0
$\begingroup$

In many datasets in the literature of both Computer Vision and Computer Graphics there is the albedo map of a scene represented as RGB images, which in principle corresponds to the diffuse reflecting component. The diffuse reflectance factor though of a material/surface is represented a one value, usually as ρ (rho), e.g. in the radiosity formulation. Thus, I am trying to understand how to go from an RGB representation to the corresponding diffuse value. I saw some people averaging the RGB values of the albedo map, e.g. for a pixel value (0, 168, 170) this would give:

((0/255) + (168/255) + (170/255)) / 3 = 0.441830065359477

however is just averaging enough/correct or there is a specific formulation.

I am puzzled quite some time now about it and I cannot really find a good answer. Thus, I would appreciate any feedback.

$\endgroup$
2
  • $\begingroup$ Radiosity (and any type of rendering formulation) needs to represent diffuse reflectance as a color, whether RGB or spectral, not as a single value for the whole spectrum. That wouldn't be correct; you wouldn't get colored bounce light e.g. in the Cornell Box. $\endgroup$ Sep 15, 2020 at 16:18
  • $\begingroup$ @NathanReed this is not necessary. What you are saying is correct if you work on the color world. But in case that you do not care to represent the radiance/irradiance as color but rather than as energy color does not matter. $\endgroup$
    – ttsesm
    Sep 22, 2020 at 8:00

1 Answer 1

1
$\begingroup$

I am not aware of what LRV is and what is a usage of it. However, I can recommend using luma instead of simply averaging RGB values with equal weight:

Human eyes are more sensitive to green colors [1], that's why 16 bit (aka high color [2]) palettes, might prefer to use R5G6B5 format.

[1] https://en.wikipedia.org/wiki/Color_vision

[2] https://en.wikipedia.org/wiki/High_color

$\endgroup$
1
  • 2
    $\begingroup$ from my research you answer is almost correct the suggested way to do it is first to linearize your image, address the gamma correction and then get the luma. $\endgroup$
    – ttsesm
    Sep 22, 2020 at 8:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.