noticed that the multitexture-passed box flickers when moving the camera…i first thought this was due to a renderQueue problem with the same box z-fighting with itself…but after some digging I found out that it was the renderOrthoBucket->renderer.setOrtho() that made the flicker…why set opengl to orthographic mode if there are no QUEUE_ORTHO-objects in the scene? (even so, it shouldnt flicker anyways…)
Added a check locally to skip a bucket if it's size is 0. Not sure why any kind of flicker would occur though. (Considering ortho is getting turned on/off on all the tests anyhow. Maybe because it is happening three times per frame in that multipass demo? (3 passes, one with ortho content)
the flicker is due to the box in the second pass not getting drawn(z stuff or something)…dunno why orthomode makes that happen though
yeah, that makes sense… I bet you have an nvidia card right? We had the same issue with shadows which was fixed with a polygon offset call. ATI cards have no issue with z checks on equals… very frustrating.
you are so right, nvidia geforce 6800…no probs on the radeon 9800 at work…thanks
does that mean your gonna add a polygon offset to fix the problem?
yes, as mentioned above…
I have ati and just wondering does this have an adverse effects on me. Even 1 frame? I'm just curious.
No it shouldn't have much if any impact. Even if it had a very small one, it's a small price to be paid for working across card types.
ya i was just thinking maybe this could be a platform to use vender specific extensions and have decriminate between the 3 major graphics companies(ATI,Nvidia,intel) and stuff like this could be avoided and there would be the use of vender specific extensions which i'm not sure but belive are faster than arb extensions. just a thought
Maybe, but there will be a time later in jME's development when it will be more appropriate to do such finetuning.