If I split a rendering between multiple computers,
and the rendering is using radiosity, should I have
one machine generate the radiosity file to be used
by all the others? Or will Pov do fine without it?
=Bob=

Post a reply to this message

> If I split a rendering between multiple computers,
> and the rendering is using radiosity, should I have
> one machine generate the radiosity file to be used
> by all the others? Or will Pov do fine without it?

You have to generate and save it before and then load the data on the
different machines. I've never tried it myself (because I've only got one
PC), but there have been numerous reports about radiosity not looking alike.
This is because POV-Ray only generates radiosity-samples in the visible
image, so when you split the image onto several machines, these gather
different samples. Gathering the data can also be done on a somewhat smaller
image IIRC.

To make sure that the machines don't gather individual additional data, use
always_sample off. It might be a good idea to slightly overlap the
distributed parts of the image though, so that you could stitch them
together in an image application if borders still don't look proper. But
that's just a wild guess, I'm just not sure if the output will be 100% the
same even when you load the data.

Regards,
Tim

-- 
"Tim Nikias v2.0"
Homepage: <http://www.nolights.de>

Post a reply to this message

"Tim Nikias" <JUSTTHELOWERCASE:timISNOTnikias(at)gmx.netWARE> wrote in message
news:41755d28$1@news.povray.org...
: > If I split a rendering between multiple computers,
: > and the rendering is using radiosity, should I have
: > one machine generate the radiosity file to be used
: > by all the others? Or will Pov do fine without it?
:
: You have to generate and save it before and then load the data on the
: different machines. I've never tried it myself (because I've only got one
: PC), but there have been numerous reports about radiosity not looking alike.
: This is because POV-Ray only generates radiosity-samples in the visible
: image, so when you split the image onto several machines, these gather
: different samples. Gathering the data can also be done on a somewhat smaller
: image IIRC.

Ok, I got that.

: To make sure that the machines don't gather individual additional data, use
: always_sample off. It might be a good idea to slightly overlap the

Ok, I got that too.

: distributed parts of the image though, so that you could stitch them
: together in an image application if borders still don't look proper. But
: that's just a wild guess, I'm just not sure if the output will be 100% the
: same even when you load the data.

Thanks Tim!
=Bob=

Post a reply to this message

For me, it's inevitable that the power will go out approximately 1 day 
before completing a 20-day render.  ;-)

Anyway, someone gave me this code a while back.  I guess you can leave all 
other radiosity settings the same (if not, someone can feel free to let me 
know, but I don't think it matters either way):

      #if (SaveRadiosity)
        save_file "Radiosity.rad" // save radiosity data
      #end
      #if (LoadRadiosity)
        pretrace_start 1           // start pretrace at this size
        pretrace_end   1           // end pretrace at this size
        always_sample off
        load_file "Radiosity.rad" // load saved radiosity data
      #end

So, the first pass (with SaveRadiosity=on), you can render at a lower 
resolution with no AA (or focal blur).  Then for the final-render 
(LoadRadiosity=on, SaveRadiosity=off), you can crank-up the resolution, turn 
on AA, and you'll get a faster render, without much difference in radiosity 
as far as I've noticed.

-- 
Jeremy
www.beantoad.com

Post a reply to this message

"Jeremy M. Praay" <sla### [at] hotmailcom> wrote in message
news:4175b40d$1@news.povray.org...
: For me, it's inevitable that the power will go out approximately 1 day
: before completing a 20-day render.  ;-)

That's my worry now!

[rad setting deleted]
Thanks! I will try these.

=Bob=

Post a reply to this message