AWS Thinkbox Discussion Forums

Comparisons

Hi guys - I know there was a page that showed legacy blobmesh vs pwrapper vs Frost on thinkboxsoftware.com but I can’t seem to find it. URL?

Khye

I can’t seem to find a URL, but there’s a copy in the old Krakatoa help files.

You will need to unblock Windows security on the .CHM file before you can open it. Copy Krakatoa1.6.0.43192.chm onto your desktop. Then right-click the file and choose Properties. In the Properties dialog, click the “Unblock” button, which is near the bottom-right corner of the window.

Now you can open the help file. Switch to the Search tab, and search for PRTMesher. The page you’re looking for is called “PRTMesher Benchmark”.

Yippie Paul found it in the old manual. :slight_smile:

I know which one you are talking about it used to be here:

software.primefocusworld.com/sof … chmark.php

I think some things did not make the transition from Prime Focus :frowning:

I also noticed that the links that describe some of the Magma Flow Krak tuts are broken. Someone might want to fix those, too.

First of all, the comparison that was posted as a hidden page in the Krakatoa manual was using PRT Mesher. That build was at least an order of magnitude SLOWER than the shipping version of Frost. So while it was impressive already, it does not represent Frost’s performance anymore.

I had a new set of benchmarks done just before the release of Frost, but the results were so much in Frost’s favor I felt it would be a bad move to release them publicly. My theory was that posting them would look like an attempt to put down the competition, while providing a trial license would actually let people experience the performance for themselves.

I am on vacation in Europe right now and don’t have access to my home machine, so I cannot even look at those files.
But you could easily set up a test using BlobMesh, PWrapper and Frost on your own machine and try to stop the time (either viewport or renderer) using the same PFlow system.

If you can find a case where Frost isn’t faster than the alternative solutions while using the same source particles and generating approximately the same geometry output, please let us know!

Please post URLs when you encounter them!

KCM links in software.primefocusworld.com/sof … dFlow2.zip

Also wanted to say I’m just now doing some Krak+Frost to do some great dissolving of solids. Was Frost used on the jeep hood’s disentegration in GI Joe?

If you mean the RnD videos with the Hummer door falling apart, I used some similar but even more powerful technology (Level Set Mesher) which was a component of Flood. Baiscally, I used a 3D texture to blend between a full Level Set version of the door and an empty Level Set, producing a volumetric animation of the door and testing particles using the Krakatoa Geometry Test operator in PFlow to release particles that were not in the volume anymore. The result looked similar to Dieter Morgenroth’s render-time voxel renders in VRay, but we actually generated a mesh in the viewport and in the renderer, so we could perform all mesh-related operations like volume tests, kd-tree ray intersection tests, stealing of normals and so on - pretty much everything Frost does, but based on voxel data.

The hood tests were done by Laszlo and he used various methods including a procedural noise texture that could be controlled by moving particles along the surface, Level Set meshing, cloth simulations and whatnot.

In the actual movie, we had to destroy a tank and a plane, and they used some of these methods, but without the Krakatoa Particles component and no Frost. The Level Set Mesher was involved somewhat.

Thanks for the report!
I fixed the CloudFlow2.zip download link (thankfully, I had a copy of the file on my laptop).
As for the missing html page, I have a full backup of those pages at home (and I am sure someone at Thinkbox has the same backup), but I cannot restore it right now.
Stay tuned!

Privacy | Site terms | Cookie preferences