[Radiance-general] Rendering a single image with multiprocessor in Windows

[Greg Ward]

Good work, Mostapha!

This is really something that should be implemented in a .cal file.  I've just never found the time.

Cheers,
-Greg

> From: Mostapha Sadeghipour <sadeghipour at gmail.com>
> Subject: Re: [Radiance-general] Rendering a single image with multiprocessor in Windows
> Date: August 5, 2014 7:04:57 PM PDT
> 
> Hi Greg and Rob,
> 
> Thanks Greg. I did a test and it works like a charm for perspective views. I haven't tested it for other types of views yet.
> 
> Rob, rpiece is not actually included in the recent Windows packages. I have it there from an old release! Oops... ;)
> 
> Let me know when you are back so we can check the results. It works perfect on my side. Maybe I should test more cases.
> 
> Mostapha
> 
> 
> 
> On Fri, Aug 1, 2014 at 4:52 PM, Rob Guglielmetti <rob.guglielmetti at gmail.com> wrote:
> Hey Mo, I can share my results with you on this when I get back from vacation; I made a spreadsheet to calculate the views based on that code but as I recall I got some funky results. 
> 
> Also thanks for the heads up on rpiece in the windows packages, I'll make sure that's not included in the future! Oops...
> 
> On Jul 31, 2014, at 11:59 PM, Greg Ward <gregoryjward at gmail.com> wrote:
> 
>> It's possible, but not at all straightforward.  You basically have to use the attached code (taken from rpiece) to adjust the -vh and -vv options to rpict, then use the -vs and -vl options to determine which part of the picture you want to render.
>> 
>> Best,
>> -Greg
>> 
>> 	switch (ourview.type) {
>> 	case VT_PER:
>> 		pview.horiz = (2.*180./PI)*atan(
>> 				tan((PI/180./2.)*ourview.horiz)/hmult );
>> 		pview.vert = (2.*180./PI)*atan(
>> 				tan((PI/180./2.)*ourview.vert)/vmult );
>> 		break;
>> 	case VT_PAR:
>> 	case VT_ANG:
>> 		pview.horiz = ourview.horiz / hmult;
>> 		pview.vert = ourview.vert / vmult;
>> 		break;
>> 	case VT_CYL:
>> 		pview.horiz = ourview.horiz / hmult;
>> 		pview.vert = (2.*180./PI)*atan(
>> 				tan((PI/180./2.)*ourview.vert)/vmult );
>> 		break;
>> 	case VT_HEM:
>> 		pview.horiz = (2.*180./PI)*asin(
>> 				sin((PI/180./2.)*ourview.horiz)/hmult );
>> 		pview.vert = (2.*180./PI)*asin(
>> 				sin((PI/180./2.)*ourview.vert)/vmult );
>> 		break;
>> 	case VT_PLS:
>> 		pview.horiz = sin((PI/180./2.)*ourview.horiz) /
>> 				(1.0 + cos((PI/180./2.)*ourview.horiz)) / hmult;
>> 		pview.horiz *= pview.horiz;
>> 		pview.horiz = (2.*180./PI)*acos((1. - pview.horiz) /
>> 						(1. + pview.horiz));
>> 		pview.vert = sin((PI/180./2.)*ourview.vert) /
>> 				(1.0 + cos((PI/180./2.)*ourview.vert)) / vmult;
>> 		pview.vert *= pview.vert;
>> 		pview.vert = (2.*180./PI)*acos((1. - pview.vert) /
>> 						(1. + pview.vert));
>> 		break;
>> 	default:
>> 		fprintf(stderr, "%s: unknown view type '-vt%c'\n",
>> 				progname, ourview.type);
>> 		exit(cleanup(1));
>> 	}
>> 
>> 
>>> From: Mostapha Sadeghipour <sadeghipour at gmail.com>
>>> Subject: Re: [Radiance-general] Rendering a single image with multiprocessor in Windows
>>> Date: July 31, 2014 2:43:17 PM PDT
>>> 
>>> I just read that rpiece is not supposed to be part of the Windows package for some reason so I probably should change the question to is it possible to render a segment of a view based on -X -Y pixels?
>>> 
>>> I can take care of running them in parallel and putting them back together separately.
>>> 
>>> Thanks,
>>> Mostapha
>>> 
>>> 
>>> On Thu, Jul 31, 2014 at 4:28 PM, Mostapha Sadeghipour <sadeghipour at gmail.com> wrote:
>>> Hi all,
>>> 
>>> I wonder is anyone has successfully rendered a single image in parallel using rpict -PP and/or rpiece in Windows.
>>> 
>>> I have a hard time to understand both the process and the syntax.
>>> 
>>> Thanks,
>>> Mostapha 
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.radiance-online.org/pipermail/radiance-general/attachments/20140805/de86663a/attachment-0001.html>