Jump to content
UltravioletPhotography

Is an Assignment of Colors to UV-Wavelengths possible?


WiSi-Testpilot

Recommended Posts

WiSi-Testpilot

I took the spectra of a CMOS sensor and extended the curves to the UV range, see image. The DCC1645C sensor was taken as an example, which I found on the Thorlabs website.

The Baader U-Venus filter for example is open for UV light from 320 to 380 nm with a maximum at 350 nm. The overlays of the filter shape and the spectra indicates, that in a first approximation UV light from 380 nm downwards generates blue, while UV light from 320 nm upwards generates red. My UV images have a great amount of blue and red and a little bit of green. This is in accord with the shown spectra.

I think, if correct, this aspect is very interesting, since the UV photos indicate which wavelength is observed.

A spectrum of a CMOS sensor with a Bayer filter that is extended to UV would be interesting.

Best regards,

Wilhelm

post-210-0-76995600-1544479016.jpg

Link to comment

NO DON'T DO THIS! We have endless wars on this topic! Please search the forum first and read the many many posts already on this topic before delving in!

 

Only madness lies here.

Link to comment

There was a group working on this on an Island, food was dropped in from time to time, but no word from them... (humor)

The angle of the lines can't be projected into the unknown UV range like that.

Link to comment

Wilhelm, you really have no idea what a can of worms this question is. Probably hundreds of forum hours have been spent on far more sophisticated efforts to get a handle on it without much agreement or end result. Forum members have even been banned over the resulting arguments.

 

One of the better attempts to grapple with the issue have been Jonathan's experiments here:

http://www.ultraviol...ctral-response/

http://www.ultraviol...n-in-uv-and-ir/

 

which Jim then attempted to use to derive colors here:

http://www.ultraviol...er-meaurements/

 

but the colors were ultimately not a perfect match. While even this effort failed, I think it was going in the right general direction. It is impossible to work backwards, though, from the colors to the wavelengths, except in a very rough and imperfect way that leaves everyone unsatisfied. Like, if you see only blues in the images you take with a certain lens, then you can deduce that the bandpass is probably not very good.

Link to comment
Hee. I think Jonathan’s numbers are probably the best we are likely to get for some time. It would be good to reproduce that experiment, however. Especially with a different camera brand.
Link to comment

I saw this topic just before I went to bed last night, and I though I'd leave it until this morning to respond.

 

This subject always causes some intense discussion. As has been mentioned I went after measuring camera response in the UV, and even built my own equipment to do it. In short, yes, camera response can be measured - I've done it, and a few others have published on it (and I recently submitted a paper on it). It certainly isn't easy, but yes camera response curves are possible. As I have monochrome converted cameras too, I can even use those to back calculate Bayer filter transmission spectra in the UV. I have even tried approaching someone who makes Bayer filters to have them make me a sample of quartz with each of the filters individually on them so I can measure transmission spectra directly. After initial discussions it has all gone quiet, so they are either unable or unwilling to do it.

 

The most up to date plot on the forum of camera response in the UV is here in post #9;

http://www.ultravioletphotography.com/content/index.php/topic/2813-camera-sensitivity-chasing-ghosts-in-spectral-sensitivity-measurements/page__view__findpost__p__21902

 

Can that response be used to predict colours for different wavelengths? Well, this makes the measuring of the camera response look easy.

 

Overall, the general colours you see can tell you something about the wavelengths present. If you get lots of blues in the raw composite file then there is likely to be a lot of the 390-400nm light for instance. Can it be used to say Pixel X is 365nm and Pixel Y is 368nm, no.

 

The threads mentioned above make interesting reads, if you have a glass of wine and want to chill out one evening.

Link to comment

Well lets not discourage anyone from testing. However, lets keep an open mind and be able to accept when we may have made a mistake or a breakthrough.

Generally what I see is less than and equal to 345nm is green. At 313nm its really green, more green than 335nm.

Above that til about 375nm is yellow. I like to think of it as golden. So a mixture of what ever colors make up yellow. Then above that is blue. 380nm to 405nm is blue with mixtures of purple or violent.

So you can stop there or try and tease it out more specifically.

What I want to test next is the color checker passport at all wavelengths that I have band pass filters for. As a follow up to the color chart MATLAB post I posted earlier.

But I think I want a new light. Cadmium may have been correct months ago recommending the 200 UVb 26W light by Exo terra. It looks to have the range I want. The temperature I want and lower IR.

My lucky herp mercury vapour lamp gets really hot, has a ton of IR. But has a really strong 313nm line and 365nm line. So has been useful.

Link to comment

I want to discourage people from jumping into a complex subject without doing the background reading. After that, new work is fine.

 

Dabateman, that color checker is useless in UV. The spectrum of the blacks and grays are not even flat- Jonathan measured those awhile ago.

Link to comment

@Andy,

Well I feel I may need to waste some shutter clicks to see. Also, I want to look at the color side. I am curious if the colors respond differently at different tight bandpasses. The red seems to have nice UVIVF as many here have shown.

 

Link to comment

I want to discourage people from jumping into a complex subject without doing the background reading.

 

With 'poison darts'?! :unsure:

 

I think you should splurge and buy the quart jars of paint, Andy. :)

I was being serious about that idea, if someone had the inclination.

However, it doesn't seem worth it to me.

Link to comment
With 'poison darts'?! :unsure:

Nah, with common sense. :rolleyes:

 

Honestly I'd love to see what tests of the paint or whatever would do, but I don't own a spectroscope and I'm sure those paints cost $$$$.

--

 

dabateman, there are tons of photos of color checkers already on the site? I mean, knock yourself out, but all they show is that the color checkers all reflect blue-ishly, without any yellows or other false colors that would be needed to use that MATLAB software. It is ridiculously hard to find a pigment that reflects shortwave UV. Only in flowers have I seen anything like that, and I almost wonder if it's structural color in their case...

Link to comment

Here is a reflected spectrum of an Australian flower, a Hibbertia

post-28-0-52396800-1544582669.jpg

 

A different Hibbertia exutiaces looks like this in Visible light

post-28-0-17756700-1544582700.jpg

 

and this in UV

post-28-0-69623800-1544582718.jpg

 

Work that one out!

 

Dave

Link to comment

Gorgeous, Dave! And yeah, nothing like that on the color checker. I've considered maybe trying to use plant dyes. You could do the classic freshman chemistry experiment and try some chromatography.

 

To use PCA to extract the bayer+sensor curves, I still think the logical thing would be to just use a sparticle. Trying to reproduce some kind of UV false color checker equivalent is a much harder task.

Link to comment

Yes those look like images of the color checker with filtration using a Baader venus U filter with wavelength range of 325/330nm to 400nm.

Now what does it look like using 300bp10, 313bp25, 335pb10, 340bp10, 370bp15 and 390bp25. These simple 6 images may look different highlight slight different reflections in the color dyes.

Or they may look identical. Otherwise its just 6 images to take and compare.

Link to comment

@Andy,

Partly yes. I am not sure what I will get and wanted to compare lighting.

Depending on the result it may say nothing or something.

To know somthing about a camera, I would need to compare cameras. But first I want to see if there is any difference.

Link to comment

Other than flowers that look yellow, and some paint/pigment/additives that look blue/violet, there are very few materials that show much if any color in reflected UV,

and other than a few odd camera sensor designs, I think most of the cameras use the same basic Bayer dyes, and record reflected UV with almost no differences.

However, the transmission graph of the Bayer in UV extended down to 300nm+ for all three colors is something that remains somewhat undefined.

Link to comment

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...