Which pixels (colors) are sensitive to which wavelengths? What does the WB do? An experiment.

[Kai]

 

Note: Several posts from JMC deal with the same or very similar questions
(e.g. https://www.ultravioletphotography.com/content/index.php?/topic/2813-camera-sensitivity-chasing-ghosts-in-spectral-sensitivity
 

 

In the following I show an investigation with sunlight, not with a monochromator. This means that the result is not so universal, but the applicability to UV photos in sunlight is quite useful - I think.

I recorded the solar spectrum at noon when the sky was clear.
For the original RAW recording, I carried out a WB against PTFE inside the camera.
On the left side, the spectrum is mainly limited by the presence (absence?) of light. The lens transmits artificial light down to almost 320 nm.

On the right side, the light is limited by the filter. Here it was a DUG11x from Schott (made available to me by ITOS 2015, which was selling Schott glasses at the time).

In the RAW converter (DxO PhotoLab 3) I developed two jpgs from it. All corrections are switched off, I just adjusted the brightness so that no channel comes over 255. I chose the generic setting together with the color mode "neutral, realistic tonality, gamma 2.2" as the color profile. This gives me the yellow to blue tones shown here in general against PTFE.

1. WB on Sunlight
The picture appears in brown, pink and purple tones.

 

2. WB according to PTFE measurement
Here you can see the usual yellow and blue tones

 

If you plot the R, - G and B values as a function of the wavelength in the spectra, you get the following distributions (left side "WB sunlight", right side "WB PTFE").
The Fraunhofer lines are a bit annoying here, the curves look "unsmoothed" but the most important things can also be seen.

 

 

 

You can see very well in the "WB sunlight" graphs that the red pixels have the highest sensitivity in the entire spectral range. The green sensitivity is much lower and the green pixels only react again from approx. 380 nm. The blue pixels only react to light up to approx. 365 nm and are also significantly less sensitive than the red pixels.
Below 365 nm, only red and green can be mixed and there are orange and brown tones.

In the range from 365 to 380 nm, all three types of pixels react to the light. Achromatic mixed colors are possible here.

Above 380 nm, only mixed colors of blue and red are possible.

 

If you now carry out a WB to PTFE, then above all red is attenuated and green (and blue) is increased. The typical false colors can now be created.

At around 370 nm, all three intensities are the same. So here an area is shown achromatic. This is the transition area from yellow to a not fully saturated blue-violet. 

 

 

[Fandyus]

Someone should pull up that chart of the actual UV passing abilities of Bayer array filters so that we could compare.

[Andy Perrin]

Fandyus, you can't, that's the thing about this, the solar spectrum isn't evenly distributed, and this is taken through a Baader so you have the Baader spectrum in there also. Kai knows all that, that's why he's giving it as an indication of the practical results you would get in sunlight with a Baader on his Canon, not claiming he found the UV passing ability of the Bayer array. If you would like something you could compare to the Bayer graph on the other page, look at Jonathan's results, which ARE corrected for all that. (To be clear, there is nothing wrong with any of the results here or with Jonathan's, but you should interpret them with care because they are not the same.)

[colinbm]

Beautiful Kai, fantastic work, thanks for sharing.

[photoni]

KAI...

I wonder what is the point of doing the white balance with PTFE in camera if you shoot in RAW.
If I overlap and balance the curves with photoshop ... they are almost the same.

the two raws should contain the same data but with different temperatures and hues

 

[ulf]

When looking at these kinds of curves the local peak and valley features are the lights fingerprint, strictly related to wavelengths.

The intensity drops are related to specific wavelengths where a specific colour  is more attenuated by a specific atom structure in the atmosphere.

 

In your overlap, especially for the R-channel there is a rather big shift between wavelengths. The general intensity shape is likely affected by the profiling and not really essential here.

[Kai]

11 hours ago, Andy Perrin said:

Fandyus, you can't, that's the thing about this, the solar spectrum isn't evenly distributed, and this is taken through a Baader so you have the Baader spectrum in there also. Kai knows all that, that's why he's giving it as an indication of the practical results you would get in sunlight with a Baader on his Canon, not claiming he found the UV passing ability of the Bayer array. If you would like something you could compare to the Bayer graph on the other page, look at Jonathan's results, which ARE corrected for all that. (To be clear, there is nothing wrong with any of the results here or with Jonathan's, but you should interpret them with care because they are not the same.)

Thanks for the clarification, Andy ;) 

11 hours ago, colinbm said:

Beautiful Kai, fantastic work, thanks for sharing.

