I know that there is similiar stuff
but I want check quality of some idea
this is my first try of universal hair generator
rather fast parsing but long rendering
(15h on Pentium 200MHz without other tasks)
instead random there is something like evenly
but customizable distributed hairs
now I want replace cones with sphere_sweeps

ABX

Post a reply to this message

Attachments:
Download 'shampoo.jpg' (48 KB)

Preview of image 'shampoo.jpg'

Wlodzimierz ABX Skiba wrote:
> 
> I know that there is similiar stuff
> but I want check quality of some idea
> this is my first try of universal hair generator
> rather fast parsing but long rendering
> (15h on Pentium 200MHz without other tasks)
> instead random there is something like evenly
> but customizable distributed hairs
	It looks pretty good, can we see some code please?

> now I want replace cones with sphere_sweeps
> 
	Try replacing them with a few triangles and make the whole thing a
mesh, it should render *much* faster...


-- 

* Abandon the search for truth, * mailto:ber### [at] inamecom
* Settle for a good fantasy.    * http://www.enst.fr/~jberger
*********************************

Post a reply to this message

>Wlodzimierz ABX Skiba wrote:
> > I know that there is similiar stuff
> > but I want check quality of some idea
> > this is my first try of universal hair generator
> It looks pretty good, can we see some code please?

#macro HairGenerator(Object,Count,Center,Range,Inside,Height)
  #local InsidePig=pigment{object{Inside color rgb 0 color rgb 1}}
  #local R0=1/sqrt(Count);
  #local way=0;
  #local theta=0;
  #while (theta<pi)
    #local way=way+2*pi*sin(theta);
    #local theta=theta+(pi/Count);
  #end
  #local part=way/Count;
  #local theta=pi/Count;
  #local last=0;
  #while (theta<pi)
    #local way=2*pi*sin(theta);
    #local startangle=-2*pi*last/way;
    #local stepangle=2*pi*part/way;
    #local way=way+last;
    #while (way>=part)
      #local way=way-part;
      #local startangle=startangle+stepangle;
      #local
PointInRange=Center+Range*<sin(theta)*cos(startangle),sin(theta)*sin(sta
rtangle),cos(theta)>;
      #local Normal=<0,0,0>;
      #local P0=trace(Object,PointInRange,Center-PointInRange,Normal);
      #if (vlength(Normal)!=0)
        #local P1=eval_pigment(InsidePig,P0);
        #if (P1.x != 0)
          cone { P0 R0 P0+Height*Normal 0 pigment { color rgb 1 } }
        #end
      #end
    #end
    #local last=way;
    #local theta=theta+(pi/Count);
  #end
#end

I plan add such parameters
1. height of hairs as pigment/pattern
2. direction of hairs as pigment/pattern
3. smoothnes of hairs as pigment/pattern
4. color of hairs as pigment/pattern

> > now I want replace cones with sphere_sweeps
> >
> Try replacing them with a few triangles and make the whole thing a
> mesh, it should render *much* faster...

yes, faster to render but not faster to parse
and I know that it should be faster, but how
interpolate sphere_sweeps to mesh ?

ABX

Post a reply to this message

Wlodzimierz ABX Skiba wrote:
> 
> #macro HairGenerator(Object,Count,Center,Range,Inside,Height)
> ...
> #end
> 
	Thank you

> yes, faster to render but not faster to parse
	Well faster overall anyway...

> and I know that it should be faster, but how
> interpolate sphere_sweeps to mesh ?
> 
	The question here is of course: do you really need the kind of detail a
sphere_sweep would produce? Unless you want to look at your hair from
very near, you'd be better off with a simple mesh...


-- 

* Abandon the search for truth, * mailto:ber### [at] inamecom
* Settle for a good fantasy.    * http://www.enst.fr/~jberger
*********************************

Post a reply to this message

> > yes, faster to render but not faster to parse
> Well faster overall anyway...

yes, I agree

> > and I know that it should be faster, but how
> > interpolate sphere_sweeps to mesh ?
> >
> The question here is of course: do you really need the kind of detail
a
> sphere_sweep would produce? Unless you want to look at your hair from
> very near, you'd be better off with a simple mesh...

but if I want build universal macro not for single use ?
I want create looking good hairs from near and from far distance
I can code this with some "level of details" but what about animations ?
zoomed face has from far mesh and from near sphere_sweeps ?
what about distortion during changing of LOD ?
sphere_sweeps has bounding therefore outside head area they
are rather fast I think
and I think that it needs less memory
each sphere_sweep need 4 centers and 4 radius
(catmull or b-spline) - 4x3+4=16 float values for one hair
mesh ? I think that I need at least 6 smooth_triangles
with coordinates and normals to simulate one hair
it means 6x(3+3)x3=108 float values
at my posted image I use over 25000 hairs

ABX

p.s. I'am not sure in my calculations, I don't know
what exactly sphsweep.c and mesh.c put in memory

Post a reply to this message

Wlodzimierz ABX Skiba <abx### [at] abxartpl> wrote in message
news:39edb44c@news.povray.org...
> sphere_sweeps has bounding therefore outside head area they
> are rather fast I think


From my experience, sphere sweeps are VERY slow to render.

Gail
--
********************************************************************
* gsh### [at] monotixcoza              * System.dat not found.         *
* http://www.rucus.ru.ac.za/~gail/ * Reformat hard drive Y)es O)k  *
********************************************************************
* If at first you don't succeed, call it version 1.0               *
********************************************************************

Post a reply to this message

Gail Shaw wrote:

> From my experience, sphere sweeps are VERY slow to render.

Same here. I always have to bound them manually, otherwise they get the
scene crawling. And the more components they have, the slower then get.
:-(
G.

Post a reply to this message

Gilles Tran wrote:

> Gail Shaw wrote:
>
> > From my experience, sphere sweeps are VERY slow to render.
>
> Same here. I always have to bound them manually, otherwise they get the
> scene crawling. And the more components they have, the slower then get.
> :-(

ditto

--
David Fontaine  <dav### [at] faricynet>  ICQ 55354965
My raytracing gallery:  http://davidf.faricy.net/

Post a reply to this message