Image luminance range to resemble actual perception

Hi All,

I have a question that perhaps originates from my wrong understanding of Radiance rendered images. I would be grateful, if you take time to read the following and share your ideas with me:

While generating rpict images of a scene without any light source other than sky, I realised that even if the external radiation is very low (for example with a global horizontal radiation of 4 W/m2 at an early morning hour), my scene still looks quite bright.

So, I used wxflasecolor application to make sure that the luminance values were rightly very low. And in fact the application allows me to change the luminance range to make the scene darker or brighter.

Well, this finding surprised me, because I expected that Radiance rendered images resemble the “actual perception” of the space in terms of brightness. But apparently, the rendered images are smilar to a false color mapping, whose range (and brightness) can be changed to get the desirable/informative appearance. Do I understand this correctly? If so, do you have any recommendations as to how one can get closer to the “actual perception” of the space? For example, with which range of luminace? Or is it altogether a wrong expectation from Radiance?

Thank you.
Farhang

Radiance will (can) produce images that are physically accurate, but displays often struggle to convey those images to a human observer in the same way we’d perceive those scenes In Real Life[TM]. So you’re on the right track using falsecolor to understand and validate the rendering by looking at the actual luminance values. I’ve not used wxfalsecolor in a long time but I suspect its operation is similar to the basic falsecolor that comes with Radiance, in that you can adjust the scale used to map colors to values. That does change the appearance of the false color image but it does not change the underlying information.

Further, you can use tools like pcond to adjust (tonemap) the image to appear as it might to a human observer; this DOES modify the pixel values, but this is an attempt to “fake it”. In scenes of very low luminance however, the default operator (with pcond -h) will tend to desaturate the colors and even blur it slightly. This is to again fake the appearance that your rods would present to your brain.

Anyway, yeah don’t confuse the falsecolor mapping scale with the actual pixel values of the raw image coming out of rpict. If the values seem correct, you’re probably on the right track.