next »

[ChrisE]

The fun never stops.

So, it seems that layering (at least as discussed as above) does not work.

Code:

RenderPassConfig rpc;
RenderPass rp;
SkySphere ss;

ss = myXith.this.createSkySphere();
ss.setLayeredModeForced(false);
ss.setUnlayeredModeForced(true);
myXith.this.xithEnv.addRenderPass(ss);

rpc = new BaseRenderPassConfig(org.xith3d.render.RenderPassConfig.DEFAULT_CAMERA_MODE);
rp = new RenderPass(myXith.this.groundBranchGroup, rpc);
rp.setLayeredModeForced(true);
rp.setUnlayeredModeForced(false);
myXith.this.xithEnv.addRenderPass(rp);

rpc = new BaseRenderPassConfig(org.xith3d.render.RenderPassConfig.DEFAULT_CAMERA_MODE);
rp = new RenderPass(myXith.this.geomBranchGroup, rpc);
rp.setLayeredModeForced(true);
rp.setUnlayeredModeForced(false);
myXith.this.xithEnv.addRenderPass(rp);

This still fails to Z-sort correctly, regardless of which combination of setLayered and setUnlayered I use (and I've tried all 8, I believe). I've also tried setting the renderer to use layering and to not use layering, and still no dice.

The results of the above, by the way, are a sky sphere in the background, as desired. In front of that is a large flat green plane I use for the ground, and then my foreground geometry always in front of the both the plane and the sphere, even when it should nominally be behind the ground plane (e.g., seen from below ground it should be occluded).

I'm unsure, but do we know for certain that this code works? There don't appear to be any unit-tests/samples in the SDK.

[Marvin Fröhlich]

For me it works. Note, that layered mode is the default. So, if you add two branch graphs / render passes, the second one will always be in front, no matter how far away from the camera they are. If you force unlayered mode on the second render pass (on the first one it won't have any effect), the two passes will be sorted (z-buffered) as if they were one big group.

If you always want something in front, you just need to put it into a separate branch graph / render pass, that is added to the scenegraph last and just use the defaults.

Marvin

[ChrisE]

So, here's where that code is used:

Code:

// if we're in layered mode, we need to clear the depth buffer.
            if ( ( layeredMode && !renderPass.isUnlayeredModeForced() ) || ( !layeredMode && renderPass.isLayeredModeForced() ) )
            {
                if ( rpIndex > 0 )
                    gl.glClear( GL.GL_DEPTH_BUFFER_BIT );
            }

(RenderPeerImpl.java, line 799)

To the best of my knowledge, that's the only place we use those two functions and the layeredMode variable. So, why don't we just us a setLayeredForced() and isLayeredForced(), instead of also including the unlayered options?

EDIT: Works now. Thanks!

[Marvin Fröhlich]

Because you can change the default .

[ChrisE]

Ah, aye, that'd do it.