Testing The Only RGBE Sensor Ever Made
18 years ago in the early 2000s and early days of digital photography, a headline article appeared on the small but growing DPReview blog announcing an exciting new type of sensor. This sensor, made by the leader of digital sensors Sony, promised a more accurate color response by using a four color filter array rather than the traditional three color filter. Red green blue (RGB) color filter arrays were standard, but this new sensor added emerald to the list (RGBE).
Sony F-828 - The Only RGBE Camera
Nearly twenty years later you are probably like me and had never heard of this type of sensor before, and thus can assume it did not prove successful for one reason or another. You would be correct in this assumption, but the reasons for this are not immediately clear.
I wanted to test this unique sensor out myself, so I recently picked up the Sony F-828: the one and only camera to be produced with an RGBE sensor. A special camera in and of itself, regardless of the sensor, the F-828 features the 8mp CCD RGBE sensor, a tilting body, and loads of high tech features for the day. It followed the critically successful Sony F-717, but failed to live up to the success of its predecessor.
Side-by-side RGB/RGBE Sensor Comparison
As part of my use of the camera I pitted it against my Canon Powershot Pro 1, an 8mp CCD fixed lens camera from the same era. The sensor is reportedly the same one as in the Sony F-828 just with a three color filter array instead of four. It seemed the closest way to see the difference in color response between an RGB and RGBE sensor, if there was any. The results are below. Please note that several factors can influence how an image appears on a digital camera. White balance, image processors, lenses among a few. But the results reflect what I have seen in the hundreds of other pictures I have taken with the Sony F-828.
'Real' vs 'Standard'
The Sony F-828 offers two image modes - Real and Standard. Standard will appear closer to traditional sensors, but maintaining less color deviation than the competitors of the day. Real will offer even less. In common terms, the Standard mode looks higher contrast and closer to the colors of the Canon Powershot Pro 1, whereas the real mode looks lower contrast and with an orange cast. Real mode is what I used for the above comparisons, because it represents what the Sony F-828 was trying to accomplish - a different color response than any other camera sensor of the time.
Here is a comparison of Real vs Standard on the Sony F-828.
And here's a comparison of the more typical appearing Standard mode from the Sony F-828 against the Canon Powershot Pro 1.
Conclusion
I'm not a scientist, let alone technical enough to understand all the factors behind the design and implementation of the RGBE color filter array. But as an artist, I can take a stab at why it did not succeed and was subsequently never released in another camera.
Looking at the results shows clearly that both the Real and Standard mode Sony F-828 images have a more brown/orange tint. This follows the original examples Sony shared at the release of the sensor. In some ways the images do look more accurate and true to life than the Canon Powershot images, which seem overly saturated. On the other hand, the greens appear to me too washed out and the blues to a lesser degree as well. Whether this is considered "better" or not is beyond my abilities to say, but certainly overall with images from the RGB sensor appear more visually appealing, generally. And I think this is what consumers found as well.
Was the relative failure of the RGBE sensor and Sony F-828 due to consumers not enjoying the images generally? Or, could it be that the hype Sony gave over the new sensor was doomed to fail no matter the results? I'm not sure we'll ever know entirely. Let me know your thoughts of the comparison and images shared above. I think the images have a unique look that could be used creatively for some great shots, but it might not be for everyone, and not for all the time.
Happy snappin'
These are all SOOC JPEGs, right? I agree that the RBGE seems a bit warmer and pop a little less, but I wonder to what extent the system got dropped because the RGB images could probably be edited to look just like the RBGE ones, if that was what you wanted. If this is the case, it's easier to do it with software then go through the technological development necessary to use a novel color filter array, with its accompanying resolution and moire issues.
Something like a Foveon sensor or X-Trans, the variations from the plain'ol Bayer filter I can think of off the top of my head, offer (supposed) benefits over the more standard color filter to a great enough extent to justify their research and development, as well as whatever resources it takes to manufacture them.
But Cynicism aside, why Emerald? Isn't that just like a green with a little more yellow? If RGB already has green, why not make a RGBY filter? Was it geared toward landscape/woodland photographers? The justification for the extra green pixels (vs red or green) in Bayer filters is that the human eye has 2x more green cones than red or blue. But then what is the advantage of going with Emerald rather than another green?
One criticism of Bayer is false color, since lines alternate between RG and BG. Would adding Emerald help with the false color problem?
The mind boggles haha.
Uploaded files:
Quote from Justin Tung on January 2, 2022, 4:12 pmThese are all SOOC JPEGs, right? I agree that the RBGE seems a bit warmer and pop a little less, but I wonder to what extent the system got dropped because the RGB images could probably be edited to look just like the RBGE ones, if that was what you wanted. If this is the case, it's easier to do it with software then go through the technological development necessary to use a novel color filter array, with its accompanying resolution and moire issues.
Something like a Foveon sensor or X-Trans, the variations from the plain'ol Bayer filter I can think of off the top of my head, offer (supposed) benefits over the more standard color filter to a great enough extent to justify their research and development, as well as whatever resources it takes to manufacture them.
But Cynicism aside, why Emerald? Isn't that just like a green with a little more yellow? If RGB already has green, why not make a RGBY filter? Was it geared toward landscape/woodland photographers? The justification for the extra green pixels (vs red or green) in Bayer filters is that the human eye has 2x more green cones than red or blue. But then what is the advantage of going with Emerald rather than another green?
One criticism of Bayer is false color, since lines alternate between RG and BG. Would adding Emerald help with the false color problem?
The mind boggles haha.
Yes, all SOOC JPEGs.
I am with you, I don't really understand what they were going for. I wish I could talk with some of the people who worked on the project at Sony and understand if it was an engineering decision that they really thought could offer greater color response, or if it was driven by a marketing team. I don't want to offend anyone who works in marketing, but as an software engineer we often get projects mostly driven by the headlines they will make and not the value they provide 🙂
I'm sure someone more technical than me could dig into why this could be more beneficial or not, but my hunch is it's not, or at least not in any practical way.
