# Conversion of RGB irradiance values to a illuminance value

There must be some kind of misunderstanding.

RAW image (or HDR image made using a calibrated response curve) pixel values are, up to a global image scaling factor, radiance values measured in W/(m^2 * sr), in the each of the three bands. If my pixels were luminance values I wouldn't have the problem of converting RGB to brightness as everything would already by weighted with the human response/sensitivity function. Luminance can't be separated into frequency bands, it's a weighted spectral integration over the entire frequency domain. If not, I am totally baffled

Irradiance is in W/m^2 and I get that from hemispherical cosine weighted integration.

I'm pretty sure we are somehow talking past each other, though.

- an iPad mail

···

On 27/03/2014, at 17.48, "Giovanni Betti" <gbetti@FosterandPartners.com<mailto:gbetti@FosterandPartners.com>> wrote:

The values of your images are bound to be luminance (cd/m2) not radiance (W/m2)

The standard radiance conversion from rgb to luminance (brightness) is 179*(R *.265 + G*.67 + B*.065)
Where 179is the standard luminous efficacy assumed by radiance.

However, evalglare will take care of all this for you.

Best,

G

////////////////////////////////////////////////
Giovanni Betti
From: Claus Brøndgaard Madsen [mailto:cbm@create.aau.dk]
Sent: 27 March 2014 16:19
To: Radiance general discussion
Subject: Re: [Radiance-general] Conversion of RGB irradiance values to a illuminance value

Giovanni,

Thanks. You're right. My camera is band-limited as yours

But then again. I don't think the light meter measures into IR nor UV, either.

The values of my HDR images are true radiance values, but for the three relatively narrow rgb bands respectively.

So, brightness values you say. But which of the many ways of converting from RGB to brightness do I choose? Which weighting if the color channels?

Best

- an iPhone mail

On 27/03/2014, at 16.23, "Giovanni Betti" <gbetti@FosterandPartners.com<mailto:gbetti@FosterandPartners.com>> wrote:
Claus,

Let me see if I can help here.
I am not sure what camera set up you have but I see it hard for you to be capturing a radiance map with your HDR images.
That would mean that your camera can record from the IR to the UV, well beyond the visible light only.
If that’s the case, I want the same camera!

Otherwise, if you have a normal SLR fitted with a Sigma 180 fisheye, you’ll likely be recording luminance maps. Integration of the brightness values (forget about colours) over the hemisphere will give you your illuminace reading at the focal point of the camera, to be compared with the reading from the lux meter.

Evalglare has a handy option to do this integration for you.

Hope this helps,

Giovanni

////////////////////////////////////////////////
Giovanni Betti
From: Claus Brøndgaard Madsen [mailto:cbm@create.aau.dk]
Sent: 27 March 2014 15:00
To: Radiance general discussion
Subject: Re: [Radiance-general] Conversion of RGB irradiance values to a illuminance value

Thanks Christopher,

My situation is a little different, I think.

I point to SIGMA 180 degree field-of-view lens vertically upwards and make a full hemisphere HDR image (radiance map) from that. I then integrate to get the vertical irradiance in each of the three color channels. These values are calibrated so they are in W/m^2.

Simultaneously, I place a light meter flat (sensor pointing vertically upwards) and take a reading. This reading is a single value in Lux.

So, how do I convert the three irradiances to a Lux value? Spectral integration, yes, but how exactly?

Cheers

From: Christopher Rush [mailto:Christopher.Rush@arup.com]
Sent: 27. marts 2014 15:45
To: Radiance general discussion
Subject: Re: [Radiance-general] Conversion of RGB irradiance values to a illuminance value

If I understand your situation correctly, I believe you would need to know the reflectance of the material at the point of the irradiance measurement to backtrack to illuminance for calibration. This works when the surface is Lambertian (diffuse), and much more difficult if there’s any glossiness. For calibration purposes, people often include a photographic grey card of known reflectance in their scene and take actual luminance or illuminance meter measurements at the grey card.

From: Claus Brøndgaard Madsen [mailto:cbm@create.aau.dk]

I use hemi-spherical HDR images to compute a vertical irradiance value (W/m^2) for each color channel.

If I wish to compare/calibrate these numbers to a reading from a light meter (giving illuminance in Lux) how do I go about converting the three (RGB) irradiances to illuminance?

Cheers,
Claus

____________________________________________________________
Electronic mail messages entering and leaving Arup business
systems are scanned for acceptability of content and viruses
_______________________________________________
_______________________________________________

I think you should try Giovanni suggestion. Meaning not to listen to but really trying to do what he is suggesting.
So, start by converting image to HDR.
HDR has 3 channels that contribute to luminance as Giovanni is saying.
The integral of the image (you can write your own script to do that, see IESNA handbook) will give you illuminance.
The same way a picture in radiance can give you illuminance. Just a conversion factor.
If the numbers are far off there is an issue with luminance calibration or simply with wignetting of the lens.
We have been doing this for years and it works… but we use posh Leicas to do so (we edit the exif file as lenses do not record f stops...).

