Very Computer

flame(P) = (1 - y/h) flame_colour(abs(x/b)) where  P = (x,y)

where the flame is defined in the 2D rectangle going form (-b,0) to
(b,h)

Flame_colour is a function that returns the color of the flame along a
scalar value. My question is how can you get the round shape of the
flame by simply multiplying the flame_colour with (1-y/h)? This will
not produce any round shape but will simply attenuate the color along
the y axis which is not what we are looking for. This looks like an
error to me. Any help would be appreciated.

There is indeed an error in the book Advanced Animation and Rendering
Techniques by Watt & Watt.

The function needed to reproduce the flame on Plate 27 should have
been something like:

flame(X) = flame_colour(sqrt((abs(x/b)^2 + (1-y/h)^2))

instead of

flame(X) = (1-y/h) flame_colour(abs(x/b))

It's no big deal really but you would expect a book with such
reputation to not have errors like that.

Quote:> > I am trying to write a renderman shader to reproduce a flame shape
> > described in Advanced animation and rendering techniques by Watt &
> > Watt. The description is at page 211. Basically they are saying that
> > you can get the flame shape by using the following function:

> > flame(P) = (1 - y/h) flame_colour(abs(x/b)) where  P = (x,y)

> > where the flame is defined in the 2D rectangle going form (-b,0) to
> > (b,h)

> > Flame_colour is a function that returns the color of the flame along a
> > scalar value. My question is how can you get the round shape of the
> > flame by simply multiplying the flame_colour with (1-y/h)? This will
> > not produce any round shape but will simply attenuate the color along
> > the y axis which is not what we are looking for. This looks like an
> > error to me. Any help would be appreciated.