Hello

I noticed a shadow anomalies, when I use the ODE Physic with shadow_stencil = 2; (see the screenshot), btw in Manual i found a mistake about "ph_selectgroup ( var bitmask );" under Remarks: Here's a sample bitmask for ph_selectgroup( 6 ), where 6 corresponds to the bits 1010: <- should be a 0110 bits.



Thanks

You're right with the mistake in the manual. But it's hard to judge from the screenshot whether the shadows are wrong or not. Can you post a close up screenshot from the shadow of a single object where the problem is more clear?

JCL: See the horizontal pole, suspended from the ceiling by two bars? See how the pole penetrates the bars?
Look at its shadow to the left. The shadow of the pole does not penetrate the shadows of the bars.
Also on that bar, two blocks are attached to it, but their shadows dont touch its shadow.

Thats all I can see (and be able to explain) ...

I think you must set shadow_stencil to 1 or 3 when you want stencil shadows to be cast from a close light. Or, you must put that light more far away.

With shadow_stencil 2 or 4, the shadow volume is a column and not a cone. So when shadows are cast in different directions, they move aways from each other but the shadow does not get bigger, I hope you understand what I mean. This way, shadows from different meshes will come apart with close lights, it was always this way.

@JCL
Yes, of course here is it.
JCL I will send the project source file to you.


@EvilSOB
I can´t better explain this.

@Spirit
Yes, I´ve casted any entities for the example, but finally I use it the stencil_shadow = 2; without cast flag for the entities.

That's kool. Which with program have you done this wich? I harly get anything like that done with 3D GameStudio! i "guess" it has something to do with programming? Woah..

Rojart: I don't see anything wrong in your screenshots. What you see is just the difference of z-fail shadows that are always parallel, and z-pass shadows that are not parallel. Z-fail treats any light source as if it were infinitely far away. This is visible in your shot. It's perfectly normal, and not a bug.

http://manual.3dgamestudio.net/shadow_stencil.htm

You're right, I read a lot, but I forget most of what I just read eg. about differences between z-pass & z-fail shadows, now work it (see the screenshot, the dynamic light is far far away)

In Manual under remarks about stencil shadows:
> Z-fail stencil shadows ... "A script can be used for tracing a ray from a light source to the camera, and automatically switching to z-fail (You mean to z-pass or I´m wrong?) shadows when the ray hits an obstacle.
Perhaps is there possible to add this as new features like this:
shadow_stencil = -2; or shadow_stencil = -4;
The z-fail stencil shadows will be automatic switching to the z-pass stencil shadows when the light source come close to the camera, without extra coding?

Why Z-fail is always parallel, what can you do with this?
z-pass shadow is impossible to use in first-person shooter, shadow drawn wrong if camera in the shadow.
For example in Doom 3 stencil shadow is camera independent, can you do that in A7 ?
For example
z-pass - shadows draw correctly

Z-fail - WRONG!!

Waiting for answer. Thanks.

Yes, why and I think that the z-fail algoritm is wrong?
Here an example with cpp sources that use it the z-fail algorithm with wonderful stencil shadow.