Comments for Brandon Pelfrey

Comment on Coding a Simple Octree by Satan

Satan — Thu, 24 Jan 2013 07:44:57 +0000

great articles.
someone only use houdini to mimic your CM advection and getting some nice results.
check here and feel free to post any of view.
http://forums.odforce.net/index.php?/topic/16415-gridless-advection/

Comment on Coding a Simple Octree by Brandon Pelfrey

Brandon Pelfrey — Sun, 06 Jan 2013 19:33:56 +0000

Excellent point For small fixed-size queries, uniform grids will likely be more efficient (I tried to make a note of this property here and in the code). One great benefit not shown here relies on the fact that we can store information in the interior nodes, as I will show in the next article on vortons. If you haven’t seen it before, I suggest you check out “Barnes-Hut Algorithm”.

Also, there will be a later article on SPH (actually more of a recreation of an old article I wrote that no longer exists and many people have asked me for), that uses this exact same spatial hashing technique you described!
Thanks for the insight!

Comment on Coding a Simple Octree by Mikola Lysenko

Mikola Lysenko — Sun, 06 Jan 2013 17:08:32 +0000

Looking at the code, it seems like you are only ever querying against a small region. Have you considered maybe using a regular grid instead? An octree has an overhead per range query of at least , and can in practice be much worse. With a regular you should be able to look up all the points within small, fixed size regions much more efficiently, and the memory overhead can be drastically reduced by storing only the filled cells using a hash table/binary search tree.

Comment on Simulating the Wave Equation on Triangle Meshes by brandon

brandon — Fri, 04 Jan 2013 18:40:22 +0000

The reaction-diffusion video strongly reminds me of smooth life gliders. I’ll have to look into their work to see exactly what they’re doing, though I have done advection-reaction-diffusion work before.
I am currently writing a couple of articles on Octrees and simulating fluids with Vortons.
It would be neat to visualize the eigenspectrum of the Laplacian on an arbitrary mesh

Comment on Simulating the Wave Equation on Triangle Meshes by Mikola Lysenko

Mikola Lysenko — Fri, 04 Jan 2013 15:39:43 +0000

Cool! It would be interesting to see some other equations on the mesh. Maybe you could try seeing if some of Tim Hutton’s work on reaction diffusion equations works well on surfaces:

https://www.youtube.com/watch?v=QJB1Jsk_oTE&list=UUVtErl9kecaaeeicumm7jXg&index=1

Another interesting thing could be to try computing the spectrum of the Laplacian operator, since it has many useful applications. Of course this isn’t anything new, but to my knowledge I’ve never seen a good WebGL demo that does it.

Comment on Simulating the Wave Equation on Triangle Meshes by Mazen

Mazen — Tue, 01 Jan 2013 11:26:28 +0000

Cool. For more updated reasons it also works very well in FF 17.0.1. at least.

Comment on Extending the Simple Fluid Simulation by Jim

Jim — Fri, 03 Aug 2012 02:56:41 +0000

Hello Brandon
Thanks for the 1st&2nd articles.Could you duplicate the FULL code of SELMA and the Characteristic Map to github with this mov shows ?(http://www.youtube.com/watch?v=71ISxEjEnOs&feature=player_embedded)

@maxim:
if you read this article carefully,you will find CM is better than wavelet turbulence.for purpose..

Comment on The Construct by eric_t

eric_t — Tue, 31 Jul 2012 13:20:08 +0000

Another similar code to look at is OpenFOAM:
http://www.openfoam.com/

It is similar to FEniCS, but uses the finite volume method instead, so you don’t use the weak form of the equations. OpenFOAM has quickly become very popular in the scientific community due to the ease of adding/modifying solvers.

Comment on Extending the Simple Fluid Simulation by maxim

maxim — Tue, 31 Jul 2012 04:40:02 +0000

Hello Brandon. Thanx for next article and greate example.
If you currently can not include code in Construct for volume rendering, then send me on email code for study, of course, if you can.
Ok. How about a more complete example, where you can calculate not only the density, but also other information (temperature, energy and so on) and get fire with smoke and rendering data, using blackbody spectrum?
And second. How about adding detail to the data obtained with low resolution and as a result to get – high resolution, without intensive computations (as an example of – wavelet turbulence for fluid, Theodore Kim). Maybe you have other studies in this area?

Comment on Extending the Simple Fluid Simulation by The Construct | Brandon Pelfrey

The Construct | Brandon Pelfrey — Mon, 30 Jul 2012 19:34:32 +0000

[...] Brandon Pelfrey Programming, Science, Gaming Skip to content Home Extending the Simple Fluid Simulation → [...]