## Search found 75 matches

Wed Sep 30, 2015 3:21 pm
Forum: General Development
Topic: Separate GPU compute kernels vs. a single GPU kernel
Replies: 10
Views: 11291

### Re: Separate GPU compute kernels vs. a single GPU kernel

I'm interested in this.
What is the data layout of intermediate buffers of your experiments, SoA or AoS?
GPU has very limited size of cache per thread compared to CPU, so data layout seems to strongly influence the performance.
Tue Jul 21, 2015 2:00 am
Forum: General Development
Topic: Multiplication between two slerped quaternion
Replies: 7
Views: 5952

### Re: Multiplication between two slerped quaternion

Interpolating between two transformation matrices using simple linear interpolation results incorrect transformation matrix. For example, let's consider two rotation matrices around X axis by 0 radians and pi radians. These matrices are below: 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 and 1 0 0 0 0 -1 0 0 0 0...
Mon Jul 20, 2015 11:04 am
Forum: General Development
Topic: Multiplication between two slerped quaternion
Replies: 7
Views: 5952

### Re: Multiplication between two slerped quaternion

Is your notation: (t - 1) * A + t * B implicitly meaning spherical linear interpolation? if you mean simple linear interpolation, it is wrong. I know that Slerp(M1, M2, t) * Slerp(M3, M4, t) is the obviously correct calculation and the result from this is what I want. However, this require two slerp...
Sun Jul 19, 2015 2:45 am
Forum: General Development
Topic: Multiplication between two slerped quaternion
Replies: 7
Views: 5952

### Re: Multiplication between two slerped quaternion

MohamedSakr wrote:everything should be "flat"
As you say I firstly convert the hierarchy to a flat object list.
What I want to know is whether it is possible to convert hierarchical animated transforms to one animated transform or not.
Wed Jul 15, 2015 6:20 pm
Forum: General Development
Topic: Multiplication between two slerped quaternion
Replies: 7
Views: 5952

### Multiplication between two slerped quaternion

Hi, I want to implement motion blur with scene hierarchy. As far as I know, the simple way to achieve motion blur is doing ray-object intersection in the object space. First transform a ray into the object space using a transformation matrix sampled at the time t, then do the intersection test and t...
Tue Apr 07, 2015 10:00 am
Forum: General Development
Topic: Out-Scattering problem
Replies: 6
Views: 6017

### Re: Out-Scattering problem

The paper for Residual Ratio Tracking will be a good reference for you. http://drz.disneyresearch.com/~jnovak/publications/RRTracking/index.html If you want to compute transmittance in in-homogeneous media in unbiased way, you can use Woodcock tracking. Furthermore, in the paper a better method RRTr...
Tue Apr 07, 2015 12:39 am
Forum: General Development
Topic: Out-Scattering problem
Replies: 6
Views: 6017

### Re: Out-Scattering problem

Let's consider the eye point x0, the nearest surface point x1, the point sampled on a light source y0. You mean your renderer samples a few points(xp) along x0x1 and connects xp and y0, add a contribution attenuated by the factors T(y0, xp) and T(xp, x0), followed by adding contribution from the sur...
Fri Mar 06, 2015 3:50 pm
Forum: General Development
Topic: Defining roughness
Replies: 6
Views: 5817

### Re: Defining roughness

My opinion is that there is no clear boundary between diffuse and glossy surfaces.
That is why methods using MIS like BPT or (Progressive) BPM are robuster on glossy-dominant scenes than other methods.

I'm sorry I'm not familiar with the other subjects.
Tue Mar 03, 2015 9:30 am
Forum: General Development
Topic: Some images from my renderer
Replies: 8
Views: 7583

### Re: Some images from my renderer

Cool!!

How much vertices do the flowers consist of?
Fri Jan 02, 2015 10:56 am
Forum: General Development
Topic: Questions on BDPT and Sampling
Replies: 3
Views: 4073

### Re: Questions on BDPT and Sampling

You don't need average the samples. Let's consider intuitively about the MC contribution of the paths you mentioned. path1: w1 * (C1 / p1) path2: w2 * (C2 / p2) The factors (C* / p*) are usual MC contributions. w* are MIS weights, so the values of both are less than 1. If p1 and p2 are accidentally ...