KriScg

I'm running into this very issue at the moment...

2012-04-09T18:51:17.741+02:00

I'm running into this very issue at the moment, and cannot even switch to 64-bit because it's an XNA game and the XNA assemblies are 32-bit only. Switching to another framework like SlimDX or SharpDX would be a painful. Makes me wonder why the XNA devs didn't release 64-bit assemblies. I wonder if it's marketing, or some technical limitations within the XNA framework... Oh well, I guess I will never know.

Re (2): You don't need to do that. Linked list...

2011-02-13T21:42:58.709+01:00

Re (2): You don't need to do that. Linked list in a freelist is only used to store unused elements. To iterate over all used freelist elements you may use an additional usage array (e.g. 1-bit per element: 0-unused 1-used). So, the good thing is that iteration is still cache friendly. But the main disadvantage here is you have to visit ~all elements to check their usage bit even when there's very few of them actually used - yep, it's just as bad as pessimistic case of iterating over all elements. You may use some tricks to speed it up by checking multiple 32-bits at once etc.

Anyway, that's what I meant by longer iteration times. I don't mean it's perfect, but it might be a good choice in certain scenarios.

(1) Yes, gradual defragmention looks like a good i...

2011-02-13T01:56:11.171+01:00

(1) Yes, gradual defragmention looks like a good idea. Worth checking for next iteration of our (FWH) engine.

(2) Linked list for particles sounds hardcore :). I don't think it's a good idea to iterate through linked list every frame.

Hi there, few more ideas: (1) Instead of defragme...

2011-02-12T15:50:24.543+01:00

Hi there, few more ideas:

(1) Instead of defragmenting whole array at once, do it gradually over multiple frames; maintain current "defragmentation pointer" and move it forward every frame by specific number; when reached the end, start again from the beginning.

(2) Don't defragment at all and instead use freelist (http://en.wikipedia.org/wiki/Free_list) to allocate and free slots for particles; particles that are already being displayed won't change their ordering, so it's just the new ones that might suddenly pop on top other ones - but in real-life this often won't matter; the downside is that the iteration takes longer now.

Whenever I want to install the SDK on Fedora 14 I ...

2011-01-15T23:26:19.569+01:00

Whenever I want to install the SDK on Fedora 14 I get this:

warning: cell-install-3.1.0-0.0.noarch.rpm: Header V3 DSA/SHA1 Signature, key ID 9ac02885: NOKEY
error: Failed dependencies:
libc.so.6 is needed by cell-install-3.1.0-0.0.noarch
libc.so.6(GLIBC_2.0) is needed by cell-install-3.1.0-0.0.noarch
libc.so.6(GLIBC_2.1) is needed by cell-install-3.1.0-0.0.noarch
libc.so.6(GLIBC_2.1.3) is needed by cell-install-3.1.0-0.0.noarch
libc.so.6(GLIBC_2.3) is needed by cell-install-3.1.0-0.0.noarch
libc.so.6(GLIBC_2.4) is needed by cell-install-3.1.0-0.0.noarch
libgcc_s.so.1 is needed by cell-install-3.1.0-0.0.noarch
libgcc_s.so.1(GCC_3.0) is needed by cell-install-3.1.0-0.0.noarch
libgcc_s.so.1(GLIBC_2.0) is needed by cell-install-3.1.0-0.0.noarch
libm.so.6 is needed by cell-install-3.1.0-0.0.noarch
libm.so.6(GLIBC_2.0) is needed by cell-install-3.1.0-0.0.noarch
libpthread.so.0 is needed by cell-install-3.1.0-0.0.noarch
libpthread.so.0(GLIBC_2.0) is needed by cell-install-3.1.0-0.0.noarch
libstdc++.so.6 is needed by cell-install-3.1.0-0.0.noarch
libstdc++.so.6(CXXABI_1.3) is needed by cell-install-3.1.0-0.0.noarch
libstdc++.so.6(GLIBCXX_3.4) is needed by cell-install-3.1.0-0.0.noarch

I guess I need to install an older version of gcc?

And also there http://connect.microsoft.com/Visua...

2010-11-13T22:20:40.698+01:00

And also there

http://connect.microsoft.com/VisualStudio/feedback/details/100686/empty-member-optimization-not-working-properly

See there http://connect.microsoft.com/VisualStudi...

2010-11-13T22:19:24.906+01:00

See there http://connect.microsoft.com/VisualStudio/feedback/details/101525/multiple-inheritance-wrong-sizeof

msvc++ fails to apply EBO in case of multiple inheritance

I see. By clipping I mean triangle cutting, not tr...

2010-10-02T20:50:48.840+02:00

I see. By clipping I mean triangle cutting, not triangle testing.

Yes, It would requite outer edge detection. It's something I'm trying to solve now. Slightly smaller occluders won't work for us, as we have death penalty for introducing stuff like occlusion false positives (or templates :)).

Clipping also consists of the first part where you...

2010-10-02T20:24:58.166+02:00

Clipping also consists of the first part where you check if the triangle has to be clipped - that has the same perf demands as the transform. The actual clipping can be slower, yes.

