Project D3D11Client Development

[asmi]

To get all features of D3D11 (terrain, atmosphere, etc.), i have to download only the latest file (in the ltest link) or I have to get some basic files?

The version with terrain and all other new features is not yet stable enough to be released to the public, so it's in closed alpha testing stage now. Once this stage is over, it will be released for public beta testing.

 

[Zachstar]

The terrain is getting quite a bit of interest it seems. I hope Martins can implement his terrain engine idea so I can be convinced to get a DX11 card and finally retire this AMD 4770

What do you think asmi? Could full working terrain and effects draw the development scene back to Orbiter in a big way? Maybe the impossible of an Project Apollo public version? :lol:

 

[asmi]

The terrain is getting quite a bit of interest it seems. I hope Martins can implement his terrain engine idea so I can be convinced to get a DX11 card and finally retire this AMD 4770

What do you think asmi? Could full working terrain and effects draw the development scene back to Orbiter in a big way? Maybe the impossible of an Project Apollo public version? :lol:

You should've gotten modern video card this a long time ago I think that unless someone bothers to develop a terrain engine, collision detection and so on and then gives it to Martin so he could integrate that, we'll be waiting long for the terrain support to be introduced - because obviously Martin has his own plans for development, and as far as I understand terrain is not included. And that's why once I'll get current iteration of D3D11Client stable enough (or maybe even earlier), I'm going to implement that engine so Martin would be able to simply integrate it into Orbiter.

 

[Zachstar]

You mean by yourself. The entire terrain engine?

 

[asmi]

You mean by yourself. The entire terrain engine?

You're saying that as if it's something big and complex It's actually pretty simple: I've already got heightmaps, and the code to load/process them (it's part of D3D11Client now), so the only thing is left is collision detection. That should not take a lot of time.

 

[Zachstar]

You're saying that as if it's something big and complex It's actually pretty simple: I've already got heightmaps, and the code to load/process them (it's part of D3D11Client now), so the only thing is left is collision detection. That should not take a lot of time.

A programmer saying anything "Next Tuesday" What has the world come to!

Well I look forward it. Real, Standardized terrain in Orbiter in my opinion is just the shot in the arm needed to really get the development scene roaring again.

I guess a difficult part is communicating to MFDs that currently rely on 0 alt for the endpoint. Yet that is where standardization helps.

 

[asmi]

A programmer saying anything "Next Tuesday" What has the world come to!

I didn't say "Next Tuesday" I'm long enough in software development industry to know that developers tend to underestimate timelines That's why I didn't tell exact time, but in this case the most complicated part is already implemented - I just need to take it out of client and make it a separate library. As for collision detection - there are tons of different algorithms out there so that shouldn't be hard either. The reality is that it will take less time for me to implement the engine than for Martin, since I've got a good piece of it already implemented and working fine.

The biggest problem here is implementing efficient paging-in and paging-out algorithm for heightmaps, since even heightmap for Earth alone won't fit within 2Gb address space, and there are other planets as well, not to mention other resources also require address space... If only Martin would build 64bit version of ng client...

Well I look forward it. Real, Standardized terrain in Orbiter in my opinion is just the shot in the arm needed to really get the development scene roaring again.

It's going to be a big pain in butt mostly for vessel devs since they will have to define and maintain convex collision hull instead of 3 TDPs as they are now, not to mention updated avionics to take terrain curvature into account.

I guess a difficult part is communicating to MFDs that currently rely on 0 alt for the endpoint. Yet that is where standardization helps.

Well majority of MFDs are all around actual spaceflight, and it won't be changed. For example Orbit MFD doesn't care if there is a terrain or not. Some other MFDs will have to be updated (adding "radio-altimeter" to Surface MFD and HUD, updating VTOL/ILS MFDs), yet some will have to be essentially re-implemented (AeroBrake MFD comes into mind)... But there is nothing unrealistic - just some things that have to be done...

 

[Zachstar]

Sorry about that. I meant to say anything except. As in I was surprised by any kind of time reference except "When it's done"

I was too distracted on that post sorry!

Edit: Ouch on the Aerobrake MFD. I remember testing it back in the day and how much it took to get it working the way it is today. I recently contacted GP about an error I had with the MFD and Fracisdrake's addons so he might need to be informed about that.

 

[Marvin42]

Weather with DX11

Could Orbiter use something like this? A bet that beeing a free, respected engine, it could get a free license so it could save some time. Seems to work with DX11 from their description.
http://www.simul.co.uk/truesky

 

[Cras]

Just got my first chance to look at D3D11 with the new post processing and terrain features.

Earth looks whitewashed. I think it has to do with the terrain somehow, as you see KSC is just fine, but the rest of the planet is just too white. The terrain looks great still.

http://dl.dropbox.com/u/68333863/TerrainTest/000.png

Saw this strange blue square of the coast of Florida as well.


http://dl.dropbox.com/u/68333863/TerrainTest/001.png

Couldnt do much on this flight because my Map MFD crashed. The screen just went blank and I could not get it back, but I am thinking this was an Orbiter crash, because when I tried to switch view to the lower 2D panel, Orbiter CTD on me.


