I read their position paper and I think some of the claims are overblown and
maybe in some cases specious:
http://www.fujifilm.com/photokina2008/pdf/release/super_ccd_exr_e.pdf
It looks like when running in 6Mpixel modes there are some advantages in some
ways, but not really in 12 mpixel mode. The mode with increased dynamic range
is an example where it maybe a fairly good tradeof, but you land up in a mode
pretty close to 6Mpixel rather than 12.
For example it appears the traditional beyer pattern, which is optimised for
highest resolution in green, since this is human eyes peak sensitivity and peak
sunlight wavelength, will have higher horizontal resolution than the new exp
pattern. This actually even adds a little more since the eye/brain aparently
also subjectively percieves it to be highest reolution if it has greatest
resolution in the horizontal direction.
The new exp patern is quite strange in that the highest resolution in green is
on the left diagonal , with all other directions being lower! This is
non-optimal since horizontal is best subjectively. The beyer has equal, but
reduced by 45% resolution on both diagonals,while the vertical and horizontal
are the the highest and equal.
The exr horizontal green resolution, is lower than the beyer, at 12Mpix because
of diagonal pattern.
Because of the unsymetrical spacing of exp, it also aliases at even lower
resolution than you would normally imagine from that raw color pixel count. If
you photographed a thin curved line object like a hair say, it would produce
some additional low frequency aliasing because sampling was more non-uniform
,at different angles than beyer. (think: if a line pair is sampled by two close
spaced pixels then they get resolved, but the next pair of pixels in exp are
now more widely spaced, so the next adjacent line pair is not resolved)
The wide dynamic range option which is really more like 6mpixel, since it
selects best exposure from two diagonal pixels, then has uniform sampling, at
least at a particular brightness level.
The claim that there is less aliasing from binning diagonally left adjacent
pixels in exr, rather than more widely separated ones (bayer), is complex to
analyse and is more orientation dependent than beyer.
However, because the beyer samples at a higher horizontal frequency than the
non-uniform exr, then binning two horizontal (or vertical) pixels in beyer,
still has lower aliasing, since the actual sampling was done at a higher
resolution (frequency) thus having lowest aliasing.
The s/n ratio improvement is exactly the same for equal area photo-diodes in
the two technologies. In this comparison for the bayer at 6mpixels, the optimal
bining for best subjective and actual resolution, would be combining vertical
pixels.
Tim Hughes
--- On Fri, 5/1/09, usher99@xxxxxxx <usher99@xxxxxxx> wrote:
> From: usher99@xxxxxxx <usher99@xxxxxxx>
> Subject: [OM] clever use of available sensor real estate
> To: olympus@xxxxxxxxxxxxxxxxx
> Date: Friday, May 1, 2009, 2:12 PM
> http://www.dpreview.com/reviews/fujifilmf200exr/page2.asp
>
> Moose might be amused that low light performance a few
> generations
> later still doesn't really exceed the F30/31--the
> latter are still
> expensive.
>
> Not many Moose sightings of late.
>
> A Trace Worried Mike
> --
> _________________________________________________________________
> Options:
> http://lists.thomasclausen.net/mailman/listinfo/olympus
> Archives:
> http://lists.thomasclausen.net/mailman/private/olympus/
> Themed Olympus Photo Exhibition: http://www.tope.nl/
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|