 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Am 22.08.2016 um 13:11 schrieb Mr:
> For the exporter from Blender to POV it is possible to have a mesh object from
> which the user will have deleted all vertices, this user may not be willing to
> add anthing to the scene, but we have to assume that he does it on purpose, and
> would indeed want to keep track of the transformations of some point, BUT he
> uses an interface and doesn't even want to know what a vector or a matrix is.
> Also, the transformations would have to happen through all the same operators as
> the standard geometry exported. because creating a particular case may have
> inconsistant results.
A vector won't fit your purpose, since you need different code to
transform it.
Given that the described use case doesn't leave a vertex to track, the
most fitting analogon would probably be an empty union.
> Would a declared vector be able to pass through all the transforms a mesh2
> object can ever go through? if so I was wrong and should use a vector, but if I
> HAVE to add a vtransform or anything that all standard geometry can't go
> through, then I can't use it, because it's not a mesh... but it's a mesh. :-)
Maybe what we should do is change the behaviour of the parser when it
encounters a mesh2, to issue "possible parse error" warnings instead of
outright errors when it encounters an empty list of faces or empty list
of indices.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
clipka <ano### [at] anonymous org> wrote:
> Am 22.08.2016 um 13:11 schrieb Mr:
>
> > For the exporter from Blender to POV it is possible to have a mesh object from
> > which the user will have deleted all vertices, this user may not be willing to
> > add anthing to the scene, but we have to assume that he does it on purpose, and
> > would indeed want to keep track of the transformations of some point, BUT he
> > uses an interface and doesn't even want to know what a vector or a matrix is.
> > Also, the transformations would have to happen through all the same operators as
> > the standard geometry exported. because creating a particular case may have
> > inconsistant results.
>
> A vector won't fit your purpose, since you need different code to
> transform it.
>
> Given that the described use case doesn't leave a vertex to track, the
> most fitting analogon would probably be an empty union.
>
> > Would a declared vector be able to pass through all the transforms a mesh2
> > object can ever go through? if so I was wrong and should use a vector, but if I
> > HAVE to add a vtransform or anything that all standard geometry can't go
> > through, then I can't use it, because it's not a mesh... but it's a mesh. :-)
>
> Maybe what we should do is change the behaviour of the parser when it
> encounters a mesh2, to issue "possible parse error" warnings instead of
> outright errors when it encounters an empty list of faces or empty list
> of indices.
Don't bother, it's already fine with zerosphere, and also there are other non
geometrical objects called empties which wouldn't get exported as meshes, so I
used also zerosphere, counting on spheres to be the most tested and optimized
pov objects :-) I'm very happy with the direction pov/uberpov/hgpov, etc is
evolving! thanks youz ! :-)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> > Thanks, I made it a 0 radius sphere with all suggested attributes
>
> Is there another use for a null than to keep track of a location after
> some transformations and/or linking other objects to that location? And
> isn't that something that can much easier be done in POV-Ray by
> declaring and vtransforming a vector?
>
> I'm just curious why you would need a null/empty object in POV-Ray.
I have a use case just now. A series of scenes (animations of particle
simulation) where particles have pseudo-random texture, but are always declared
sequentially, thus the pseudo-random texture will be the same. As a particle
might disappear during the simulation, creating hole in the sequence, I need to
call the texture procedure (using null object) so that the state of the
pseudo-random generator is advanced and subsequent particles get the same
pseudo-random texture parameters as in preceding steps.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>
>>> Thanks, I made it a 0 radius sphere with all suggested attributes
>>
>> Is there another use for a null than to keep track of a location after
>> some transformations and/or linking other objects to that location? And
>> isn't that something that can much easier be done in POV-Ray by
>> declaring and vtransforming a vector?
>>
>> I'm just curious why you would need a null/empty object in POV-Ray.
>
> I have a use case just now. A series of scenes (animations of particle
> simulation) where particles have pseudo-random texture, but are always declared
> sequentially, thus the pseudo-random texture will be the same. As a particle
> might disappear during the simulation, creating hole in the sequence, I need to
> call the texture procedure (using null object) so that the state of the
> pseudo-random generator is advanced and subsequent particles get the same
> pseudo-random texture parameters as in preceding steps.
>
>
>
>
Well, a zero radius sphere prety well IS a null object.
Just push that disappeared particle far behing the camera, with a zero
momentum, or give it a zero radius, and still give it it's texture.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
