Sunday, February 17, 2013

Panasonic Announcement - Is This the Next Big Technology Improvement for Camera Sensors?

In choosing to enter the Micro 4/3 camera format a few years back, I knew that I was making a bet - a bet on how far sensor technology would improve in the near future. I'm sure that most of you know that image quality is generally correlated with sensor size - the bigger, the better. This is why people still compare sensors to "full-frame," which means the 24x36mm size of the image produced in a standard 35mm film camera. Most sensors in cameras less than $10,000 are smaller than that.
So, buy the biggest sensor you can afford, right? Wrong. Some other things are correlated with sensor size - camera (and lens) size, weight, and cost.
So, when Micro 4/3 came along, the question for me was whether or not small sensor technology would likely improve to the point where the Micro 4/3 sensor would perform as well as the benchmark at that time - which for me was the Nikon full-frame sensor. So far, it isn't there. My Nikon D700 is still miles above any Micro 4/3 sensor in terms of image quality. That said, Micro 4/3 has made significant improvement. The latest sensors in the Panasonic Lumix DMC-GH3 and Olympus OM-D are really very good. Even the previous round of technology, like the sensor in my DMC-G3, is pretty darned good. I would say that APS-C sensors (larger than Micro 4/3) are close to the benchmark of older full-frame sensors, and Micro 4/3 is where APS-C used to be. I'm still holding out hope for the next generation of Micro 4/3 sensors to achieve my initial wish, or get much closer.
What does this mean for me, practically? It means that when I must have the very best image quality, especially when shooting in low light, I still grab my Nikon D700. When I'm going to be traveling a lot, and generally shooting in good light, Micro 4/3 is the right choice for me.
One area where companies have been challenging the standard sensor technology is in the color filters on top of the sensors, necessary for making color images. The way color images are formed on most sensors is by using a set of four pixels to capture all of the color information, which is then combined into a single pixel in the resulting image. The standard filter array is called the Bayer filter. The Sigma Foveon sensor and the XTrans sensor from Fujifilm are attempting to improve upon the Bayer filter design with at least some success. Panasonic has announced the development of another technology that could really improve the performance of their sensors by improving how color separation is done. It uses miniature light splitters to deliver specific colors to the sensor, rather than filters.
(Click Here) to read a post about this, and Panasonic's press release, on
The idea here is that the filters used to enable the Bayer concept are lossy - they lose approximately half of the light that hits them, transmitting only half to the sensor. Panasonic's concept uses optical splitters, which can be manufactured using semiconductor processes, to separate the light into its component colors, without so much loss of intensity. If the loss could be cut from 50% to 0% (impossible), that would mean the equivalent of one full-stop of image quality improvement. In other words, if this works extremely well, a picture taken at iso3200 with the Bayer filter could be taken at iso1600, which would give the associated improvement in noise performance.
What is not clear is whether this approach would allow a reduction of the number of sensors in each color pixel to be reduced from four (as used in the Bayer pattern) to three. That could potentially improve the resolution available in a given sensor size.
One could even imagine a futuristic version that scans the splitters during exposure, potentially limiting it to one sensor per image pixel. Now, that's simply imagination on my part - Panasonic has not hinted at any type of scanning splitter. And, it would have to happen extremely fast. But, if you could have it, it would allow image resolution like that available in the Leica Monochrom (which only shoots black and white, providing the ability to use every sensor on the chip for gathering gray level information, and yielding incredible image quality), but with the option to shoot color that the Monochrom does not have.
Or maybe the splitters could be switched off when shooting black and white? That would provide excellent color response when you want it, and improved resolution when you want to shoot black and white. Again, pure conjecture on my part. But wouldn't it be cool?
Oh, by the way. I've been neglecting my Lumix Micro 4/3 gear. So, I'm taking it to the California desert in a few weeks for a one-day workout. More to come on that subject.