Friday, 11 February 2011

This week

I kind of have 1 week for my mid term review and as always  i am a last minute worker. So here is my plan for this week, going through that real time photon map thesis, and see for alternatives for real time subsurface scattering and also i will have to try and implement the paper so i will start again on that.

Testing the Results

In my implementation there are two parameters absorb and scatter and at every event point whether there is scattering or absorption is decided by these coeffs.
i.e we take the ratio absorb/(absorb+scatter) at each event point and then generate a random number between zero and one. If the number is less than the ratio the event is absorption else it is scattering. Let me play with the coeffs. If my ab coeff is 0 the image should be black.
The wet spheres are black.
So that is correct.
Now let me try for case scatter coeff is 0. I should get an image similar to my earlier diffuse image.Here i get an additional layer of rings above some radius on the wet surfaces. Why are these rings appearing?

A curious thing i observe is that when i do not generate the other hit points in the media the circles do not appear. What can be the reason for this?

Thursday, 10 February 2011

My implementation

Ok my initial implementation of the wetness paper is as follows.  I am just doing the subsurface part first.
In my hand i have a progressive photon map implementation. This implementation has ray tracing pass first which generate hit points on diffuse surfaces. The whole area is divided into grids and each grid point has a sphere associated with it to collect photons(surrounding ,some initial radius). Each hitpoint is mapped into grids with the help of these spheres.
There are two things i have to deal with in case of subsurface scattering.

  • Photon tracing
In case of photon tracing you have to trace the photon with a monte carlo model(this model is suggested to be efficient because here you only have photons of equal energy). At every distance given by absorption coefficient you do a random number generation to see whether the photon is absorbed or scattered.
If photon is absorbed it is stored there in photon map else it is scattered in random direction. My scene consists only of spheres so i check after the photon hits a surface that whether the photon is inside or outside the sphere at every event distance(given by ab coeff)if it is inside the surface i do the scatt or ab otherwise it is just a normal photon.

  • Ray tracing.
While doing ray tracing i kind of do a ray marching along the ray that enters the wet media. This is done by having hit points at fixed distances(given by the ab and scatt coeffs) and these hit points corresponds to the same pixel.

Wednesday, 9 February 2011

Wetness Rendering Paper

I have my meetings with Prof Subodh Kumar on Tuesdays 11.30 am.
So i will update my blog at least before the meetings.

I am currently working on wetness rendering and my aim is to render wetness as realistically as possible in real time(30 fps). I am currently trying to implement the paper rendering wet material by Jensen. This approach uses photon mapping to render wetness. Since it uses photon mapping it is obviously not real time(Although i saw a paper on real time photon mapping).I have to somehow improvise and innovate it to real time. I am using a basic implementation of progressive photon mapping to help me render the images.

So what does this paper have?
It talks that in order to render wetness it has two models
  • Surface model.
  • Subsurface model.
In case of Surface model it talks about the amount of photons that pass through the layer of water.A thin film of water is on the surface of material due to wetness. The effect of this is given by fresnal's equations(I am confused about some of the concepts). This is obtained with the help of refractive indexes, the incident angle, the transmitted angle(actually the refractive indexes give the angles with the help of Snell's law).
 The Surface model is trying to simulate total internal reflection in the layers causing further absorption.

In case of Subsurface model it basically talks about subsurface scattering. In case of subsurface scattering light enters the object at some place of the object and exits at some other place. Inside the material light can be scattered or absorbed.
The scattering and absorption are given by scattering coefficient and absorption coefficient respectively. The scattering and absorption are given in units of cm-1 and helps us to decide each scattering event. The probability of absorption is given with the help of exponential distribution(some doubts in this area also need to be cleared).

Monday, 7 February 2011


The purpose of this blog is as a way of analyzing my project. By writing down the progress of my project i am planning to think loudly. Hope this helps my project.