The rasterization clipping introduces per-scanline overhead as opposed to per-triangle, but it's quite tiny.

Apologies for the confusion, I use post projection depth for the depth buffer.

Occlusion false positives are almost impossible to avoid, as far as I understand (you'll have to teach the rasterizer to distinguish inner edges vs outer edges), but it was not a problem for us - since we don't (obviously) render the actual level geometry in occlusion buffer, the occlusion geometry can be made slightly smaller than the real volume.

1. transform & clip. I separate them because v...

2010-10-02T20:16:46.799+02:00

1. transform & clip. I separate them because very small percentage of triangles needs to be clipped.

2. clipping during rast and triangle setup. IMHO on PC it could be faster to clip with 5 planes. Need to check performance of this one.

4. z in view space. Does it mean that You store view space depth in software depth-buffor? Also why view space and not screen space?

BTW did You solve problem with false positives? I mean the one because of rendering in lower resolution. Last time when I through about it I understood that it's not enough to shift pixel center when interpolating depth. There can be for example 1 pixel size occluders which will occlude bigger pixel block. It's the main issue which prevents us from replacing beamtrees.

Here is my experience (we're using this at wor...

2010-10-02T19:58:20.332+02:00

Here is my experience (we're using this at work on SPUs):

- transform & clip are more or less the same wrt optimization demands - not sure why you separate them
- you can only clip against the near plane, the rest is done easier in raster stage
- backface culling is essential
- scanlines were ~2x faster than halfspace on SPUs for me
- we test using aabb screen space rect against the min aabb z (view-space)
- 256x144, no HOM - did not bother as object tests were fast enough

Very similar method was described in that NVIDIA p...

2010-09-08T13:54:43.471+02:00

Very similar method was described in that NVIDIA presentation (http://developer.nvidia.com/object/nvision08-DemoTeam.html).

You could modify depth in early-Z pass (using some...

2010-09-08T12:02:36.172+02:00

You could modify depth in early-Z pass (using some parallax method like POM or CSM) but I guess it won't be "early-Z" anymore :)

@Dab that's an interesting idea. Now I remembe...

2010-09-08T11:26:04.465+02:00

@Dab that's an interesting idea. Now I remember that NVIDIA presentation about normal reconstruction from depth. The question is how to include normal maps into this process.

It just occurred to me: what if you would use norm...

2010-09-08T11:09:07.722+02:00

It just occurred to me: what if you would use normals in lighting pass? You can reconstruct normals from depth (so no need to render anything but early-Z first) and it looks quite well: http://dabroz.scythe.pl/2010/05/02/ssao-revisited

@Dab why did You have to remind me about it at mid...

2010-09-08T00:42:30.319+02:00

@Dab why did You have to remind me about it at midnight. Instead of going to bed after good evening with software rasterization programming I had to prototype it :). It does look quite strange at intersections of lights with different color and direction. So it doesn't look too practical, but I think that decoupling lighting from normals and aggregating lights is a good idea. Maybe instead of directional lights I should use virtual point lights or smth...

Did you implement it? Have any screenshots? Have a...

2010-09-07T23:45:22.479+02:00

Did you implement it? Have any screenshots? Have any screenshots with "bad" cases? :) I just wonder about practical application.

Thanks for fixing this quickly, should be correct ...

2010-08-20T22:43:13.283+02:00

Thanks for fixing this quickly, should be correct now.

Another thing to be aware of when using this technique is that you also push forward light sources that are completely behind you. This should not be an issue though if you do proper frustum culling for lights.

Thanks, I fixed example code. Now min/max is in ca...

2010-08-20T21:05:25.580+02:00

Thanks, I fixed example code. Now min/max is in camera (view) space (previously it was erroneously done in screen space).

Thanks for the interesting bit. But why are you ac...

2010-08-20T20:46:32.276+02:00

Thanks for the interesting bit. But why are you actually using the minimum? The coordinates are in clip space (0 >= z <= w) after the matrix multiplication, right? Or are you using GL conventions?

Yes, You are right that most of the quality degrad...

2010-08-18T21:06:25.101+02:00

Yes, You are right that most of the quality degradation comes from range decrease. I didn't think about it when I wrote this post. Scaling could work, but it would require a change of texture pipeline (16 bit source textures, some tool or smart artists).

It seems to me, the main cause of quality degradat...

2010-08-18T19:18:57.809+02:00

It seems to me, the main cause of quality degradation in your sample is color range decrease. You should scale premultiplied image to full range in order to have better compression.

This sounds like a very interesting idea! I can...

2010-08-15T12:22:10.228+02:00

This sounds like a very interesting idea! I can't wait to see if it works, because I'm not so sure about it :)

just wanted to note for others which may be trying...

2010-07-31T07:29:09.198+02:00

just wanted to note for others which may be trying to do this:
I also had to install 32bit packages of libXtst and cairo in order for eclipse to work.

Hi, it looks like You need to install sysroot_imag...

2010-07-30T09:27:08.775+02:00

Hi, it looks like You need to install sysroot_image ( rpm -ivh sysroot_image-3.1-1.noarch.rpm ).