Correspondence between the pixel value and the illuminance value


I am calculating the illuminance produced by the reflection of light on a curved surface and for this I have generated a model with as many polygons as possible so that the result is as continuous as possible. I have generated the render with rpict with the option -i and a resolution of 4096px x 4096px as follows:

rpict -i -vf 03-Vista.vf -x 4096 -y 4096 -av .5 .5 .5 02-Escena.oct > 04-IImagen%mes%-%dia%-%hora%SinFiltro.hdr

The falsecolor image is this one:

In addition, I have made a series of images varying the %mes% and %dia%.

Subsequently, I calculated the values of several specific pixels as follows:

pvalue -o 04-IImagen%mes%-%dia%-%hora%SinFiltro.hdr | rcalc -e "eq(a,b):if(a-b+.5,b-a+.5,-1);and(a,b):if(a,b,a)" -e "cond=and(eq($1,1250),eq($2,1650))" -e "$1=$3;$2=$4;$3=$5" >> ValoresRAW-4096-Puntos.txt

Comparing the results obtained with pvalue and the concrete pixel values in the image, the results do not match obtaining the following curves:

That is, the values jump in two curves but the pattern do not match.

Therefore, the doubts that arise to me are the following:

-What is the correspondence between the pixel value in the photo and the illuminance value in the numerical output, i.e. how can I check that the values obtained with pvalue correspond to those in the image? Because as I am looking at it they do not correspond.

-How does Radiance take the averaged value in a pixel when the resolution of the 3D model is higher than the resolution of the image?

I hope you can solve my doubts.

Thank you very much.
Best regards.

Hi Irene,

I’m not sure I can help you with the correspondence between the HDR image and pvalue output. However, you may want to look into the photon mapping extension built into RADIANCE, which is specifically designed for specular reflections from curved surfaces (caustics, see attached piccie). You will need to generate a caustic photon map for this. You can find the manual online here, though it should also be included in the official distribution.

Hope this helps. Good luck & best regards,


P.S. and caveat: If your curved geometry is faceted, the caustic photon map may reveal this as banding artefacts. This is a modelling issue, and not a limitation of the light transport algorithm.

[Crappy ancient rendering from 1st RADIANCE photon map release…]

1 Like

I can’t say for sure, but I suspect your eq(a,b) definition is at fault. Though it works in most cases, the accuracy of color pixels in Radiance is only about 1%, so using a 0.5 margin of error on values over 1000 is problematic. Use the following definition, instead:


which only works for positive values (OK in your case).


Thank you both for your comments. I’ll try it.