General

Re: XNA Procedural Trees

Klasker — Tue, 20 Nov 2007 11:12:29 GMT

Chazsoft:
First off, great tool you have created here, but I have an issue with it, I am runing it on my laptop, the manky old card on it doesnot support 32bit index buffers so I dont get any trees drawn, I have tried setting the buffer type to short but ended up killing the project :(

Any chance you can rebuild it to use shorts for the index buffers??

Sorry, I won't add support for 16-bit indices at the moment - extending the Mesh class to support both would be a serious hassle without making it immutable, and I dislike having a several mesh-classes and immutable meshes aren't my favourite kind either.

Re: XNA Procedural Trees

Charles Humphrey — Fri, 16 Nov 2007 10:54:57 GMT

First off, great tool you have created here, but I have an issue with it, I am runing it on my laptop, the manky old card on it doesnot support 32bit index buffers so I dont get any trees drawn, I have tried setting the buffer type to short but ended up killing the project :(

Any chance you can rebuild it to use shorts for the index buffers??

Re: XNA Procedural Trees

Charles Humphrey — Mon, 12 Nov 2007 16:40:26 GMT

Mail link is now up and you can subscribe to the Dark Omen Games News feed here.

Re: XNA Procedural Trees

DaphydTheBard — Mon, 12 Nov 2007 10:52:07 GMT

The ZMan:
FYI your 'mail' and 'feed' links on http://www.darkomengames.co.uk/ do not work

Thanks, I was aware. One of our other guys (Nemo) is responsible (at the moment) for getting the site up and running - it's early days.

Cheers for the heads up though!

This tree component is truly awsome.

[TODO] Add Klasker to credits screen in OctaneX...

Now if you can give me leaves that move in the breeze..... :-)

Klasker - fancy joining our little team?

Re: XNA Procedural Trees

The ZMan — Mon, 12 Nov 2007 10:37:48 GMT

FYI your 'mail' and 'feed' links on http://www.darkomengames.co.uk/ do not work

Re: XNA Procedural Trees

DaphydTheBard — Mon, 12 Nov 2007 09:13:30 GMT

Klasker, you're my hero :-)

No, seriously! This is awsome, thanks for fixing it so quickly.

If anyone wants to see Klaskers tree's in action in a nearly-complete game, please check this out:

www.darkomengames.co.uk

And download the Beta for OctaneX.

Re: XNA Procedural Trees

Klasker — Sun, 11 Nov 2007 06:00:26 GMT

Any comments on perf? I've not profiled it to see which bits are taking up time yet, but from 1300Hz to 100Hz to draw one tree is a serious hit.

That is a very poor measure of performance you have given me. Drawing nothing at all has a theoretical framerate of infinite frames per second, so comparing one tree to nothing at all does not measure anything. The 100 Hz by themselves does not mean a lot without knowing your video card at least. Try rendering a model of similar polygon count and see how they compare.

I tried profiling it with nVIDIA PerfHUD, but the frame profiler seems to be broken - it tells me that render call #1 takes several million seconds to render. A quick run with nProf didn't bring up anything suspicious. It says that TreeDemo.Draw() only takes 1.48% of the total CPU time, but my CPU usage is only ~4% when running the demo, so that is also a poor measurement.

With that said, I am not an expert on profiling, and neither on billboards. If anything can be optimized, it would definitely be the leaves (since the trunk is just a mesh, nothing special about it).

If the performance is a big let-down, you can use LOD to increase performance. Generating a high-poly and low-poly version of the same tree is easy. To render a very low-poly version, set radialSegments to 5 and cutoffLevel to 1 when calling GenerateTree. The high-poly and low-poly should still look alike (apart from the level of detail) if they were generated using the same seed.

Re: XNA Procedural Trees

Andy Patrick — Sat, 10 Nov 2007 19:50:04 GMT

That sorted it - brilliant, and thanks!

Any comments on perf? I've not profiled it to see which bits are taking up time yet, but from 1300Hz to 100Hz to draw one tree is a serious hit. I'll let you know what I find as I investigate more, of course.

Re: XNA Procedural Trees

Klasker — Sat, 10 Nov 2007 14:21:47 GMT

Version 1.2 is now available which should fix the problem of leaves not scaling.

Uniform scales are now supported, but non-uniform scales are not. The reason is that billboards do not really have a sensible reaction to non-uniform scales if they are to face the camera at all times. I walked the middle-ground by simply using the WorldView matrix's X-scale and assuming that the matrix is scaled unifomly using that scale.

If you are making water reflections of the trees using a negative Y-scale, this might give you problems, but I think you should be ok as long as the absolute Y-scale is equal to the X-scale (e.g. 0.05 and -0.05 should work).

And by the way, thank you all for your comments (and bug reports). I appreciate the feedback.

Re: XNA Procedural Trees

DaphydTheBard — Sat, 10 Nov 2007 13:59:24 GMT

Andy Patrick:

Yes, I'm getting the same problem. I'm having to scale my trees down to 0.02 to fit in my game world (scale of metres) and that works fine for the trunk, but not the leaves - the leaf shader doesn't seem to deal with scale at all.

Apart from that, very effective, very easy to get into the game, very well commented code - top stuff!

Cool, glad it's not just me, I thought maybe I was doing something wrong :-)

Klasker, really can't wait to see an update to this component, it's awsome.

Re: XNA Procedural Trees

Andy Patrick — Sat, 10 Nov 2007 13:37:38 GMT

Yes, I'm getting the same problem. I'm having to scale my trees down to 0.02 to fit in my game world (scale of metres) and that works fine for the trunk, but not the leaves - the leaf shader doesn't seem to deal with scale at all.

(I'm also finding that drawing one tree cuts my frame rate from 1300Hz to 100Hz, but I'll have a check for myself to work out why and how to avoid that).

Apart from that, very effective, very easy to get into the game, very well commented code - top stuff!

Re: XNA Procedural Trees

DaphydTheBard — Fri, 09 Nov 2007 23:03:26 GMT

Klasker

Firstly, this is awsome. I've integrated your trees into OctaneX, the second beta for which we'll be releasing over this weekend. So thanks!

However

I've found a small, but rather significant problem, which I'm not sure how to fix. I need to scale the trees down to about 0.05 of their starting size to make them fit my game.

However, the problem is that the leaves don't scale with the tree trunk, so I end up with a big green ball of BIG leaves in my scene, and I can't see the tree (if I don't render the leaves I can see the scaled tree fine).

Same in reverse, if you scale the tree up, then the leaves look tiny.

Any chance you can take a look at this problem when you get a chance? I think it lies in the particle cloud FX file, as I've had a root through the project code and can't see anything else that might be making this happen (I'm not too ofay with shaders atm, I leave that to Nemo).

Thanks. And I'll obviously be sticking your name in the credits for OctaneX from now on!

Re: XNA Procedural Trees

Klasker — Wed, 07 Nov 2007 15:04:30 GMT

An updated version is available, which fixes the issue with systems using comma as decimal separator.

Re: XNA Procedural Trees

Takuan Daikon — Wed, 07 Nov 2007 00:04:31 GMT

I'm surprised that Shawn Hargreaves didn't mention this, as I found it a fascinating read recently, and surely can provide some interesting thoughts about the problem mentioned by steffanp : Bitfield renderstates

Re: XNA Procedural Trees

Klasker — Tue, 06 Nov 2007 14:13:00 GMT

@Caetris:
All you need is the random seed. If you generate a tree using profile X and the seed 12345, you will always get that same tree with profile X and seed 12345, on any computer at any time. The only possible exception to this is if you move a new version of XNATrees.