My guess is due to the dynamic nature of animated entities, Mark is only frustum culling the bounding box of the initial T-Shaped animated figure. So, obviously once animated the figure does not naturally extend to within that region, so you get unwanted culling. Or worse he simply does a point in frustum type check.
I guess it's pointless asking/making a feature request these days, but here's what I suggest.
1.Take original Model and load into a 3d app you used to EDIT/Modify the skeleton/whatever.
2.Find the anim frame that takes up the most 'spatial' space.
3.Find out the distance of the furthest vert from the centre/root node.
3.Add a dummy joint on each axis that is that distance from the centre.
But this joint shouldn't be visible, and it should NEVER move with the animation frames.
Now when b3d loads, it'll use a bigger bounding box because of these dummy joints, one that is big enough to surround the whole entity.
Of course, worse case scenario is mark is simply doing PointInFrustum checks, in which case I have no idea how to work around it.(Without access to b3d's source ;p)
I guess it's pointless asking/making a feature request these days, but here's what I suggest.
1.Take original Model and load into a 3d app you used to EDIT/Modify the skeleton/whatever.
2.Find the anim frame that takes up the most 'spatial' space.
3.Find out the distance of the furthest vert from the centre/root node.
3.Add a dummy joint on each axis that is that distance from the centre.
But this joint shouldn't be visible, and it should NEVER move with the animation frames.
Now when b3d loads, it'll use a bigger bounding box because of these dummy joints, one that is big enough to surround the whole entity.
Of course, worse case scenario is mark is simply doing PointInFrustum checks, in which case I have no idea how to work around it.(Without access to b3d's source ;p)