But I did get a look at this:

Which I have to say is just awesome.

 

[dumbo2007]

Wow!! that surface looks smooth and made of marble !

Hope to see the DG rolling on it soon

 

[Columbia42]

I just tried the D3D11 client but when I launched Orbiter, it hung up on loading taxiway.dds and concrete.dds. Is this the kind of thing that would happen if my GPU is not DX10/11 capable? Because I was running it on my laptop (only computer with Win7) which just has an integrated Intel chip so I have no idea if it is capable or not.

 

[Cras]

Here is pics of KSC and the Moon with the post-processing engine turned off. There seems to be no difference to me. The Earth is clearly still whitewashed and the strange blue square still sits off the coast of Florida.

http://dl.dropbox.com/u/68333863/TerrainTest/004.png

http://dl.dropbox.com/u/68333863/TerrainTest/005.png

 

[asmi]

I just tried the D3D11 client but when I launched Orbiter, it hung up on loading taxiway.dds and concrete.dds. Is this the kind of thing that would happen if my GPU is not DX10/11 capable? Because I was running it on my laptop (only computer with Win7) which just has an integrated Intel chip so I have no idea if it is capable or not.

This is likely due to the fact that you've taken wrong version of the client. Post here contents of Orbiter.log after that "hang"

Here is pics of KSC and the Moon with the post-processing engine turned off. There seems to be no difference to me. The Earth is clearly still whitewashed and the strange blue square still sits off the coast of Florida.

http://dl.dropbox.com/u/68333863/TerrainTest/004.png

http://dl.dropbox.com/u/68333863/TerrainTest/005.png

Yea, I've already uploaded a fix for "popped-out" textures, and working on terrain issue now.

 

[Columbia42]

This is likely due to the fact that you've taken wrong version of the client. Post here contents of Orbiter.log after that "hang"

I can't upload an Orbiter log because Orbiter freezes and I have to force close it, meaning no log is written. The version I downloaded from bitbucket is 8f50441cb4f6.

 

[DaveS]

This is likely due to the fact that you've taken wrong version of the client.

What is the tag of the the latest stable version for the Orbiter beta?

 

[asmi]

What is the tag of the the latest stable version for the Orbiter beta?

The latest version lacks binary, I will build and upload it tonight.

As for integrated video cards, I'm afraid current features and effects bring client beyond their performance envelope, so I might be forced to drop that support... Will see what I can do, but if this support will preclude some important features, I will drop that. As I've mentioned before, D3D11Client is supposed to be at cutting edge, and inegrated videos aren't quite there... At the end of the day, you can't play Crysis on integrated video, can you?

Of course that drop won't mean, that I will take measures to specifically prevent the client to be run on that HW. It just will be sloooow...

 

[Zachstar]

In my opinion Integrated is fully in the DX7 and DX9 client camp. The only thing that is built into Mobo or CPU that can even touch DX11 these days is the APUs from AMD.

In my opinion just say it flat out. No APU, no good DX11 card = Try another client. Because you know people will ask why their Intel chip from 3 years ago cant handle heavy DX11 Terrain

Edit: And in my opinion that is the beauty of the client system. DX11 = King of terrain and heavy shading effects. DX9 = King of midrange DX7 = King of Integrated Graphics Software= King of Pentium 2 NG= King of... Everything else.

 

[asmi]

In my opinion Integrated is fully in the DX7 and DX9 client camp. The only thing that is built into Mobo or CPU that can even touch DX11 these days is the APUs from AMD.

In my opinion just say it flat out. No APU, no good DX11 card = Try another client. Because you know people will ask why their Intel chip from 3 years ago cant handle heavy DX11 Terrain

Edit: And in my opinion that is the beauty of the client system. DX11 = King of terrain and heavy shading effects. DX9 = King of midrange DX7 = King of Integrated Graphics Software= King of Pentium 2 NG= King of... Everything else.

Yes I completely agree. I made it absolutely clear in the very beginning that the client is targeted at the high-end systems, which means that sooner or later the moment would've came to let slower systems go. Well, looks like this moment is coming, which kinda makes me sad, but on the other hand I'm eager to add more and more new beautiful and exciting features and effects to make Orbiter's experience a truly immerse one, adding modern visuals to already awesome simulation.

So my apologizes to you guys with slower systems, but it's time for this client to move on...

 

[Zachstar]

Well that is the thing. The guys with the less capable systems have a choice now. And as you are writing the standardized terrain system it will mean those with other clients can still get things like terrain. It just wont be as amazing as DX11.

In the DX9 Topic the very first reply was an off topic by me saying that DX10 should be the goal. As back then the only OVP client I thought would see the light of day was Artlav's OpenGL version. I could not imagine back then that someone would not only take on DX11 but on top of that write a standardized terrain system for Orbiter itself! The choice folks are going to have soon enough will be staggering and I believe that those with integrated graphics will have plenty of the Orbiter experience.