Irradiance Contour Lines Calibration Problem

[Jan]

I'm trying to check the Irradiance Contour Lines will make on the XY plane (10mx10m) when illuminated by a 1 Candela Point Light source at 1 meter distance and perpendicular to center of the plane.

My understanding or assumptions are as follows:

* Luxcore Luminous Efficiency is 179 lm/W
* The Lux value of the illuminated XY plane at center and at the zenith from the Point Light source of 1 candela (1 lumen/steradian) and at 1 meter distance or radius is calculated to be 1 lux (1 lumen/sq. meter).


Under Irradiance Contour Lines settings are the following:

*Scale: ????
*Range: 1 assuming the lux range on the illuminated XY plane is to be from maximum of 1 lux (inside the red contour) at its center to minimum 0 lux (outside the purple contour region) at infinite edge of the XY plane
*Steps: 20

I tried to use the default Scale of 179 but I cannot get the red contour line nearest the center of the plane even with Steps set at 20.

Are there something wrong with the above assumptions? What is the correct Scale setting and what value to set there to correspond to the actual lux values in the illuminated XY plane?

Thanks

[Jan]

This is what I have found.

For the red and innermost contour to be exactly at the center and which correspond to 1 lux the Scale has to be set at 3.14159 or exactly the value of PI.

This is the math function:

z*Tan[ArcCos[lux^(1/3) z^(2/3) where lux = lux + 1/19 until lux = 1; z = 1 the distance from point source and perpendicular to the center of the plane.


These are the contour radii in measurements using the above math function:

{ComplexInfinity, 2.47394, 1.86696, 1.55664, 1.35119, 1.19797, \
1.07537, 0.972535, 0.883228, 0.803531, 0.730778, 0.66302, 0.598719, \
0.536539, 0.475177, 0.413145, 0.348407, 0.277433, 0.191578, 0.}


Plotted and then overlay and compared to LuxRender Irradiance Contour

radiosity5.blend

(830.7 KiB) Downloaded 136 times

Is this correct? Why scaled to 3.14159..... or Pi?

[lighting_freak]

Hello,

the right photometry for that is:
A point light source emits an intensity of 1 cd into the full space. The solid angle for full space is defined as 4*Pi.
So this 1cd (lm/sr) multiplied with the solid angle of 4Pi is the amount of luminous flux that is emitted. Its around 12.57 lm (lumen). Now just devide it by squared distance to get the lux value. So in your case its 1m and the result is 12.57 lux.

This is the amount of light that reaches the target surface. It's the illuminance... but the camera can't see illuminance. It sees only the amount of light that is reflected from that target surface and from that light only that portion that is reflected back into the camera. This is called luminance [common unit is cd/m^2].

You can see that easily on that fact. A real illuminance wouldn't change by changing the color of the target surface. But the luxcore implementation does... something is not really accurate handled with this feature.

I wish that developers will take care this one day. A burial luminance camera that could handle real world values would be huge improvement and also an USP for luxcore.

With best regards

[Jan]

the right photometry for that is:
A point light source emits an intensity of 1 cd into the full space. The solid angle for full space is defined as 4*Pi.
So this 1cd (lm/sr) multiplied with the solid angle of 4Pi is the amount of that is emitted. Its around 12.57 lm (lumen). Now just devide it by squared distance to get the lux value. So in your case its 1m and the result is 12.57 lux.

Hello too. But 4Pi*r^2 is area of a sphere and 4Pi is area of a unit sphere therefore 12.57 lumen is the total luminous flux on the unit sphere with total area of 12.57 sq meter. We are projecting this luminous flux onto the plane. 1 cd is 1 lm/1 sr. Also 1 sr is equal to 1 sq meter for a unit sphere or 1 lux. If an infinitesimal portion of this plane subtend this infinitesimal area of a sphere and perpendicularly we should also get 1 lux on the plane.

This is the amount of light that reaches the target surface. It's the illuminance... but the camera can't see illuminance. It sees only the amount of light that is reflected from that target surface and from that light only that portion that is reflected back into the camera. This is called luminance [common unit is cd/m^2].

I concur and this what makes LuxCoreRender unique. It has additional illuminance layer beside reflected luminance. Also light source can also be luminance for Area Lights

You can see that easily on that fact. A real illuminance wouldn't change by changing the color of the target surface. But the luxcore implementation does... something is not really accurate handled with this feature.

Luxcore also has a unique feature called Normalize by Color Luminance. The Irradiance Contour Lines is consistent despite adjusting the hue or saturation if this option is enabled. There are also some questions or problems also with the IES lights too which I would like to raise. Thank you for the input.

[lighting_freak]

Hi,

My bad, I was wrong. Sorry...
Yes the illuminance should be 1lx in 1m distance (in your case only in plane's centre point). Of course the area of the unit sphere is this 4pi m^2...

4pi lm/4pi m^2 = 1 lx

I'm still not convinced that the illuminance calculated in luxcore right, but I'll run some tests on that.

Let's assume the illuminance is already implemented, would it be possible to get another option for luminance with its contour lines?

BR

[Jan]

Hi,

My bad, I was wrong. Sorry...
Yes the illuminance should be 1lx in 1m distance (in your case only in plane's centre point). Of course the area of the unit sphere is this 4pi m^2...

4pi lm/4pi m^2 = 1 lx

I'm still not convinced that the illuminance calculated in luxcore right, but I'll run some tests on that.

Let's assume the illuminance is already implemented, would it be possible to get another option for luminance with its contour lines?

BR

For luminances of surfaces I think it has to be pixel-based because of BSDF. Aside from different surface reflectances reflected luminances also varies with every single pixel depending on the angle of the light ray it sees as well the angle it is viewed. Smooth contour probably may not be possible for majority of cases unless maybe the surface is Lambertian.