In this short post I am going to talk about how we used and integrated built-in XSI particle system to make smoke simulation for our pipe smoker character in Vishka Studio’s current short animated movie Baran.
read the full text and post here :: http://www.vishka.com/blog/?p=87
Good Luck!

Using the QuarticBlur filter was a good idea. Most people wouldn’t have thought to use that.

Oh what a great idea ! Much more faster than volume shader rendering !

I read your article about Smoke Simulation. And found it awesome. So i also suggest Using the QuarticBlur filter was a good idea. Most people wouldn’t have thought to use that.

Ok, so from the stand point of a compositing Noob, what is the Quartic Blur doing exactly. The docs say it’s a constant time blur. What does this mean in noob terms?

Thanks,
Eric T.

Haha, :D, I want to know too, so I’m a noob.

its cool xD

I read your article about Smoke Simulation. And found it awesome. So i also suggest Using the QuarticBlur filter was a good idea. Most people wouldn’t have thought to use that.

Hi guys,

Just been looking around and trying to find out what Quartic Blur actually is? I know from where i’ve looked that it has to do with blurring pixels and it will take more time the more pixels you are doing it with.

If someone could just drop a few lines that would be much appreciated :)

N.De Bond

From overhearing the mathematics of it, and my horrific simplification of that, to a noob it would be answered like this…

In both Gaussian and Constant time blurs you have a blur radius, if you take a large image eg 4k, and blur it with Gaussian the time it takes to do the blur is dependent on the radius chosen.

Eg Gaussian: radius = 20, it would do it quickly. Radius=4444, it would take a long time.

The Constant time functionality is a way to reasonably know (for the user) a constant (ie the same) amount of time is used irrespective of the radius chosen, within certain limits.

Eg Constant time: radius = 400, it would do it in ‘n seconds’. Radius=4444, it would do it in approx ‘n seconds’ and the approx delta is very small.

The typical exception to this is over the smaller end of the radii, eg 2. Where the time would be less than the next quantum, which maybe at 40 (case dependent). That said, the next larger quantum, the more approximate the time is likely to be to that previous quantum and similiar to the next quantum etcera.

I suppose the above is not so noob (ie? y?) an answer...but think of it like this; a step function, where the steps at the begining of the radii, change a little (and the height is the time taken), as the radii get larger and larger the time to do them remains pretty much the same.

The ultra simplification has to be; if you want to use large radii don’t use Gaussian use Constant time blurs.

I hope that helps.

Terry Jones
Customer Support
Letterbox Animation Studios
http://www.LetterboxAnimationStudios.com

Awesome post, thanks! [RIGHT][U]simulation emprunt immobilier[/U][/RIGHT]