Yash Polke


About Me

Technical Director



Houdini Skills


Not Specified

Recent Forum Posts

Dual rest in flip 2020年12月7日14:23

Hello all,

I am facing a dual rest issue, where with the default values(Frame delay = 50, Frame offset = 1) I get popping in my renders.

What are the ideal values the dual rest attributes could be set to, for a 100 frame simulation with 50 frames pre-roll, so that the render does not pop?


How to distribute/combine flip sims on a single machine using H18.5? 2020年11月18日20:27

Hello all,

I am trying to do a very long twisted / bent river simulation by dividing the terrain into a few sim sections. Using HQueue (as per Houdini Help), at the moment, unfortunately is not a solution that I can use.

So my questions are:
A. Is there a way in which I can simulate different sections of the river terrain, and combine them coherently, diminishing the boundary layer artifacts generated due to Upper and Lower padding inside the Flip Sims?

B. Is there another method for distributing simulations, not necessarily on the farm?


How to normalize distance within a point cloud? 2020年4月16日15:31

Hello all,

I hope someone knows the answer to this.

1: I have a horizontal line aligned to Z-axis.
2: I create a tube out of it using the polywire node.
3: Using mountain sop using the defaults I create some noise on it(almost looks like periodic noise) and I just reduce the frequencey
of noise by increasing the Element size parm, nothing fancy.
4: Using points from volume node, I scatter some points inside this noisy tube.
5: I calculate the near distance, from the line to the point cloud like so:
int nearpt = nearpoint(1, @P);
vector nearpos = point(1, ‘P’, nearpt);
f@near_dist = distance(@P, nearpos);
6: Now that I have the distance, what I want to do is normalize this distance. Which means creating value of 0(from the horizonatal line) to 1(outer edge/periphery of the noisy tube).
7: To “normalize” this value of f@near_dist I use the fit function, fitting f@near_dist from its min_value to it's max_value to 0-1.

This here is the problem. I don't seem to get proper distribution of values from 0-1 within this point cloud. The value on the outer edge of the noisy tube is anywhere from approx 0.1-0.98. I need those values to be 1 so that the point cloud has values from 1(from outer edge) progressively decreasing to 0(which is the core of the point cloud/line). Imagine if you take a cross-section of this tube in any section, the value should always go from 0(in the center) to 1(outermost radius), no matter where the cross-section is.

I also tried using sdf's. But I can't have an sdf so high-res as to give me an accurate gradient within the point cloud.

So my question is what is the way in which I can achieve this normalization of the f@near_dist?

Looking forward to your answers, I have attached a file for your reference.