What are you working on?

[Shpuld]

Fair point frag.machine, most of the graphics are still very much WIP, so I'll probably have the enemy fire have more interesting colors to make it stand out.

[Barnes]

update sslr
http://www.youtube.com/watch?v=lKoE2-iwGIQ

[mankrip]

I've finished the procedural dithered particle code, and I'm damn pleased with the result. I may post some screenshots tomorrow; it seems that Internet access is being cut in my job during the weekends…

The particle metadata takes a lot of RAM. About 100 MB in full HD (1080 particle sizes), 32 MB in 800*600, and 5 MB in 320*240. The number of sizes is given by the vrect's min(width, height).

But they're blazing fast to draw, even the translucent ones. Rendering hundreds and maybe thousands of particles on screen isn't a problem.

Now I'm working on some BSP drawing code…

[qbism]

update sslr
http://www.youtube.com/watch?v=lKoE2-iwGIQ

Looks like it is reflecting geometry in view and cleverly fades-out the rest? Fading due to the shader? In any case not much FPS impact.

:D I've finished the procedural dithered particle code, and I'm damn pleased with the result. I may post some screenshots tomorrow; it seems that Internet access is being cut in my job during the weekends…

The particle metadata takes a lot of RAM. About 100 MB in full HD (1080 particle sizes), 32 MB in 800*600, and 5 MB in 320*240. The number of sizes is given by the vrect's min(width, height).

If system is new enough to run HD, then 100MB isn't much RAM

[mankrip]

update sslr
http://www.youtube.com/watch?v=lKoE2-iwGIQ

Looks like it is reflecting geometry in view and cleverly fades-out the rest? Fading due to the shader? In any case not much FPS impact.

Indeed. It seems to reuse the scene after the polygons have been clipped, so there's no need for clipping again. Maybe all it does is telling the GPU to redraw the on-screen polygons using a different projection matrix, which should be very efficient. And blurring the original screen limits of the reflection.

[Barnes]

update sslr
http://www.youtube.com/watch?v=lKoE2-iwGIQ

Looks like it is reflecting geometry in view and cleverly fades-out the rest? Fading due to the shader? In any case not much FPS impact.

Indeed. It seems to reuse the scene after the polygons have been clipped, so there's no need for clipping again. Maybe all it does is telling the GPU to redraw the on-screen polygons using a different projection matrix, which should be very efficient. And blurring the original screen limits of the reflection.

SSLR is a completely screen space technique, the only things it needs are screen depth, normal & color buffers. Yes, it fades out the missing info, but it can be made to fade into Q3-style env map or cubemap probes.

[Barnes]

Radiosity normal mapping (not deluxe mapping) & specular approximation + SSAO
Vanilla:


Ground Zero:


The Reckoning:


Zaero:


There is a huge need in texture re-mastering, especially bump maps. So if anyone wants to help, welcome

[mankrip]

SSLR is a completely screen space technique, the only things it needs are screen depth, normal & color buffers. Yes, it fades out the missing info, but it can be made to fade into Q3-style env map or cubemap probes.

A normal map/buffer. I didn't think of that. That's really a genius use for normalmapping.

[Barnes]

Our implementation needs view-space normal, which can be additionally distorted by a UV offset or normal map.

[Spiney]

One thing I think might be cool for Q2 is to add masks to the textures that define the 'metalness' of a pixel.
So then you could change some parameters of your reflection model for some cheapish physical based rendering.
So no more diffuse reflection for the metals and you could use the existing color texture as the specular color -- well, I SUPPOSE
Non-metals just use the color of the light as specular color (though ideally you need to gamma correct or it'll look conveniently metallic anyway).

I noticed Quetoo doesn't actually do diffuse shading with it's normal maps (only specular) which I guess is why the metals look kind of nice.
The downside is that you can't do things like brick walls and such very well, but Q2 is mostly metal so it's a natural fit.

The radiosity normal maps are nice, do you get them out of a compiler or render them in engine somehow?
Also did you have phong specular on the surfaces before? I guess you can't do that with 'vanilla' RNM unless you make a deluxel variant.
But SSLR magic trickery is probably better

[krigs]

One thing I think might be cool for Q2 is to add masks to the textures that define the 'metalness' of a pixel.
So then you could change some parameters of your reflection model for some cheapish physical based rendering.
So no more diffuse reflection for the metals and you could use the existing color texture as the specular color -- well, I SUPPOSE
Non-metals just use the color of the light as specular color (though ideally you need to gamma correct or it'll look conveniently metallic anyway).

I noticed Quetoo doesn't actually do diffuse shading with it's normal maps (only specular) which I guess is why the metals look kind of nice.
The downside is that you can't do things like brick walls and such very well, but Q2 is mostly metal so it's a natural fit.

Shader takes 1 min to change, what is really funny is that there are no metalness/roughness textures and we have no time to draw them for 1488 textures of Quake2 (estimate without non-partial duplicates). Know anyone with such a wealth to do at least some?

The radiosity normal maps are nice, do you get them out of a compiler or render them in engine somehow?
Also did you have phong specular on the surfaces before? I guess you can't do that with 'vanilla' RNM unless you make a deluxel variant.

A modification to qrad3 allowed to use Blinn half-angle approximate specular. Not reflection probes, but better than no specular.
Not sure what you mean by 'vanilla RNM', Q2 had only flat lightmaps.

But SSLR magic trickery is probably better

Currently, SSLR is only on the water surfaces. The visual impact from RNM is much larger, as finally the lighting code is making use of that little nice amount of re-texture present.

[Jay Dolan]

I noticed Quetoo doesn't actually do diffuse shading with it's normal maps (only specular) which I guess is why the metals look kind of nice.
The downside is that you can't do things like brick walls and such very well, but Q2 is mostly metal so it's a natural fit.

[jim]

Thought of some double weapon system for TigerCake. Shields & Tools on left arm, Weapons on right arm.

The shield is a bit big, and should be transparent... Shield probably replaces armor.


The medikit can be used to heal self or other players in cooperative or leech health from enemies.


Don't know what that lance does. It's a magic fantasy lance, so it could shoot something. Maybe it's a bit incomplete.

[toneddu2000]

So cool jim, so cool... Can't wait to play it!
Can I ask you how do you realize double wielding system? It's kinda intriguing!

[Barnes]

Radiosity normal mapping with indirect specular approximation and high resolution lightmaps
http://www.youtube.com/watch?v=00pLjnl66R8