Culling issue #223
Culling issue #223
Conversation
| @@ -2104,7 +2104,7 @@ define([ | |||
| } | |||
|
|
|||
| var uniformState = context.getUniformState(); | |||
| var mv = uniformState.getModelView(); | |||
| var mv = uniformState.getView(); | |||
There was a problem hiding this comment.
The variable name mv for "model view" no longer makes sense, right?
Are you sure about this change. In my OpenGlobe RTC implementation, I use the model-view matrix. This is also how Deron described it.
@kring changes here will affect your terrain code.
There was a problem hiding this comment.
Right, I'll change the variable name.
It would be the model-view matrix if the central body had a model matrix. I think there might be a bug where the uniform state doesn't update the model matrix to the identity if the primitive doesn't have one.
There was a problem hiding this comment.
I'll take a look. I actually don't like the model matrix the way it is now. It is needed in two locations - once returned by update and once in the uniforms list.
So do you think this is the right workaround? Why not add the model matrix to the central body instead of changing the RTC algorithm to look like it is potentially doing something else?
|
This also fixes the "Camera Reference Frame" Sandcastle example, which didn't draw its polylines. |
|
Tests still pass (It would be nice if we had a new test for this specific issues, but I know |
Fix vectors after merging open-source.
The culling volume should use the camera position and orientation in world space. Looks like I introduced this when adding the culling volume. You can test it with the Camera Reference Frames Sandcastle example. This fixes part of issue #209.
Also, the central body was using the model-view matrix of the uniform state instead of the view matrix. I fixed that which addresses issue #117. @mramato This also fixes the problem in the viewerCamera branch where you were seeing parts of the central body orbiting with a satellite.