Invisible wrote:
>> Basically yes. Although there are many ways to reduce the amount of
>> rays needed to achieve certain noise level without affecting
>> "unbiasness". The point is that the rays keep on going without
>> stopping them artificially.
>
> Right. Well presumably there must be *some* limit on the maximum
> recursion depth though?
One unbiased way is to do it like this:
R = randomNumber(0...1)
if(R < diffuseProb)
//Do diffuse stuff and return the traced value;
else
return 0; //So it got absorbed.
So if diffuseProb is (say) 0.9. You have 10% probability that a ray
stops bouncing anymore. This methods let's you still trace to unlimited
depth without the need to specify max trace depth.
Post a reply to this message
|