Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[cameni]

Sorry I shortened the term so it may have led you to a wrong conclusion - correctly it's occlusion culling, and it's an optimization that removes stuff that is not visible. So in mountains where the distant terrain is occluded, it would bring a serious performance boost.

For example here this view of Lukla:


But here's what is currently being sent to the card:




Note that it isn't that bad as it looks, the scene is drawn from front to back so there's almost no overdraw, nevertheless there's additional processing as the engine has to compute terrain tiles even for stuff that is not visible, consuming more memory as it should etc.

[razgriz 53992]

Sorry I shortened the term so it may have led you to a wrong conclusion - correctly it's occlusion culling, and it's an optimization that removes stuff that is not visible. So in mountains where the distant terrain is occluded, it would bring a serious performance boost.

For example here this view of Lukla:
SNIPPED PHOTOS

But here's what is currently being sent to the card:


Note that it isn't that bad as it looks, the scene is drawn from front to back so there's almost no overdraw, nevertheless there's additional processing as the engine has to compute terrain tiles even for stuff that is not visible, consuming more memory as it should etc.

Very nice shots btw (well the first one)

So are you saying, without occlusion culling the card is rendering massive areas of land, even when most of it is not visible? And that occlusion culling gets rid of this, increasing performance? I might be wrong, but that sounds like it would increase FPS a fair bit...

[cameni]

So are you saying, without occlusion culling the card is rendering massive areas of land, even when most of it is not visible? And that occlusion culling gets rid of this, increasing performance? I might be wrong, but that sounds like it would increase FPS a fair bit...

Well it's not actually rendering the hidden areas of land, because the pixels get rejected earlier, but it's processing the meshes and generating them, which may be costly mainly during a movement. There will be raw FPS increase for sure, just not that huge as it may look from the screenshots. There's also very effective LOD working, so what may seem as large stretches of land are just several levels of terrain.
But yes, it should help a lot.

[razgriz 53992]

Cool stuff.

So if I could get 30fps BEFORE occlusion, that's got to be good. The only issue may arise if add ons create densely populated areas such as cities, but even then it'll probably be good considering the effective LOD.

Sorry to ask so many questions, but when the LOD is developed further and we start having more detail close up like 3D trees and stuff, will the frame rate be affected much? I would doubt it, just clarifying

Merry Christmas btw, you really should take a break!

[cameni]

It depends on the amount of detail added, surely one can add so much that it would kill performance on any card
But of course we want to continue using the same techniques that are giving us the performance on existing parts of the engine.

Merry Christmas to you all too.

[ZeosPantera]

It depends on the amount of detail added, surely one can add so much that it would kill performance on any card
But of course we want to continue using the same techniques that are giving us the performance on existing parts of the engine.

Merry Christmas to you all too.

It is Xmas and Santa Claws is coming for us all.

Now onto business. The first picture shows the side of a mountain obstructing everything in the distance but In the second picture from above, all the land we can't see behind the mountain is still shown. Is that this..

it's not actually rendering the hidden areas of land, because the pixels get rejected earlier, but it's processing the meshes and generating them

I presume that is what I am referring too. Will that also be culled to nothingness? Or would the intense disappearance / reappearance of that much land screw things up smoothness-wise?

[cameni]

Yes the land that is shown with different camera will be culled out by occlusion culling algorithm. Sudden reappearance should not affect the smoothness too much, the occlusion culling mainly deals with not sending invisible chunks to renderer at all which enhances the performance.

However, it's not entirely problem free if we also want it to ensure that terrain from certain detail level won't be generated and it won't consume GPU memory uselessly. A sudden reappearance could lay a sudden burden on the generator part that may affect the smoothness if not handled properly.

[Marek]

When Demo will be out?
Som fanatik do Tatier   a myslim tím Autá   ale aj hory su fajn

[ZeosPantera]

When Demo will be out?
Som fanatik do Tatier   a myslim tím Autá   ale aj hory su fajn

When ATi fix the issues with OpenGL.

[pico]

...that could be in Years.  

[pico]

Do u plan Engineupdates for the Demo? If u do, u can bring the ATI Support whit an later Update, when its done.  

Just an Idea.  

[cameni]

IMO tactically it would not be a good idea to release the demo without ATI support. I still believe the issues will be resolved soon .. for some definition of the word soon
We are also reconsidering how to position the demo with regards to our recent game plans. Should be something that lures players towards it.

[razgriz 53992]

We are also reconsidering how to position the demo with regards to our recent game plans. Should be something that lures players towards it.

That would be difficult I imagine due to the fact none of the game plans are integrated in any way. I'm sure anyone here who knows about it already would be fine with it just how it is, with no plot, just the freeroaming world, but that does make it difficult to appeal to lots of people, specifically the market(s) you aim to target.

[cameni]

Well, yes. I mean, we have to think how to execute this. One thing is a technology demo, and another one is the game and a game demo. But by the time all issues are fixed the demos could as well converge into one. Or not, given the different interests there are for the technology.

Btw, what are your expectations from the demo, what do you want it to bring to you? It may help us to separate it so that the expectations are addressed better.

[ddenn]

For me the demo with all features that was already shown in videos is OK. But I'm just a civil flightsim enthusiast, not a possible investor or big developer, though I'd like to make at least some vehicle addons for Outerra, like planes, boats and cars.

I'd like to see in the demo mostly a possibilities of the engine for addon makers, I always thought that procedural technology is not very addon-makers friendly (it may have a lot of restrictions I guess). Hope that's not the case with Outerra