Bayer - Foveon comparison

Started Nov 23, 2008 | Discussions thread
DMillier Forum Pro • Posts: 21,002
Re: Resolution versus detail......

Hi Lin

It's funny how these debates often hinge on mis-understandings or different interpretations of words!

I won't get into the second part of your response just now, but with regards to the DR issue, I have this to say:

I think we agree that in this pair of images, at least in some parts, the Sigma shadows are more 'open' - by which I mean not as dark. Whether this translates in practice to more shadow detail depends - for these images only - whether lifting the shadows of the 50D image can reveal additional detail. I can't process the 50D raw so I can't check this for myself.

On the wider sense of what is the cause of the lighter shadows in the Sigma image, it is important to avoid leaping to conclusions. One could easily believe that the Sigma has more shadow detail because it has more dynamic range and that this is an advantage in favour of the Foveon. Many people have made this claim.

However, this could be an illusion. One can easily improve shadow detail by exposing hotter. This will work fine for low contrast subjects but fail for high contrast subjects. Likewise, we don't know whether the lighting changed, even fleetingly and marginally. It doesn't take much of a cloud to drop the lighting level by half a stop.

Without knowing exactly what the luminance range of the subject is and whether the correct exposure was used AND whether subsequent processing was correct, it is difficult if not improssible to draw a reliable conclusion as to DR from looking at somone else's images.

I think if I draw any clear conclusion from what I see in this comparison, it is more of what I saw in the comparison with the Kodak 14nx. Cameras in the better than 6-8MP category provide pretty good image quality whatever the brand and technology. Trying to find one approach (or brand) to be overwhelmingly superior seems pointless; sometimes images from one look better than the rest, sometimes it's the other way around. We are big time in nit picking territory.

For the moment, Foveon and bayer approaches will continue to co-exist with their respective flaws and qualities.

And both systems continue to be good enough to challenge the abilities of most photographers. I have to say that the technology is definitely NOT the limitation in my own work, LOL.

Lin Evans wrote:

Hi David,

The "claim" of greater DR is for these twp photographs, period. There
is no reason to extend the argument or implication beyond what I
said. If you read my original statement that is quite clear. The
SD14 conversion demonstrated greater DR than the 50D conversion in my
opinion whether looking at the OP's originals or later conversions
from RAW. Shadow detail can be pulled from the SD14's conversion
which I can't see in the 50D conversion. No attempt was made to
determine "why" because there are no controls over variables so there
is no reason to attempt to make any more of the statement than
expressed or intended. It's not a scientific experiment it's an
observation.

It's also clear that John has no idea of how to discern the
difference between resolution and rendered detail - to him they are
identical. To me they are absolutely not identical. Resolution is
measured in the relevant sense in line pairs per mm or in lines per
image height. It's quite possible in one case to "measure" a known
quantity of poorly defined lines which are only recognizable because
of apriori knowledge that the subject matter is composed of lines. In
a corresponding case there may not be lines visible at the same
converging point on a resolution chart thus indicating lower measured
resolution, but the lines which are visible and to the point of
Nyquist being sharper and better defined and with greater contrast
means better detail rendition. Detail and resolution are not
synonymous. It's very much like early higher resolution but lower
contrast Zeiss lenses versus higher contrast but lower resolving
Japanese lenses. The Japanese lenses produced better detail but lower
measured optical resolution. I don't believe John will ever
understand the difference because to him it's all "numbers" and to
the real world it's what one "sees" clearly versus some semblance of
"something" present which is so fuzzy and indistinct as to be
unrecognizable in character.

In some sense it's back to the mush at Nyquist from a CFA sensor with
AA filter versus a Foveon sensor which displays so called "false"
detail beyond Nyquist. Whether one chooses to call a pixel
representation of five blades of grass "false detail" when there are
actually nine blades of grass, or imagine that there "may" be grass
where the AA filter blurs the green mass into an unrecognizable blob
is exactly the issue. Whether there are 18 upright supports
represented by an indistinguishable blur or reduced to 10
identifiable upright supports via "false" detail. Which has greater
"detail?" If I can identify structures, fences, supports, rocks,
grass, leaves, etc., in a lower "resolution" capture by a Foveon
sensor than in a higher resolution capture by a sensor with AA filter
then the lower resolution capture reveals greater "detail" even
though it has lower measured resolution.

Lin

-- hide signature --
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow