Converting Illumination Output from Nits to W/m² per Channel

[liranhareli]

Hello everyone,

I’m working with the PtDirectIllumination annotator in Omniverse Simulation and need some clarification on how to interpret and convert its output.

Context:

• Annotator Output: The PtDirectIllumination annotator provides per-channel (RGB) values in units of nits (cd/m²). (Source: Obtaining Accurate Illuminance Values in IRay/RTX Accurate)
• Measurement Location: The measurements are taken on the sensor, which suggests they should represent irradiance (W/m²) rather than luminance (cd/m²), as nits are units of luminance.
• Observation with White Light: When using a white light source, I notice that the annotator outputs equal RGB values. This seems counterintuitive because if the values are in nits (which are weighted by the human eye’s sensitivity), I would expect different values for each channel due to the photopic luminous efficiency function V(λ).

Questions:

1. Unit Clarification: Does the PtDirectIllumination annotator output represent luminance (cd/m²) or irradiance (W/m²) per channel? If it represents irradiance, why are the units given in nits?
2. Conversion Method: How can I accurately convert the per-channel output from nits to W/m² per channel, considering the observations above?
3. Equal RGB Values Explanation: Why does the annotator output equal RGB values with a white light source? Shouldn’t the values be different due to the varying sensitivity of the human eye across different wavelengths?

Thank you!

[Richard3D]

1. we output luminance per channel in nits (prior to tone mapping)
2. to convert something to irradiance from nits for things in focus (or pinhole model)
   E = pixel_area / focal_distance^2 * L
3. this question ties into the first, we don’t provide response curves, and in turn the material isn’t authored per-wavelength. This goes more into spectral rendering territory

[liranhareli]

@Richard3D

1. we output luminance per channel in nits (prior to tone mapping)
2. to convert something to irradiance from nits for things in focus (or pinhole model)
   E = pixel_area / focal_distance^2 * L
3. this question ties into the first, we don’t provide response curves, and in turn the material isn’t authored per-wavelength. This goes more into spectral rendering territory

Thank you!

I have a question about unit consistency:
Since L is luminance per channel in nits (cd/m²) and E is irradiance in W/m², should we include the luminous efficacy factor of 683 lm/W to convert luminance to radiance before applying the formula?
This adjustment would account for the difference between photometric and radiometric units.

[Richard3D]

Here is some updated formulas

Exposure is calculated as H = 1.3361 * cm2_factor * film_iso / (100 * shutter * fstop * fstop)

So if you want to calculate irradiance at the sensor from pixel values, you would first take the sRGB luminance of the pixel, L = 0.2126 * R + 0.7152 * G + 0.0722 * B , which corresponds to H, from which you get E_v = H * shutter / 1.3361 and then E_e = 683 . E_v

[liranhareli]

@Richard3D

Here is some updated formulas

Exposure is calculated as H = 1.3361 * cm2_factor * film_iso / (100 * shutter * fstop * fstop)

So if you want to calculate irradiance at the sensor from pixel values, you would first take the sRGB luminance of the pixel, L = 0.2126 * R + 0.7152 * G + 0.0722 * B , which corresponds to H, from which you get E_v = H * shutter / 1.3361 and then E_e = 683 . E_v

1. in H, I think there is a typo, and it should be multiplied by L
2. In the formula “E_e = 683 . E_v”, do you mean E_e = E_v / 683?
3. what do you mean in “cm2_factor”?

Thanks

[Richard3D]

1. Yes
2. Yes
3. cm^2 factor is a multiplier of the exposure. It is by default left at 1.