Pages: [1]   Go Down

Author Topic: IR photography: Sigma SD14 (Foveon) without IR filter only produces B&W images  (Read 534 times)

Guillermo Luijk

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1879
    • http://www.guillermoluijk.com

This goes to IR photography experts in the forum: a member of another forum had his Sigma SD14's (Foveon) IR filter removed. Later he bought a 720nm IR filter to do some IR photography (gray foliage, false colour in the skies, etc...).

As a surprise he is getting RAW file without any colour information (no differentiation between the signal absorbed by the R, G and B Foveon layers). This is a RAW file:

https://drive.google.com/file/d/1kJMtnPooyX-N8uGW4PhVNq8xsG8ZKyPi/view?usp=sharing

I have extracted the RAW data and plotted the histogram. Indeed the three histograms are the same, just for an exposure factor:



If we take the individual per channel RAW data and match exposure, they become undistinguishable (just the expected differences in SNR)



I calculated the relative exposure histograms:



G is 1,2EV over B, and R is 1,55EV over G.

The question is: is this behaviour expected? is it because it's a Foveon? why are the Foveon layers not making any wavelength discrimination over the IR spectrum?. If I am not wrong, this guy will never get any colours on his IR shots using this camera and filter combination.

Regards

kers

  • Sr. Member
  • ****
  • Online Online
  • Posts: 3090
    • Pieter Kers

Maybe the sensors ( R, G and B ) are all almost equally sensitive to IR...
I would like to know what happens without the filter; i think it could be a soft coloured image?
and how sensitive the sensor is with the whole spectrum and without.
Logged

EricV

  • Full Member
  • ***
  • Offline Offline
  • Posts: 236

Silicon is fairly transparent to long IR wavelengths.  Most IR photons will probably go through the entire sensor without interacting, and those that do interact are equally likely to interact at any depth.  Foveon sensors get their color information from the depth at which photons interact, so IR photons will provide little "color" information.  I guess the relative intensities of R>G>B are due to the relative thickness of those silicon layers, plus any scaling which Foveon applies to those signals.

What is the spectral response of the IR filter used by the photographer?  If it is a narrow bandpass filter, there is no "color" information to begin with.  Assuming it is a fairly broadband filter, the shorter the wavelength (closer to visible red), the better the chance of getting some differential IR absorption with depth and hence some color information.  Short wavelength IR will prefer the blue channel for example.  But for long wavelengths, this preference will disappear, as noted above.
Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2349
    • The Last Word

Maybe the sensors ( R, G and B ) are all almost equally sensitive to IR...
I would like to know what happens without the filter; i think it could be a soft coloured image?
and how sensitive the sensor is with the whole spectrum and without.

There are not individual color sensors. There is no CFA. I'm guessing that blocking the visible light will result in material response only in the longest-wavelength layer, which is then, through the off-diagonal-heavy Foveon compromise matix, turned into what you saw in the OP.

Jim
Pages: [1]   Go Up