|
 |
This paper looks interesting from Intel, don't know if POV uses something
like this already, but the speed-up factor compared to OpenRT looks very
impressive (3x - 10x). They claim that their algorithm results in only 1/3
of the traversal steps being required during the trace, by grouping rays
with similar paths in the scene together (I haven't studied the exact method
yet).
ftp://download.intel.com/technology/computing/applications/download/mlrta.pdf
BTW I stumbled across this while looking at real time raytracing for the
Cell processor.
Post a reply to this message
|
|
|
|
"scott" <sco### [at] laptop com> wrote:
> This paper looks interesting from Intel, don't know if POV uses something
> like this already, but the speed-up factor compared to OpenRT looks very
> impressive (3x - 10x). They claim that their algorithm results in only 1/3
> of the traversal steps being required during the trace, by grouping rays
> with similar paths in the scene together (I haven't studied the exact method
> yet).
>
> ftp://download.intel.com/technology/computing/applications/download/mlrta.pdf
>
> BTW I stumbled across this while looking at real time raytracing for the
> Cell processor.
I'm kind of curious about OpenRT. Could somebody sum up its current status? How
fast is it, and what kind of features are missing?
-Mike
Post a reply to this message
|
|
