|
 |
Am 27.02.2011 13:19, schrieb gregjohn:
> clipka<ano### [at] anonymous org> wrote:
>>
>> The slope pattern is not a true 3D pattern - it is only defined on the
>> surface of an object (after all it needs a normal vector); so it doesn't
>> make sense to move it around separately from the object. Instead, the
>> slope pattern always "sticks" to the object it is applied to, as far as
>> translations go. (Note however that the slope pattern's "up" vector
>> stays fixed all the while.)
>
> Just making sure I understand the "not make sense." It could be either
> geometrically indeterminate or a massive coding headache to enable, I'm not able
> to say.
It would be geometrically indeterminate for virtually all primitive
types (with a few exceptions like blobs or isosurfaces where the normal
vector can be defined as the gradient of some potential).
> But artistically there could be cases were allowance of warps/
> transforms to the slope pattern could be very useful. One application was in
> making a radius of curvature texture for applying high and low values of SSS.
That may well be - but enabling this would be a major /definition/
headache (as well as an inconsistency headache), because the way warps
work on any other pattern is totally unsuited for perturbing slope patterns.
Maybe the easiest way to implement something close would be to add a
mechanism to directly add, multiply or average multiple patterns, so
that you could modulate the slope pattern with a bozo pattern for instance.
Post a reply to this message
|
|
