, the majority of users will probably have computers that are easily capable of handling four or more concurrent threads.

GaryG

[twpol]

A comment about the long train pauses, might it be possible for the sounds to be handled as a separate CPU thread?

This is one option we have been considering, though it is too early to say whether it will alleviate this problem. We are also concious of the effect of such a change on 1 and 2-core systems, which we don't want to prematurely make unusable for OR.

[conrail1]

This is one option we have been considering, though it is too early to say whether it will alleviate this problem. We are also concious of the effect of such a change on 1 and 2-core systems, which we don't want to prematurely make unusable for OR.

James,

I forgot to ask this earlier but do you know of any changes in v360 that would take advantage of hyper-threading? If not in v360 anything planned for future builds?

[twpol]

I forgot to ask this earlier but do you know of any changes in v360 that would take advantage of hyper-threading? If not in v360 anything planned for future builds?

Nothing in OR specifically targets hyper-threading and I don't expect that to change; the OS can do a better job of working out what to run where. We still use 2 threads on a single logical core and 3 on >1 logical core, though, so a hyper-threaded single core machine will run with the 3-thread OR.

What, exactly, could be taken advantage of by knowing if a machine has hyper-threading?

[conrail1]

Nothing in OR specifically targets hyper-threading and I don't expect that to change; the OS can do a better job of working out what to run where. We still use 2 threads on a single logical core and 3 on >1 logical core, though, so a hyper-threaded single core machine will run with the 3-thread OR.

Thanks, I haven't bothered to test v360 with hyper-threading enabled and with v128 it did nothing to help performance so that's why I asked.

What, exactly, could be taken advantage of by knowing if a machine has hyper-threading?

Other than the game and OS seeing double the amount of physical cores that are available I don't know, that's what I was questioning.

[jfrydom]

James,

I forgot to ask this earlier but do you know of any changes in v360 that would take advantage of hyper-threading? If not in v360 anything planned for future builds?

They could use the extra threads for particle effects. (smoke, rain etc.)

[conrail1]

They could use the extra threads for particle effects. (smoke, rain etc.)

For particle effect work load it's probably better to use the GPU in the future by way of say PhysX as an example.

[OTTODAD]

If any programmer would like to join the Open Rails team and work on some of the aforementioned topics of interest, please fill out an application at http://openrails.org/coders.html.

Although I have XNA 3.1 and Visual C# 2008 I am nowhere near being qualified to attempt coding anything as yet and without ORTS Project Files can not have a look-see.

This is as far as I got. Go on, have a laugh: http://www.otto-wipfel.co.uk/video/xna-sample.wmv

So the basic requirements for anybody thinking about joining the team would have to be a thorough knowledge of how the above work and an even more thorough knowledge of what MSTS / MSTSBin and all it's support files do.

Will the "Open Source Code" VC#2008 Project Files of ORTS be made available eventually ?

O t t o