DPR Forum

Welcome to the Friendly Aisles!
Register now and use your old dpreview username.
Enjoy this modern, easy to use software. Look also at our Reviews & Gallery!

Bayer cameras color blind?


Active Member
I am just wondering how a bayer sensor can accuratly see yellow, magenta, and cyan if the filters only let through red, blue, and green.
I am sure there is some overlap in the response of the pixels but I don't see how it's possible a bayer type sensor to accurately measure these colors.

The foveon chip does not filter out any of the photons so they will all contribute to the luminance output and the number stopping at each layer determine the color. I think this is why the pictures look like Kodachrome slides.

The first thing I noticed when using the SD14 is that pictures of my lawn showed yellowish grass in some areas while my Olympus made all the grass the same color green. I went outside to look at the lawn and sure enough there was yellowish grass. Also, my Olympus and Fuji cameras never get flesh tones right . The yellow component of the skin seems to be replaced with an orange red funk that you just sort of get used to and start to think people really look that way.
Taz....every sensor has there + and -....I use a Bayer type sensor on my Sony DSLR's...and your right I like the coloring from my SD14 better...but on the flip side...I can't get the speed out of the SD14 or 15 that you can get from some of the other camera's...there is always a trade-off...

When I had a Fuji S3 - (when I was shooting Nikon) - the S3 really shines in RAW or TIFF mode...the JPG output is really sub-par...just like the SD14...on my Sony's...I shoot RAW due to the fact I can always play with the image more and edit-it to where I like it...

But yes...I think the Foveon chip is better at colors -

Tony C. :z04_cowboy:
I am just wondering how a bayer sensor can accuratly see yellow, magenta, and cyan if the filters only let through red, blue, and green.
Wrong assumption.

Filters do let all colors through, just in different amounts. Pure green may result in 0.2R+0.7G+0.1B, pure yellow in 0.46R+0.46G+0.08B and so on. Search net for Kodak sensors spectral response curves - they are freely available. Some sensors have narrower spectral response curves than Kodak ones, but they always overlap.

Same for Foveon - every layer registers a bit of every color, just proportions differ. Spectra overlap a bit more than for bayer though, this increases sensitivity to subtle color variations (good) and color noise (bad) (compared to bayer sensors).

And if you have some spare time, read papers and articles abut color vision basic, about color space transformations and so on. Initially you don't understand anything, it takes many evenings to catch general ideas :)
Now I can understand why there is no free lunch with sensor design. I prefer greater color sensitivity over higher speeds, I guess it's because I shot enough film to think ISO 400 is fast.

I wonder if the human eye can change it's response curves as needed to compensate for lighting conditions.

We have a long way to go before we can match the processing power of the human brain.
Color blind?

To see how the sensor in your camera "sees" yellow, cyan, and magenta, look at your monitor, a device made up of Red, Blue, and Green elements. Your monitor displays yellow by combining green and red, cyan is a combination of green and blue, magenta is red and blue. the other 16 million colors are combinations of the three. Likewise, your sensor sees all colors as some combination of red, green, and blue. The difference between your monitor and a sensor is that one emits light and the other collects it, the color combinations are the same in both cases.