There is no one would rather
have 3 color samples
we are not talking about the sd9 here. we are talking about the
resolution capability of a vertical color filter pixel relative to
a lateral
color filter pixel.
Who doesn’t believe that 3 samples at one special location is better? The real/current market issue is how an X3 at 3.4MP X3 compares to Bayer at something like 6MP (and going to about 9MP). We all know Bayer sampling has problems.
the vertical color filter pixel represents the best solution to the
physics problems: quantum efficiency and diffraction limited
resolution.
You have a very good point I think about diffraction limited resolution. Optics is not my field but I do know to worry about diffraction limits. I think the market is going to move from “megapixels” to the quality of each pixel in terms of color accuracy, dynamic range, and noise/ISO. The issue will then become as it was for high quality film, what technology can best capture light PER unit area.
we have some work to do engineering some of the more practical
problems such as noise floor. but I expect that progress will be
more rapid than the CCD guys
Was it deliberate that you left of CMOS Bayer Cameras? I agree that CMOS will eventually rule and squeeze out CCD, but it could be Bayer/Spatial color filtered still. The only current product with the X3 is competing mostly with Canon’s CMOS DSLRs (and to a lesser extent Nikon’s CCD). I guess it could be that you are more thinking about competing in the P&S world where Sony dominates with CCD.
I think it is a good idea to do a reality check with REAL cameras
the 300d image has no color moire because there are no resolved
features near nyquist- that is there is no single pixel contrast at
all.
no place in the image shown is there a black pixel / white pixel
/ black pixel where pattern pitch = pixel pitch. since they cannot
resolve minimum pitch, they should not be claiming so many pixels!
What? This is a test chart that sweeps through all different pitches of the distance between black and white transitions. For comparison take a look at the older D30 chart. As the lines converge it goes into rather distinct color moiré as would be expected (and shown in your demonstration figure). I was quite surprised that the 10D/300D did not have this as well. I don’t think that it is a function of the AA filter as they would have to blur the image too much to get it this clean, but rather some form of moiré detection algorithm (the 10D/300D are “too clean”).
the SD10 image shows classic chromatic aberration (CA), a radially
symetric pattern. if single pixel color features are going to be
resolved,
then chromatic aberration is going to be resolved too. a better
sensor will show more lens problems. the CA is just being blurred
out of the image from the 300D.
Why is it showing some violet color fringing on the Black to White transitions in the SD10 photo even in the center of the photo where the aberrations should be minimal? Per my previous comments, if the 300D AA filter was blurring out the color this much, it could not resolve as much as it does. I’m not sure what Canon is doing, but they seem to have some other algorithm at work.
the vertical color filter has fewer radial color artifacts than a
color
mosaic sensor because there is no lateral color crosstalk possible
(all pixels are the same)
That would seem to be true in theory, but I would like to see it in practice.
Historically for a new technology to win out over established high
volume technology, the new technology has to show clear and
indisputable advantages
this raises a little discussed issue with the vertical color filter
technology,
which is manufacturing. the plastic color filters are a messy and
difficult
to control wafer process technology. rework rates are high and color
depends on incoming QA, layer thickness, etc etc. by contrast, the
color filter crossovers of the vertical color filter depend only on a
fundamental property of the most perfect crystal made by humans
in mass production. this is very important for "high volume
technology".
That sounds nice in theory, but they are stamping out those messy color filters by the many millions. They also have the advantage of being able to “tune” their characteristics and even use extra colors (ex. Sony’s “E” filter) to expand their gamut.
technologies that are fundamentally easier to make have a way of
succeeding in the long term.
It depends on how significant the differences may be. It would seem that the X3 process is more complex than say Canon’s CMOS sensor process at least at the Silicon level. The X3 has multiple wells and is squeezing 3 times the transistors per pixel (from 3 for Canon to 9 for the X3). Thus the Canon’s CMOS would seem to be cheaper per unit area for the Silicon.
happily America is still populated by large numbers of people who
foolishly believe that if they invent a better mousetrap,
Hey, I’m all for the better mousetrap and American inventiveness. I’m an independent inventor that is selling a much better way to build high resolution displays for near eye and projectors.
I agree that better dynamic range is important to consumers.
Once again, others are working on the CMOS sensors at both the high and low end too with Bayer filters. I agree that CMOS is going to win out in the long run.
