About the implementation of Joint Importance Sampling

Practical and theoretical implementation discussion.
Post Reply
shiqiu1105
Posts: 138
Joined: Sun May 27, 2012 4:42 pm

About the implementation of Joint Importance Sampling

Post by shiqiu1105 » Tue Jan 20, 2015 10:09 pm

Recently I am very interesting in this Joint Importance Sampling: http://www.ci.i.u-tokyo.ac.jp/~hachisuka/jis.pdf

It seems to reduce the variance of volume rendering a lot.
But it looks quite complicated to implement, especially when considering the tabulation in anisotropic phase function.
So is this implemented anywhere, so I could use as reference?

Thanks,

ingenious
Posts: 282
Joined: Mon Nov 28, 2011 11:11 pm
Location: London, UK
Contact:

Re: About the implementation of Joint Importance Sampling

Post by ingenious » Wed Jan 21, 2015 1:03 am

Unfortunately, there is no public implementation available, due to potential legal problems (the work was done at Disney Research). However, implementing the analytical importance sampling routines is actually quite easy. For this what you really need are the boxed equations in section 5. The tabulated importance sampling routines are a little more involved, but the analytical ones will get you a long way.
Image Click here. You'll thank me later.

shiqiu1105
Posts: 138
Joined: Sun May 27, 2012 4:42 pm

Re: About the implementation of Joint Importance Sampling

Post by shiqiu1105 » Wed Jan 21, 2015 1:12 am

ingenious wrote:Unfortunately, there is no public implementation available, due to potential legal problems (the work was done at Disney Research). However, implementing the analytical importance sampling routines is actually quite easy. For this what you really need are the boxed equations in section 5. The tabulated importance sampling routines are a little more involved, but the analytical ones will get you a long way.
Thank you for the hint ingenious! I will look into it.
One more quick question, I understand that BDPT is one of the estimators used in your latest UPBP, so is this joint sampling used in that framework as well?

ingenious
Posts: 282
Joined: Mon Nov 28, 2011 11:11 pm
Location: London, UK
Contact:

Re: About the implementation of Joint Importance Sampling

Post by ingenious » Wed Jan 21, 2015 2:11 am

shiqiu1105 wrote:I understand that BDPT is one of the estimators used in your latest UPBP, so is this joint sampling used in that framework as well?
The joint importance sampling techniques weren't actually used there (couldn't do it on time), but they can certainly be added to improve sampling.
Image Click here. You'll thank me later.

shiqiu1105
Posts: 138
Joined: Sun May 27, 2012 4:42 pm

Re: About the implementation of Joint Importance Sampling

Post by shiqiu1105 » Thu Jan 22, 2015 1:31 am

ingenious wrote:
shiqiu1105 wrote:I understand that BDPT is one of the estimators used in your latest UPBP, so is this joint sampling used in that framework as well?
The joint importance sampling techniques weren't actually used there (couldn't do it on time), but they can certainly be added to improve sampling.
Thanks for the info.
Okay I will try to implement the analytic sampling first to do monte carlo subsurface scattering, hopefully post some results.

Been reading about dipole diffuse approximation and just couldn't understand it :(

Post Reply