Thank you for your kind comment, Colin :)

[Andy Perrin]

Photini, there is never a need to do anything IN CAMERA if you have the RAW and also snap a photo of the PTFE standard to white balance from later. Many cameras cannot even do the white balance inside the camera anyway. But you need a photo of the PTFE if you want to be at all standardized even within your own shots. 

[photoni]

9 hours ago, ulf said:

When looking at these kinds of curves the local peak and valley features are the lights fingerprint, strictly related to wavelengths.

The intensity drops are related to specific wavelengths where a specific colour  is more attenuated by a specific atom structure in the atmosphere.

 

In your overlap, especially for the R-channel there is a rather big shift between wavelengths. The general intensity shape is likely affected by the profiling and not really essential here.

 

I honestly don't understand what you're saying

 

2 hours ago, Andy Perrin said:

Photini, there is never a need to do anything IN CAMERA if you have the RAW and also snap a photo of the PTFE standard to white balance from later. Many cameras cannot even do the white balance inside the camera anyway. But you need a photo of the PTFE if you want to be at all standardized even within your own shots. 

 

You say one thing right is logical, but impractical.

I don't walk in the woods, or by the sea, with a piece of PTFE to calibrate every photo taken in the shade or in the sun.

Yes, I did it for 25 years in the studio for some difficult sets... with light, camera, lens, an aperture ... constant, but it didn't give perfect results, just a slight variation in the brightness of the subject is enough to change it.

I do not trust the internal balance with the camera, it is factory calibrated for the VIS, it is questionable with UV and IR.

For software balancing ... it depends what you use

with Photoshop you have tight parameters but luckily you can edit individual channels

Nikon ND-X I tried it, it's very flexible but I don't know it well.

DxO PhotoLab they say is good but has elementary functions for my work

i prefer Capture One, has limitations, but for me it works best with UV and IR

[Andy Perrin]

I’ve been walking around with a piece of PTFE in my camera bag for years and it’s very easy to shoot one photo of it. Editing individual channels will give different results every time and is much harder. You also only need one photo of the PTFE per day. Then you just transfer the white balance to all photos shot that day. Even if there is a difference between shade and sun…you never take UV photos in the shade. (At least I do not. I have no flash.)
 

Capture One is fine. 

[dabateman]

Photoni,

Your alignment is bad. You need to match up exactly the peaks and valleys.  Your red curve for the one on the top needs to be shifted to thd right alot. Then when you look at the peaks you will see big differences. 

Even little difference add up to major color differences.  Just look at the filter catalog,  you will see slight differences in spectrum leading to big changes in color. 

[Fandyus]

On 10/28/2021 at 6:24 PM, Andy Perrin said:

Photini, there is never a need to do anything IN CAMERA if you have the RAW and also snap a photo of the PTFE standard to white balance from later. Many cameras cannot even do the white balance inside the camera anyway. But you need a photo of the PTFE if you want to be at all standardized even within your own shots. 

Not entirely true, setting proper white balance will help the camera determine correct exposure values/prevent specific color channels from being pushed too far.

[Andy Perrin]

Fandyus, it doesn’t work that way with RAWs assuming you capture in manual. (I never use anything but manual for UV work.) The white balance is just stored as metadata and the RAW renderer (eg Capture One) does the actual computation of the pixel colors. It does affect histogram though so it’s recommended to set the camera to monochrome if you can’t use in camera WB. 

[Fandyus]

On 10/30/2021 at 5:03 AM, Andy Perrin said:

Fandyus, it doesn’t work that way with RAWs assuming you capture in manual. (I never use anything but manual for UV work.) The white balance is just stored as metadata and the RAW renderer (eg Capture One) does the actual computation of the pixel colors. It does affect histogram though so it’s recommended to set the camera to monochrome if you can’t use in camera WB. 

If you use full manual then it for sure doesn't matter. If you use aperture priority like me, it is benefitial.

[photoni]

On 10/28/2021 at 11:09 PM, dabateman said:

Photoni,

Your alignment is bad.

it is a quick and unpretentious assembly
you have to ask KAI if using the same settings the two raws are the same

 

On 10/30/2021 at 5:03 AM, Andy Perrin said:

Fandyus, it doesn’t work that way with RAWs assuming you capture in manual. (I never use anything but manual for UV work.) The white balaso it’s recommended to set the camera to monochrome

I also see the shot in BW

 

I'm lazy :)

I always use RAW and auto iris priority

often Auto-ISO Automatic white balance

if I have time ... maximum allowed ° K for UV and minimum ° K for IR

 

If I like the photo, after the first shot I check the histogram of the 3 channels and take a second or third more or less sharp photo.