Originally, I had created what I called "naked primitives" by using 
cylinder elements in blobs to create sort of a wire frame of a cube, 
pyramid, and a sphere, with some radiosity for effect.  The blobbing 
created a bit more mass at the vertices, which was the effect I was 
going for.  Then, I noticed that the sphere didn't much resemble a 
sphere, but rather this odd shape that looked kind of cool.

Then I decided to take the sphere-like shape on its own, make an array 
of them, make every other one reflective, put it over a checkered plane, 
and add some focal blur.  And this is the result.

I don't like it though...the focal blur isn't quite right.  The settings 
I used were 200 samples, confidence 0.99, variance 1/100000, aperture 4, 
with the focal point in the very center of the sphere.  The problem is 
that the focal blur still looks grainy, and there are couple black 
pixels.  I'm assuming the black pixels are from the max_trace_level 
being reached.  I'm using the default.

What also gets me is some of the edges in the checker pattern are abrupt 
in the focal blur.  What would cause that?

Post a reply to this message

Attachments:
Download 'rsocp_fp.png' (251 KB)

Preview of image 'rsocp_fp.png'

DJ Wiza nous apporta ses lumieres en ce 05/04/2006 04:33:
> Originally, I had created what I called "naked primitives" by using 
> cylinder elements in blobs to create sort of a wire frame of a cube, 
> pyramid, and a sphere, with some radiosity for effect.  The blobbing 
> created a bit more mass at the vertices, which was the effect I was 
> going for.  Then, I noticed that the sphere didn't much resemble a 
> sphere, but rather this odd shape that looked kind of cool.
> 
> Then I decided to take the sphere-like shape on its own, make an array 
> of them, make every other one reflective, put it over a checkered plane, 
> and add some focal blur.  And this is the result.
> 
> I don't like it though...the focal blur isn't quite right.  The settings 
> I used were 200 samples, confidence 0.99, variance 1/100000, aperture 4, 
> with the focal point in the very center of the sphere.  The problem is 
> that the focal blur still looks grainy, and there are couple black 
> pixels.  I'm assuming the black pixels are from the max_trace_level 
> being reached.  I'm using the default.
> 
> What also gets me is some of the edges in the checker pattern are abrupt 
> in the focal blur.  What would cause that?
> 
> ------------------------------------------------------------------------
> 
You may try with a lower aperture value, like in the 1~2 range. Smaler aperture should
make the 
grainyness less apparent, or you can increase the sample number, along with the render
time.
You probably need to increase Max_trace_level, and maybe use a larger adc_bailout.
For the black leaking on the white, you may try using Gray05, Gray10 or Gray15 for the
dark squares. 
Don't know if it will help.

-- 
Alain
-------------------------------------------------
Coming soon: Windows for Nintendo!

Post a reply to this message