As a fun project, I'm also working on sharing out POVRay rendering over 
a TCP/IP network (such as the Internet). Project doc is up at 
http://www.newerawd.co.uk/parpov

James Jackson.

Gergely Vandor wrote:

> "Hugo" <hua### [at] post3teledk> wrote in message
> news:3c28f2b2@news.povray.org...
> 
>>I'd like to have such a program. Maybe it wouldn't even have to be
>>restricted to share power for Povray. But how will you split a Pov
>>
> rendering
> 
>>into pieces? The wise people say you can't use multiple CPU's in one
>>computer, so how will you do it over the Internet?
>>
> 
> Povray is able to render only a specified region of an image. It should be
> fairly easy to combine these regions to the whole image. Although as I'm
> writing this I'm becoming a bit uncertain since AA, radiosity, and photons
> (and maybe some other features too) can be problematic.
> 
> 
>>And there has to be done a lot for security as already said. Especially if
>>it's a broad application that allows other programs than Povray.
>>
> 
> If we're talking about povray, the only  real risk is file IO, which can be
> restricted to a single dir.
> 
> Cheers
> 
> Gergely
> 
> 
>

Post a reply to this message

2nd reply - sorry :)

This sounds like it'a gonna turn into a very professionl job - my system 
is very much an amateur job. I'm starting simple (ie only able to 
specify render resolution, whether AA is turned on or off). When the 
system is coded, I then intend to add further options (handling of 
animation rendering, radiosity etcetc).

I make no claims to be an expert programmer (For a start I'm 17, still 
at school, and never had any formal C++ training :) and I'm sure the way 
I go about some things is a little arse-about-face, but It's good fun!

I welcome comment on my project document - anything is helpful.

Also, I have a little problem with calling POVRay. To calculate values 
for the load sharing, I am performing timed test renders of each row in 
a (scaled width) x 100 version of the scene. If I call POVRay with +SR1 
+ER1, the whole image gets rendered. However, if I use a value greater 
than 1, such as +SR6 +ER6, I only get that row (as required). What is 
going on in the +SR1 case?

I'm using the latest 3.1 POVRay MSDOS binaries.

Cheers,
James Jackson.

Gergely Vandor wrote:

> "Mike Williams" <mik### [at] nospamplease> wrote in message
> news:Md7### [at] econymdemoncouk...
> 
>>Wasn't it Gergely Vandor who wrote:
>>
>>
>>>Povray is able to render only a specified region of an image. It should
>>>
> be
> 
>>>fairly easy to combine these regions to the whole image. Although as I'm
>>>writing this I'm becoming a bit uncertain since AA, radiosity, and
>>>
> photons
> 
>>>(and maybe some other features too) can be problematic.
>>>
>>AA is not a problem. The pixels just outside the specified region are
>>calculated and used for the AA calculations, but are not output.
>>
>>For radiosity, what you probably need to do is arrange for a radiosity
>>savefile to be created once for the whole scene. The partial renderers
>>would then all load the same save file.
>>
>>I can't see that photons would be a problem at all. Photons interact
>>quite happily with objects that are outside the frame, so they should
>>work just as well with objects that are inside the frame but outside the
>>selected region.
>>
> 
> Nice then, thanks for the info. :)
> 
> Knowing this I believe that what I was thinking of is far from impossible.
> If there's demand for such a program, and there is no similar program in
> development right now (maybe the IMP project will give birth to the right
> thing), we could start to recruit programmers for the job :)
> 
> And if we are talking about IMP, can anyone give details about the software
> they are making? How it will work, to what extent it is capable of doing
> what I was thinking of (or what other pov users would expect).
> I find their website confusing.
> 
> ---
> Cheers
> Gergely
> 
> 
> 
>

Post a reply to this message

James Jackson <jam### [at] jamesjacksoninfo> wrote:
[snip]
> Also, I have a little problem with calling POVRay. To calculate values
> for the load sharing, I am performing timed test renders of each row in
> a (scaled width) x 100 version of the scene. If I call POVRay with +SR1
> +ER1, the whole image gets rendered. However, if I use a value greater
> than 1, such as +SR6 +ER6, I only get that row (as required). What is
> going on in the +SR1 case?

(as per reply in comp.graphics.rendering.raytracing) - try using +ER1.1
(+ER1 seems to be interpreted as 100%, so the entire image gets rendered).

Post a reply to this message

Thanks very much - works a treat!

Chris Colefax wrote:

> James Jackson <jam### [at] jamesjacksoninfo> wrote:
> [snip]
> 
>>Also, I have a little problem with calling POVRay. To calculate values
>>for the load sharing, I am performing timed test renders of each row in
>>a (scaled width) x 100 version of the scene. If I call POVRay with +SR1
>>+ER1, the whole image gets rendered. However, if I use a value greater
>>than 1, such as +SR6 +ER6, I only get that row (as required). What is
>>going on in the +SR1 case?
>>
> 
> (as per reply in comp.graphics.rendering.raytracing) - try using +ER1.1
> (+ER1 seems to be interpreted as 100%, so the entire image gets rendered).
> 
> 
> 
>

Post a reply to this message

Chris Colefax <chr### [at] tagpovrayorg> wrote:
: +ER1 seems to be interpreted as 100%

  Could this be a place for a bug report?

-- 
#macro N(D)#if(D>99)cylinder{M()#local D=div(D,104);M().5,2pigment{rgb M()}}
N(D)#end#end#macro M()<mod(D,13)-6mod(div(D,13)8)-3,10>#end blob{
N(11117333955)N(4254934330)N(3900569407)N(7382340)N(3358)N(970)}//  - Warp -

Post a reply to this message

I <chr### [at] tagpovrayorg> wrote:
> : +ER1 seems to be interpreted as 100%

Warp <war### [at] tagpovrayorg> wrote:
>   Could this be a place for a bug report?

The documentation explicitly states that row/column values <= 1 will be
interpreted as a percentage of the image, so the behaviour is correct from
that point of view.  This does mean, though, that you can't render the first
line only without resorting to hacks like specifying +ER1.1 or +ER0.

However, the behaviour of rendering the entire image when +SR1 +ER1 is
specified is probably buggy - by the documentation, this should render the
last row of the image, or possibly the first, but not all.

Post a reply to this message