Post Your Irrlicht Screenshots / Render Here.
Dual paraboloid shadow mapping handling transparent caster & receiver as you can see on the grass. VSM & PCF supported but not enabled on this screenshoot. Also, it's only the shadow render target
Next effect on my list is SSAO but I need to finish my antialiasing with edge detection shader. Then I hope I'll find some time to implement god rays effect if not too expansive (And global illumination with spherical harmonics, Idk exactly how it's working however a lot of people here are playing with it and I'm pretty curious :p)
Next effect on my list is SSAO but I need to finish my antialiasing with edge detection shader. Then I hope I'll find some time to implement god rays effect if not too expansive (And global illumination with spherical harmonics, Idk exactly how it's working however a lot of people here are playing with it and I'm pretty curious :p)
-
- Posts: 1215
- Joined: Tue Jan 09, 2007 7:03 pm
- Location: Leuven, Belgium
Crepuscular rays (or god rays) aren't that demanding, nor hard to implement if you have a decent post-processing framework set upstefbuet wrote:Dual paraboloid shadow mapping handling transparent caster & receiver as you can see on the grass. VSM & PCF supported but not enabled on this screenshoot. Also, it's only the shadow render target
Next effect on my list is SSAO but I need to finish my antialiasing with edge detection shader. Then I hope I'll find some time to implement god rays effect if not too expansive (And global illumination with spherical harmonics, Idk exactly how it's working however a lot of people here are playing with it and I'm pretty curious :p)
Nvidia has a really nice article about it in one of their GPU gems books: http://http.developer.nvidia.com/GPUGem ... _ch13.html
Mmm... that implementation of the godrays has some tricks on it... you really need a special set up for the scene to make it useful.
It relies on the fact that the background has to be much brighter than the foreground, so it actually samples several times the screen in the direction of the light position, and makes a radial blur. All in all, that postproduction effect is really useful when the sampled image is small or is blurry
Surely you've read this, but this article explains many aspects of the VSM so it can overcome its weaknesses
http://http.developer.nvidia.com/GPUGem ... _ch08.html
(oh my! GPUgems is a real treasure! )
It relies on the fact that the background has to be much brighter than the foreground, so it actually samples several times the screen in the direction of the light position, and makes a radial blur. All in all, that postproduction effect is really useful when the sampled image is small or is blurry
VSM + blurred shadowmap is better than any PCF you can use. With VSM you don't really need PCF because VSM gives a perfect match for the PCF with just one sample of the shadowmap (at the cost of posible light bleeds, which, anyway, can be minimized, or completely corrected using simple operations) and you can still do effects on the shadowmap. Blurring the shadowmap with VSM softens a lot, and quite well, the results of the shadowmappingVSM & PCF supported
Surely you've read this, but this article explains many aspects of the VSM so it can overcome its weaknesses
http://http.developer.nvidia.com/GPUGem ... _ch08.html
(oh my! GPUgems is a real treasure! )
"There is nothing truly useless, it always serves as a bad example". Arthur A. Schmitt
Yes I've read this article many times
In fact I'm using VSM with blurred shadow maps, and I said PCF because I'm also using a 5x5 PCF when sampling the blurred shadow map. This idea came from the XEffect framework, however Idk if I'll keep this or try something else to speed it up. However dual paraboloid is very fast compared to the basic 6 x render
In fact I'm using VSM with blurred shadow maps, and I said PCF because I'm also using a 5x5 PCF when sampling the blurred shadow map. This idea came from the XEffect framework, however Idk if I'll keep this or try something else to speed it up. However dual paraboloid is very fast compared to the basic 6 x render
-
- Posts: 175
- Joined: Wed Dec 20, 2006 12:04 pm
eden update
Hey Everyone,
Been digging away at irrlicht in wxWidgets, managed to get the UI below going. It's irrlicht working in a wxAUI pane with their nice new ribbon thrown in for kicks...
Been digging away at irrlicht in wxWidgets, managed to get the UI below going. It's irrlicht working in a wxAUI pane with their nice new ribbon thrown in for kicks...
Unreal Engine style backgrounds
The planet, the star and that enormous thing are a background mesh. The good thing is that they are still simple animated nodes, so anything you can do with an ISceneNode can be done to them
The point is to overcome the limits of a skysphere/skycube, the textures don't have to be large anymore to get a good background.
The planet, the star and that enormous thing are a background mesh. The good thing is that they are still simple animated nodes, so anything you can do with an ISceneNode can be done to them
The point is to overcome the limits of a skysphere/skycube, the textures don't have to be large anymore to get a good background.
"There is nothing truly useless, it always serves as a bad example". Arthur A. Schmitt
-
- Posts: 1215
- Joined: Tue Jan 09, 2007 7:03 pm
- Location: Leuven, Belgium
I didn't set it off, but it looks quite well
This is the background only.
http://i52.tinypic.com/ft6s9.png
I leave the link only because the image is quite large and may take some time to load.
This is the background only.
http://i52.tinypic.com/ft6s9.png
I leave the link only because the image is quite large and may take some time to load.
"There is nothing truly useless, it always serves as a bad example". Arthur A. Schmitt
-
- Competition winner
- Posts: 1123
- Joined: Sun Jun 10, 2007 11:14 pm
They look well. But these games had a lot of 2D animation behind. In fact, when the "puyos" got together, their shapes somehow blended. It is a sort of puzzle, it isn't really complex, you need all the posible shapes and put them together. Joining them with beams is too simple. But it is nice. Maybe if the beams were a bit thicker, they would be more noticeable.
update... shader driven skin! almost 60.000 skinned vertices with normal mapping, 2 dynamic lights, spherical harmonics environment lighting, and all. I have pushed irrlicht over the limit, and i say well. There is not enough room for all the data this kind of animation requires. Irrlicht vertices aren't enough for this as they are currently. I think that, at least, another type of vertex which, besides tangent and binormal, stored 4 weights and their corresponding indices would be welcome, if things have to go on without FVFs, that is. Though, i guess that Irrlicht 2 will include the later, so, no worries... The good part... The bottleneck is still the pixel rendering, at 640x480 the speed rises over 65 fps.
The next target?... break the limit of just 59 bones (the DX shader model 3 only allows 1024 floats as constants, which, including the other data, leaves room just for 59 or so bone matrices)
BTW... the skinning i did to the character is LAME, it is suposed to look like that with the ankle broken and all With a well done skinning, there are no troubles.
update... shader driven skin! almost 60.000 skinned vertices with normal mapping, 2 dynamic lights, spherical harmonics environment lighting, and all. I have pushed irrlicht over the limit, and i say well. There is not enough room for all the data this kind of animation requires. Irrlicht vertices aren't enough for this as they are currently. I think that, at least, another type of vertex which, besides tangent and binormal, stored 4 weights and their corresponding indices would be welcome, if things have to go on without FVFs, that is. Though, i guess that Irrlicht 2 will include the later, so, no worries... The good part... The bottleneck is still the pixel rendering, at 640x480 the speed rises over 65 fps.
The next target?... break the limit of just 59 bones (the DX shader model 3 only allows 1024 floats as constants, which, including the other data, leaves room just for 59 or so bone matrices)
BTW... the skinning i did to the character is LAME, it is suposed to look like that with the ankle broken and all With a well done skinning, there are no troubles.
"There is nothing truly useless, it always serves as a bad example". Arthur A. Schmitt