Tuesday, January 01, 2008

Photographic color rendition under varying illuminants

Sorry once again for the non-fun blog. This will stop soon, promise...

I always wanted to do this experiment - take a sample, photograph it under several different light sources, correct the white balance for each, and once this is done, see how much different things look in the corrected photographs. From a practical standpoint, this translates to "how important is it to use daylight to photograph art or a still life?".

I am going to skip the discussion of color rendering index (CRI) and other related topics - if you don't know what they are, you probably aren't reading this anyways... Some day I'll perhaps provide at least some references. Meanwhile, on to the experiment.

G, my watercolor paint swatch obsessive wife, unwittingly provided a color test target - full strength tube watercolor paints. My goal was to take three nearly identical photographs, varying only the light source.

The first light source is the reference standard - mid-day daylight, in this case natural light under a cloudy sky (no direct sunlight of the target). The second source was a compact fluorescent light fixture (sorry, don't know the model) with a specified 5000K, 91 CRI 70W bulb (actually three 23 w. bulbs combined). Finally, I used a quartz halogen source. For this, I started with a halogen flood lamp, but didn't like the light distribution, instead using two 75w halogen floor lamp fixtures.





The targets were photographed using a Nikon D80 at f8, 50 mm, using Nikon RAW mode. Once the images were captured, Adobe Camera Raw was used to convert the images to Adobe RGB files. The following procedure was used for each image:



  • Set the white balance using the white center of the color wheel. This is done by clicking in ACR with the white balance eyedropper.

  • Set the exposure (essentially adjusting the histogram) as follows: view in threshold mode, and advance exposure until the first sign of clipping (this always happened in the oranges, but the paper white, interestingly enough...), then back off .05 exposure units.

  • Rotate and crop the images so they are approximately the same size.

  • Specify a conversion to Adobe RBG (which provides a wider gamut than sRGB). Then, for web viewing, (the picures you will be viewing below), also convert a copy to sRGB mode and save as a separate .jpg.


No other manipulation was done in Photoshop, save adding a text layer to identify the light source. So, here are the three resultant pictures:


First, the reference daylight picture:


Next, the Halogen picture:

Finally, the High color rendering index Fluorescent lamp:

Analysis - lots of thoughts, lots of things to say, no hard conclusions - but those will have to wait for a bit, as it is now time for bed....









3 comments:

Carol Douglas said...

That is NOT what I expected, since I thought the discontinuous compact fluorescent would have gaps in the color wheel but from visual comparison (tough when they're lined up vertically) it appears that is actually the closest to reality.

Anonymous said...

I notice that the labels on the bottoms of the images don't match the text. For example, the image under "First the daylight reference picture" says "compact fluorescent", and vice-versa. I suppose this means the halogen one is correct.

Carol Douglas said...

Well, Doug, that explains EVERYTHING.

Happy B'day, Chuck.