Page 1 of 1
Posted: Wed Aug 24, 2005 5:51 pm
by stonelli
Did you respect the rule of RGB at no more than 214,214,214?
That gives you an 85% efficient material which is just about as efficeint as we can get any material.
There is another thread that explains how going higher confuses the RS1 engine.
Posted: Wed Aug 24, 2005 7:55 pm
by Maya69
you canost a detail of emiter
i explain :
the size of emiter is portant for noise
if your emiter is behind a glass (+ noise)
etc,
Posted: Wed Aug 24, 2005 9:02 pm
by Micha
... try to use a low polycount in the scene ...
Posted: Wed Aug 24, 2005 9:22 pm
by Matthew Schrock
Did you check to make sure that the emitter plane normals are pointing out of the fixtures? This one got me once...
Matthew
Re: Speed Optimization? (WIP question)
Posted: Wed Aug 24, 2005 9:22 pm
by Mihai
adehus wrote:The space is approx 9000 sqft
I think this is the problem. Try to section the space, delete objects until the total scene size for each shot is about as big as what is seen by the camera.
Posted: Wed Aug 24, 2005 10:10 pm
by Micha
I'm not sure, but I try to use a low polycount if I can. I'm sure, that a MXI sphere around a scene with 50 polgons was 30% faster as the same scene with a 5000 polygon sphere emitter.
Posted: Wed Aug 24, 2005 10:13 pm
by Mihai
Yeah, sectioning is a bad work around with a renderer that's supposed to make our work easier. I really hope it's just a temporary issue with Maxwell.
Personally I would first make a small preview to check the light, then another small one sectioned to verify how alike they look, and then make the big render....but this really isn't a nice way
Posted: Thu Aug 25, 2005 1:11 pm
by Hugh
Try reducing material reflectance still further, 85% is still very, very high and the higher it is the more light gets bounced and the longer you wait for convergence. For lighting calculations in general, a higher than average set of surface reflectances would 70% for the ceiling (178) 50% for the walls (128) and 30% for the floor (77). If you know the actual reflectances to be used, better yet.
Posted: Fri Aug 26, 2005 11:12 am
by Hugh
adehus wrote:Or just play with the camera settings to let more light in?
That's the one!
Posted: Fri Aug 26, 2005 1:41 pm
by Micha
... I mean the poly count of anything. In Rhino, a NURBS modeler, all objects will be converted to polygons. The meshing setting are important for the poly count. Maybe, your software need some settings too.
One hint more you could try: use big plane extra emitters, for example in for the camera invisible windows (simulate an evening light from outside - blueish). Or a softlight from other directions like an other room beside with open door (yellowish). Do you have seen my last museum rendering? For example you can see a blue light from the left on the lightbox side. Also I have set a yellow light from right. So, you could get live in the scene and more big lightsources against the noise.
http://www.maxwellrender.com/forum/viewtopic.php?t=6092