Has anyone though about creating a utility to optimise POVRay code?

I'm thinking along the lines of something that will parse an object
and simplify the various rotations and translations.  I tend to fiddle
with things until I get them laid out exactly as I want them, then
find that I've got a big heap of rotations and translations scattered
all over the place that could probably be simplified to just a few of
each.

Just an idea...

Defective O-O

Post a reply to this message

Defective <def### [at] homecom> wrote:

> Has anyone though about creating a utility to optimise POVRay code?

> I'm thinking along the lines of something that will parse an object
> and simplify the various rotations and translations.  I tend to fiddle
> with things until I get them laid out exactly as I want them, then
> find that I've got a big heap of rotations and translations scattered
> all over the place that could probably be simplified to just a few of
> each.

Well, since POV does this internally, turning all of that stuff into a
single transform, it would only make a difference in the parsing, and
parsing is fast enough that I don't think it's high on anyone's priorities
(this coming from someone who does images with 2+ hour parses)

Geoff

Post a reply to this message

> (this coming from someone who does images with 2+ hour parses)

2 hours??  You mean you need 2 hours to before the render starts and figure out
you missed something??  Geez!...  My own parse alwasy take less than a few
seconds, but my renders often take a few days/week!!  I dislike to many objects,
I rather like a few very nice ones with transparency, textures, reflection,
refraction, etc... all those effects that require a good computing CPU, but
almost no memory! ;)

-- 
+-------------------------+----------------------------------+
| Simon Lemieux           | Website : http://www.666Mhz.net  |
| Email : Sin### [at] 666Mhznet | POV-Ray, OpenGL, C++ and more... |
+-------------------------+----------------------------------+

Post a reply to this message

On 14 Nov 2000 07:44:26 -0500, Geoff Wedig
<wed### [at] darwinepbicwruedu> wrote:

>
>Well, since POV does this internally, turning all of that stuff into a
>single transform, it would only make a difference in the parsing, and
>parsing is fast enough that I don't think it's high on anyone's priorities
>(this coming from someone who does images with 2+ hour parses)
>

Actually...

I'm thinking along the lines of generating something I can read
myself.  my machines all parse faster than I do.  An example of the
problem would be my current project...I'm rendering all the parts I've
built for my truck.  The parts that mount to the box have all been
generated through much fiddling.  Now when I look at the code, I spend
a lot of time wading through the various rotations, translations and
additions and subtractions that have been used to place the bits and
pieces where I want them.  I've been cleaning up the individual
objects and it's taking me hours on some parts.  This should be
something that can be automated.

Defective O-O

Post a reply to this message

Simon Lemieux <lem### [at] yahoocom> wrote:

>> (this coming from someone who does images with 2+ hour parses)

> 2 hours??  You mean you need 2 hours to before the render starts and figure out
> you missed something??  Geez!...  My own parse alwasy take less than a few
> seconds, but my renders often take a few days/week!!  I dislike to many objects,
> I rather like a few very nice ones with transparency, textures, reflection,
> refraction, etc... all those effects that require a good computing CPU, but
> almost no memory! ;)

I do extremely detailed objects, with lots of pieces, designed by macros for
the most part.  This sometimes means my computer goes into thrashing with
swaps.  Once the render starts, swapping isn't an issue, though.

The pic that was worst on this took close to 500 hours to render, all told
(it was a stop start over weekends and at night)

And I do use transparency, textures and all that as well, so both the parse
and the render are *slow*.

Lately, I've been experimenting with and learing all the new Megapov
features.  Many of these (especially isos) have reduced my parse and render
time greatly, but I still look for parses in the tens of minutes.

Geoff

Post a reply to this message

I'm frequently in a similar position.  Often my scenes develope as I work on
them and they rarely end up like what I first envisioned.  This means when I
go back to change something in the code it is often much different than the
neat, "orderly" script that I first intended to create.  I know there is no
excuse for messy coding but when I get seriously get the creative juices
flowing neatness somehow falls to the back burner ;)

I too, would find something along those lines useful.  Something similar to
an html "beautify" utility, that might do the above forementioned
("condense" rotations and translates, etc.) as well as make comment
seperators and such uniform, etc.


"Defective" <def### [at] homecom> wrote in message
news:elf41tkr79q176h6et1t7uro4hhn6a71na@4ax.com...
> On 14 Nov 2000 07:44:26 -0500, Geoff Wedig
> <wed### [at] darwinepbicwruedu> wrote:
>
> >
> >Well, since POV does this internally, turning all of that stuff into a
> >single transform, it would only make a difference in the parsing, and
> >parsing is fast enough that I don't think it's high on anyone's
priorities
> >(this coming from someone who does images with 2+ hour parses)
> >
>
> Actually...
>
> I'm thinking along the lines of generating something I can read
> myself.  my machines all parse faster than I do.  An example of the
> problem would be my current project...I'm rendering all the parts I've
> built for my truck.  The parts that mount to the box have all been
> generated through much fiddling.  Now when I look at the code, I spend
> a lot of time wading through the various rotations, translations and
> additions and subtractions that have been used to place the bits and
> pieces where I want them.  I've been cleaning up the individual
> objects and it's taking me hours on some parts.  This should be
> something that can be automated.
>
> Defective O-O
>
>

Post a reply to this message