Kenneth <kdw### [at] earthlinknet> wrote:
> (Remove_Bounds=on needs to be
> in the INI file, though; I don't know how that might affect 'bigger' scenes.
> With it off--the default--POV issues a non-fatal warning, "Unnecessary bounding
> object removed" and eliminates the bounding object.)

  IMO that's a design mistake in POV-Ray worth of a bug report.

-- 
                                                          - Warp

Post a reply to this message

Warp <war### [at] tagpovrayorg> wrote:
> Kenneth <kdw### [at] earthlinknet> wrote:
> > (Remove_Bounds=on needs to be
> > in the INI file, though; I don't know how that might affect 'bigger' scenes.
> > With it off--the default--POV issues a non-fatal warning, "Unnecessary
> > bounding object removed" and eliminates the bounding object.)
>
>   IMO that's a design mistake in POV-Ray worth of a bug report.
>

Very interesting.  I've been living with that for years, scratching my head
about it, and just thought it was 'the way POV-Ray worked.' (I never *could*
make complete sense of the docs' section about bounding, due to this particular
behavior.) Good to know that I wasn't crazy after all!

Ken

Post a reply to this message

"Kenneth" <kdw### [at] earthlinknet> wrote:
> Warp <war### [at] tagpovrayorg> wrote:

> >   Manually specifying a bounding box for the sphere sweep object should fix
> > that problem.
>
> Did a test, and that does indeed work; the OP's object renders MUCH faster, and
> the artifacts are confined to the bounding shape. (Remove_Bounds=on needs to be
> in the INI file, though; I don't know how that might affect 'bigger' scenes.
> With it off--the default--POV issues a non-fatal warning, "Unnecessary bounding
> object removed" and eliminates the bounding object.)

Sorry, I made a major blunder there (I hate it when that happens!) The default
for Remove_Bounds is ON, not off. So it's natural that POV will remove the
manually-added bounded_by object, when using the default setting.

To clarify:  The bounded_by object is used correctly by setting
Remove_Bounds=off, OR by setting Bounding_Threshold to 1 or 0. (The OP's scene
has only one object, so Bounding_Threshold needs to be set at that or lower for
automatic bounding to kick in.) In which case, it doesn't seem to matter
if Remove_Bounds is on OR off. A bit of a mystery, but a bit-off topic too.

Ken

Post a reply to this message

Kenneth <kdw### [at] earthlinknet> wrote:
> Sorry, I made a major blunder there (I hate it when that happens!) The default
> for Remove_Bounds is ON, not off. So it's natural that POV will remove the
> manually-added bounded_by object, when using the default setting.

> To clarify:  The bounded_by object is used correctly by setting
> Remove_Bounds=off, OR by setting Bounding_Threshold to 1 or 0. (The OP's scene
> has only one object, so Bounding_Threshold needs to be set at that or lower for
> automatic bounding to kick in.) In which case, it doesn't seem to matter
> if Remove_Bounds is on OR off. A bit of a mystery, but a bit-off topic too.

  I understood your mistake the first time, and it doesn't change my opinion
that POV-Ray removing by default your manual bounding object from the sphere
sweep is a design mistake worth of a bug report.

  It may be a good idea to remove a manual bounding box eg. if it has been
applied to a sphere primitive, because in that case the bounding box will
only make rendering the sphere slower. However, in the case of a sphere
sweep this is clearly not a case, hence a special case is warranted.

-- 
                                                          - Warp

Post a reply to this message

On 18.03.10 17:29, Warp wrote:
> Kenneth<kdw### [at] earthlinknet>  wrote:
>> (Remove_Bounds=on needs to be
>> in the INI file, though; I don't know how that might affect 'bigger' scenes.
>> With it off--the default--POV issues a non-fatal warning, "Unnecessary bounding
>> object removed" and eliminates the bounding object.)
>
>    IMO that's a design mistake in POV-Ray worth of a bug report.

It is an ancient but with the spline interpolation bounding computations not 
working. It has been known ever since sphere_sweeps where added to POV-Ray 
(as the bug existed in the original source). It was even discussed back then 
( 3.5 beta), but nobody ever volunteered to implement the somewhat 
complicated mathematical solution to computing the correct bounding for 
spline based sweeps :-(

	Thorsten

Post a reply to this message

Thorsten Froehlich <tho### [at] trfde> wrote:
> On 18.03.10 17:29, Warp wrote:
> > Kenneth<kdw### [at] earthlinknet>  wrote:
> >> (Remove_Bounds=on needs to be
> >> in the INI file, though; I don't know how that might affect 'bigger' scenes.
> >> With it off--the default--POV issues a non-fatal warning, "Unnecessary bounding
> >> object removed" and eliminates the bounding object.)
> >
> >    IMO that's a design mistake in POV-Ray worth of a bug report.

> It is an ancient but with the spline interpolation bounding computations not 
> working. It has been known ever since sphere_sweeps where added to POV-Ray 
> (as the bug existed in the original source). It was even discussed back then 
> ( 3.5 beta), but nobody ever volunteered to implement the somewhat 
> complicated mathematical solution to computing the correct bounding for 
> spline based sweeps :-(

  My point was that POV-Ray *removing* the manually-specified bounding box
from the sphere sweep by default is a design mistake which should be changed.

  I understand it's not trivial to calculate a bounding box automatically,
but removing the user-defined one shouldn't be done. There should be some
exception for sphere sweeps (and perhaps some other heavy objects) so that
manually-specified bounding boxes are always preserved for them.

-- 
                                                          - Warp

Post a reply to this message

>   My point was that POV-Ray *removing* the manually-specified bounding box
> from the sphere sweep by default is a design mistake which should be changed.

As far as I understand it, the point of removing manual bounding
is to speed up the rendering of scenes designed before POV-Ray 3.0.
In all later scenes manual bounding was added with the intent of
overriding the automatic bounding anyway. So maybe 3.7 might be
a good time to change the default for Remove_Bound to "off"?

Post a reply to this message

Warp schrieb:

>   It may be a good idea to remove a manual bounding box eg. if it has been
> applied to a sphere primitive, because in that case the bounding box will
> only make rendering the sphere slower. However, in the case of a sphere
> sweep this is clearly not a case, hence a special case is warranted.

Sounds like opening up too many special cases, and the question which 
objects should be treated that way and which shouldn't.

Why not simply output a warning whenever user-specified bounding boxes 
are removed, regardless of object type?

If a user decides that he needs the bounding box, he can then just 
change the .ini file or command line parameters accordingly and live 
happily ever after.

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> Why not simply output a warning whenever user-specified bounding boxes 
> are removed, regardless of object type?

  Because default behavior should be that of the least surprise.

  Nowadays nobody adds bounding shapes to objects (this was only the case
with POV-Ray 2.x and earlier, where automatic bounding was poor or even
inexistent), so it would be better to *not* remove bounding boxes by default,
and have the automatic removal as an option to be specified in the ini file.

  I'm not exactly sure anymore, but I have a faint memory that POV-Ray doesn't
actually remove bounding objects from infinite objects (such as quartics).
If that's the case, then sphere_sweep could simply be flagged as "infinite"
to get that behavior.

-- 
                                                          - Warp

Post a reply to this message

On 23.03.10 22:21, Warp wrote:
>    I'm not exactly sure anymore, but I have a faint memory that POV-Ray doesn't
> actually remove bounding objects from infinite objects (such as quartics).
> If that's the case, then sphere_sweep could simply be flagged as "infinite"
> to get that behavior.

That would cause various other problems along the way...

	Thorsten

Post a reply to this message