Is it possible to use a DSLR as a cost effective spectrometer?

[Anand Sivaram]

TL;DR

Is it possible to use a DSLR as a cost effective spectrometer?

Is it possible to use a DSLR as a cost effective spectrometer?

To find out:

• The wavelength of different monochrome LED lights.
• The color temperature of different white lights

To elaborate, take a DSLR and keep the white balance to fixed 6500K to get D65 white point. Take the photograph of the monochrome LED light source or the white light. Now, find the RGB pixel values using photoshop or similar and calculate the wavelength or color temperature.

Basically the reverse of what is given below.
https://academo.org/demos/wavelength-to-colour-relationship/
https://academo.org/demos/colour-temperature-relationship/

Is it practically doable?

 

[Charles Link]

The spectrum from a diffraction grating spectrometer provides much more spectral information than what you can get with a silicon photodiode and 3 simple bandpass filters. The results are interesting, but no, the DSLR cannot replace the results of a spectrometer.

 

[pinball1970]

Is it possible to use a DSLR as a cost effective spectrometer?

To find out:

• The wavelength of different monochrome LED lights.
• The color temperature of different white lights

To elaborate, take a DSLR and keep the white balance to fixed 6500K to get D65 white point. Take the photograph of the monochrome LED light source or the white light. Now, find the RGB pixel values using photoshop or similar and calculate the wavelength or color temperature.

Basically the reverse of what is given below.
https://academo.org/demos/wavelength-to-colour-relationship/
https://academo.org/demos/colour-temperature-relationship/

Is it practically doable?

I've been messing around with something similar using a spectromaster but these are about £1500. I am not sure how accurate these camera readings are.
CCT may be ok but not sure about anything else. Sophicenture is expert at this

 

[Andy Resnick]

Is it possible to use a DSLR as a cost effective spectrometer?

Is it practically doable?

Yes, but not the way you describe for several reasons. OTOH, if you use a dispersive element (like a transmission grating) in the imaging path, you can use the camera as an imaging spectrometer and determine, for example, emission wavelengths of fluorescent lights (Hg), and streetlamps (Na).

 

[sophiecentaur]

Now, find the RGB pixel values using photoshop or similar and calculate the wavelength or color temperature.

A DSLR colour analysis works in a similar way to our eyes (it's designed that way). It is not a spectrometer and all it will do is tell you what COLOUR the light is, on the basis of the outputs from the three wideband colour sensors. It cannot tell you the wavelength unless you know that you have monochromatic light. The RGB values can tell you roughly where a monochromatic source sits on the spectral curve on a CIE Chromaticity Diagram, relative to the white point that the camera uses. If the spectrum of the light is not monochromatic then the RGB values can't give any more than a Colour Match.
The CIE diagram shows the chromaticity (x,y) co ordinates for approximate colours (depending on your monitor etc etc). The DSLR analysis does not attempt to produce accurate representation of spectral colours because three display primaries can only create colours within the 'triangle' of those primaries. This is not a five minute job so I suggest you read around about how the RGB values in any colour system relate to the XY co ordinates. You could perhaps get somewhere if you had some LEDs with known wavelengths to calibrate with.
You can see the line that different temperature White Points sit on so your RGB values could give you an idea of the colour temperature of your white LEDs.
The method that @Andy Resnick suggests would give you a much better result but it would involve using what is effectively a home made spectrometer. That would require quite a bit of skill and ingenuity, I think.

Sophicenture is expert at this

I wish - but thanks anyway. All I know is that it's a lot lot harder than most people think.

 

[Andy Resnick]

The method that @Andy Resnick suggests would give you a much better result but it would involve using what is effectively a home made spectrometer. That would require quite a bit of skill and ingenuity, I think.

It's really not difficult:

Left is an incandescent bulb, right are compact fluorescent bulbs. The grating is taped to the rear lens element.

 

[sophiecentaur]

@Andy Resnick I agree that example is not difficult. I was thinking in terms of something with a bit sharper resolution. The tiny hand held spectroscope I bought on eBay (for gemologist mainly) gives a great visual display. I haven’t tried to photograph the display but perhaps I should. Not sure what lens to use. Perhaps my iPhone would do better than the dslr.

 

[Anand Sivaram]

Thank you very much all. Gained a lot of information

 

[pinball1970]

Hi Anand

Just one last thing

Hi-jack a colourist/photographer and have a fiddle with his spectro if you know anybody or hire one from the uni?

A quick reading with a spectromaster on some 4000K LEDs gave me this data

You don't get chromaticity co-ordinates but you do get the SPD, RA, CCT and lx if that is important to you.

Good luck

 

[Anand Sivaram]

Hi Anand

Just one last thing

Hi-jack a colourist/photographer and have a fiddle with his spectro if you know anybody or hire one from the uni?

A quick reading with a spectromaster on some 4000K LEDs gave me this data

You don't get chromaticity co-ordinates but you do get the SPD, RA, CCT and lx if that is important to you.

Good luck

View attachment 242485View attachment 242486

Thank You