[Bf-committers] water painting
Robert Wenzlaff
rwenzlaff at soylent-green.com
Sat Nov 27 03:57:13 CET 2004
On Friday 26 November 2004 13:51, Jeremy Wall wrote:
> implemented as a rendering plugin it makes sense. blender already has
> toon style and traditional raytracing renders why not a watercolor
> style. or as a postprocess plugin for the compositor which would
> probably fit the code he wrote better.
Toon is a shader because Toon needs to be a shader.
Toon shading is a 3D effect, even if it tends to obscure the 3D data with
minimalistic shading. It still uses the 3D data to calculate that effect and
thus could not be implimented as a sequence plug-in.
As I understand it, the watercolor effect being spoke of is a 2D effect.
Not only does the watercolor effect not use 3D info, I'm not sure it can be
implimented as a shader. A shader does not determine the color of a pixel
directly. Shaders determine what each object's color is at a given vector
from the lights/camera. The pixel is then colored with a weighted average of
all the objects that intersect there.
Color boundaries would bleed in watercolor reguardless of their object's
separation on the Z axis. This bleed is done on a pixel to pixel basis
without reguard to where the object is in the scene. There is no mechanism*
in the shading system to allow an object to effect the color of a pixel that
does not intersect the object. You would end up with a nice, photo sharp 3D
scene where it looked like all the objects were covered in watercolor....
Anyway, it can be done now as a sequence plugin. I've always thought that a
single image (or single + static image(s)) sequence plugin should be allowed
to be specified as a post-process. This would make things like zblur much
easier to use.
*OK, Radiosity does this (and glow effects if we ever get them), but it is
still a 3D effect (volume based), and requires a lot of special code. Why
add a lot of special code to the 3D shading tree, when the watercolor effect
can be done so simply in 2D?
--
************************************************************
If the source code is of any help to a cracker in decrypting
a data stream, then the binary wasn't particularly helpful
to the user when encrypting it.
************************************************************
Robert Wenzlaff rwenzlaff at soylent-green.com
More information about the Bf-committers
mailing list