
MSAFluid vs. FluidSolver -- Processing vs. Flash (Comparison FAIL, watch http://vimeo.com/4738408)
7 months ago
(Comparison FAIL, watch vimeo.com/4738408 ~ Processing improperly set in this video)
This is a test run of MSAFluid and FluidSolver ran on my multitouch table. Computer stats:
* Windows XP Pro desktop
* Intel Celeron Processor: 1.6GHz
* 2.0 GB RAM
* NVIDIA GeForce 8800 GT Graphics Card
* Projector: Dell 3300MP
Not a fast computer by any means, but a definite difference in performance between Processing and Flash.
You can read a little more about it here: cyancdesign.com/2009/05/flash-vs-processing-fluid-solver/
NOTE: This video uses an unedited MSAFluid Processing example. But the example was saved intended to be used with the multi-touch iPhone app, where the screen is so small, to use normal velocities could make the display go haywire. vimeo.com/4738408 uses high velocities to produce expected results.
This is a test run of MSAFluid and FluidSolver ran on my multitouch table. Computer stats:
* Windows XP Pro desktop
* Intel Celeron Processor: 1.6GHz
* 2.0 GB RAM
* NVIDIA GeForce 8800 GT Graphics Card
* Projector: Dell 3300MP
Not a fast computer by any means, but a definite difference in performance between Processing and Flash.
You can read a little more about it here: cyancdesign.com/2009/05/flash-vs-processing-fluid-solver/
NOTE: This video uses an unedited MSAFluid Processing example. But the example was saved intended to be used with the multi-touch iPhone app, where the screen is so small, to use normal velocities could make the display go haywire. vimeo.com/4738408 uses high velocities to produce expected results.
-
Vimeo: About / Blog / Developers / Jobs / Community Guidelines / Community Forums / Help Center / Site Map / Merchandise
/ Get Vimeo

Previous Week
BTW if you have any question on optimizing it for multitouch - feel free to ask
A few points about the velocity issue. CCV was outputting velocities as distance per frame, whereas I wrote the processing demo as expecting velocities per second so it's very sluggish. Seth probably wrote the flash demo as expecting velocties per frame, so it behaves nicely. In the processing demo, if you want the same behaviour, you'll need to modify the code to multiply the TUIO velocities by 60 (or whatever your tracker is running at). Although I believe the new CCV on the SVN is now sending velocities per second.
On a note regarding performance comparison, I don't think the flash demo has the same number of fluid cells (was it 40x40?) as the processing demo (was it 120x80?), so to compare the performance of the two demos would be unfair. But I must add that I am very impressed with the flash demo's performance anyway, it flies on my macbook pro and looks great, so great job Eugene!
I was surprised to see processing showing such a bad visuals... I'm sure it is due to options mistakes.
Of course Processing is faster than flash. it cant be compared ;) but at good hardware flash runs very smooth and fast. BTW flash resolution is 60 cells width