Does R300 have same shadow problem in SS:SE with R200?

[Zephyr]

A guy told me that:

1、If he runs it in Extreme Quality mode with nVidia cards, such as G2/3/4, SS:SE can have a shadow for dynamic object like this.

This shadow can change shape following object's action, and will disappear when object moves in building's shadow . At same time, this shadow consumes system resource seriously(FPS drops heavily?). If using ATi's card, such as R100 and R200, he cannot see such shadows for dynamic objects.

2、If running SS:SE in non-Extreme Quality mode, he can see shadows like fading disk-shape for all dynamic objects in both nVidia's cards and ATi's cards, including shadow for bullet.

The guy want to know whether R300 can resolve this problem.

I dont have SS:SE, so I can only judge it by these words. At first, I think SS:SE uses shadow mapping in non-Extreme Quality mode, and maybe uses shadow volumes in Extreme Quality mode. Because shadow volumes has more strict requirement for quality of driver, I think whether R300 or R200 can display those shadows depends on ATi driver's quality.

However, someone posts some pictures(Extreme Quality mode) which confuse me:

Anyone can explain it? Thx.

 

[OpenGL guy]

SS changes how it renders stuff based on a config file. If the config file is not modified, then some features might not be available on all cards.

I don't know if this answers your question, but it's worth noting.

 

[gking]

R200 doesn't support shadow mapping.

R300 does, but the driver doesn't support the SGIX_ shadow map extensions, according to digit-life's review.

So, unless Croteam adds an ARB_shadow path to SS:SE (or ATI adds the older non-ARB extensions to their driver), dynamic shadows won't be available.

 

[DemoCoder]

Is SS really using shadow buffers? If so, it should exhibit full self shadowing and realistic occlusions. I would have thought that those shadows are simple projective shadows using a stencil mask to clip them to the projection plane.

e.g. http://www.opengl.org/developers/code/glut_examples/examples/dinoshade.jpg

 

[Zephyr]

R200 doesn't support shadow mapping.

R300 does, but the driver doesn't support the SGIX_ shadow map extensions, according to digit-life's review.

So, unless Croteam adds an ARB_shadow path to SS:SE (or ATI adds the older non-ARB extensions to their driver), dynamic shadows won't be available.

How to explain "If running SS:SE in non-Extreme Quality mode, he can see shadows like fading disk-shape for all dynamic objects in both nVidia's cards and ATi's cards, including shadow for bullet. " ?

 

[Zephyr]

Is SS really using shadow buffers? If so, it should exhibit full self shadowing and realistic occlusions. I would have thought that those shadows are simple projective shadows using a stencil mask to clip them to the projection plane.

e.g. http://www.opengl.org/developers/code/glut_examples/examples/dinoshade.jpg

It seems that SS:SE dont use stencil buffer. U can check the console output: (0 for stencil).

 

[DemoCoder]

They could simply do projective shadows without the clip part. It works until you run into an intersection and the shadow slides under the wall, or off the edge of a cliff.

 

[gking]

Democoder wrote:
Is SS really using shadow buffers?

I believe so - but I'm not 100% sure. It does explain the behavior, though.

If so, it should exhibit full self shadowing and realistic occlusions

Only if you do the shadow test for all objects. If the only object doing the depth comparison is the ground, you won't get self-shadowing or players casting shadows on each other.

Zephyr wrote:
he can see shadows like fading disk-shape for all dynamic objects

You mean little circles under all objects? That could be accomplished pretty easily using a technique like the one used for drawing "bullet holes" into walls.