···

On 27 Mar 2014, at 17:31, Claus Brøndgaard Madsen <cbm@create.aau.dk> wrote:

There must be some kind of misunderstanding.

RAW image (or HDR image made using a calibrated response curve) pixel values are, up to a global image scaling factor, radiance values measured in W/(m^2 * sr), in the each of the three bands. If my pixels were luminance values I wouldn't have the problem of converting RGB to brightness as everything would already by weighted with the human response/sensitivity function. Luminance can't be separated into frequency bands, it's a weighted spectral integration over the entire frequency domain. If not, I am totally baffled

Irradiance is in W/m^2 and I get that from hemispherical cosine weighted integration.

I'm pretty sure we are somehow talking past each other, though.

- an iPad mail

On 27/03/2014, at 17.48, "Giovanni Betti" <gbetti@FosterandPartners.com> wrote:

The values of your images are bound to be luminance (cd/m2) not radiance (W/m2)

The standard radiance conversion from rgb to luminance (brightness) is 179*(R *.265 + G*.67 + B*.065)
Where 179is the standard luminous efficacy assumed by radiance.

However, evalglare will take care of all this for you.

Best,

G

////////////////////////////////////////////////
Giovanni Betti
From: Claus Brøndgaard Madsen [mailto:cbm@create.aau.dk]
Sent: 27 March 2014 16:19
To: Radiance general discussion
Subject: Re: [Radiance-general] Conversion of RGB irradiance values to a illuminance value

Giovanni,

Thanks. You're right. My camera is band-limited as yours

But then again. I don't think the light meter measures into IR nor UV, either.

The values of my HDR images are true radiance values, but for the three relatively narrow rgb bands respectively.

So, brightness values you say. But which of the many ways of converting from RGB to brightness do I choose? Which weighting if the color channels?

Best

- an iPhone mail

On 27/03/2014, at 16.23, "Giovanni Betti" <gbetti@FosterandPartners.com> wrote:

Claus,

Let me see if I can help here.
I am not sure what camera set up you have but I see it hard for you to be capturing a radiance map with your HDR images.
That would mean that your camera can record from the IR to the UV, well beyond the visible light only.
If that’s the case, I want the same camera!

Otherwise, if you have a normal SLR fitted with a Sigma 180 fisheye, you’ll likely be recording luminance maps. Integration of the brightness values (forget about colours) over the hemisphere will give you your illuminace reading at the focal point of the camera, to be compared with the reading from the lux meter.

Evalglare has a handy option to do this integration for you.

Hope this helps,

Giovanni

////////////////////////////////////////////////
Giovanni Betti
From: Claus Brøndgaard Madsen [mailto:cbm@create.aau.dk]
Sent: 27 March 2014 15:00
To: Radiance general discussion
Subject: Re: [Radiance-general] Conversion of RGB irradiance values to a illuminance value

Thanks Christopher,

My situation is a little different, I think.

I point to SIGMA 180 degree field-of-view lens vertically upwards and make a full hemisphere HDR image (radiance map) from that. I then integrate to get the vertical irradiance in each of the three color channels. These values are calibrated so they are in W/m^2.

Simultaneously, I place a light meter flat (sensor pointing vertically upwards) and take a reading. This reading is a single value in Lux.

So, how do I convert the three irradiances to a Lux value? Spectral integration, yes, but how exactly?

Cheers

From: Christopher Rush [mailto:Christopher.Rush@arup.com]
Sent: 27. marts 2014 15:45
To: Radiance general discussion
Subject: Re: [Radiance-general] Conversion of RGB irradiance values to a illuminance value

If I understand your situation correctly, I believe you would need to know the reflectance of the material at the point of the irradiance measurement to backtrack to illuminance for calibration. This works when the surface is Lambertian (diffuse), and much more difficult if there’s any glossiness. For calibration purposes, people often include a photographic grey card of known reflectance in their scene and take actual luminance or illuminance meter measurements at the grey card.

From: Claus Brøndgaard Madsen [mailto:cbm@create.aau.dk]

I use hemi-spherical HDR images to compute a vertical irradiance value (W/m^2) for each color channel.

If I wish to compare/calibrate these numbers to a reading from a light meter (giving illuminance in Lux) how do I go about converting the three (RGB) irradiances to illuminance?

Cheers,
Claus

____________________________________________________________
Electronic mail messages entering and leaving Arup business
systems are scanned for acceptability of content and viruses

_______________________________________________