[Radiance-general] slightly higher irradiance values for -b 1 option

[Greg Ward]

Hi Ali,

I agree that this is an interesting and unexpected result.  I've looked over the code, and I suspect the cause is having so many ambient samples for a small starting radiance is causing the ambient super-sampling code to hit the 1e-7 (FTINY) threshold for ignoring differences between adjacent ambient divisions.

What happens to your series when you use -as 0.  I would expect the errors to go up for all brightnesses, but hopefully you will at least get the same errors for all gensky -b settings at that point.  This would confirm my suspicion, and allow me to implement a fix.

Cheers,
-Greg

> From: Christopher Rush <Christopher.Rush at arup.com>
> Date: April 21, 2017 7:46:08 AM PDT
> 
> I don’t think I’ll have the answer to your question, but to be sure we all understand… although you get more consistent values between –b 10, 100, and 1000, when you increase –ad they all converge toward the trend you would expect from –b 1?
>  
> -Chris
>  
> From: Ali Fatoorechi [mailto:ali at surveymbs.com] 
> Sent: Friday, April 21, 2017 10:15 AM
>  
> Yes the posted figures are sort of avg of multiple runs. 
>  
> I noticed for -b 10, 100,... options the -ad has to be increased to some x4 more to get same accuracy as -b 1. 
> But I am not sure if this is expected in Radiance.
> 
> Regards,
>  
> Ali Fatoorechi
> MBS Survey Software Ltd
>  
>  
> On Fri, Apr 21, 2017 at 2:38 PM, Christopher Rush <Christopher.Rush at arup.com> wrote:
> Ali,
> If you run the same commands multiple times, do you always get similar results? Does -b 1 always give proportionally 2% higher value than the higher orders of magnitude if you repeat the process multiple times, or does it vary?
>  
> If you change the –ad or –lw parameters do you get any different degree of consistency?
>  
> -Chris
>  
> From: Ali Fatoorechi [mailto:ali at surveymbs.com] 
> Sent: Friday, April 21, 2017 7:57 AM
>  
> Hello guys,
>  
> I am a bit struggling to make sense of numbers I get. I might have done something wrong.
> The rtrace gives roughly same irradiance figures(apart from the exponent component) for overcast sky generated with -b 10, -b 100, -b 1000, 
> or any other power of 10, while -b 1 gives slightly higher value.
>  
> rtrace -w -h -I -ab 12 -ad 10000 0 -as 4096 -aa 0 -ar 0 -lw 1E-05 room.oct < pnt.txt 
>  
> !gensky 3 21 10 -c -a 51 -o 0 -m 0 -b 1
>  
> -b 1:       9.245390e-002   9.245390e-002   9.245390e-002
> -b 10:     9.045287e-001   9.045287e-001   9.045287e-001
> -b 100:   9.023538e+000   9.023538e+000   9.023538e+000
> -b 1000: 9.039278e+001   9.039278e+001   9.039278e+001
>  
> This is a simple room with an opening and no glazing. 
>  
> Thanks,
> Ali
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.radiance-online.org/pipermail/radiance-general/attachments/20170421/5978adf1/attachment.html>