Bayer with NO anti-aliasing (Kodak Pro 14n)

If you take a picture using film; are
the pictures then sharp or unsharp?


Roland

 

uhhh ???

Each time there is a complex colored pattern on an image from a 6MP
Bayer based camera, *there are* chromatic artefacts. Their level vary
from camera to camera, depending on the strength of the AA filter. But
the AA filters are always too weak to remove all the artefacts, which
come basically from the aliasing of the chromatic components. Strong
enough AA filters to remove all the aliasing would blur too much the
images and the resolution (even in luminance) would drop to something
equivalent to 1.5Mpix.


Claiming that the SD9 Foveon sensor does not suffer at all from
aliasing is not fair, but claiming that the usual cameras with 6Mpix
bayer sensors do not suffer at all from chromatic artefacts is not
fair as well (besides the facts that both claims *are* false from a
pure signal processing point of view).

 

Look at it, if it is empirically sharp, then it is sharp.

 

You should be worried, he responded to you words by accident, obviously,
that is what he really thinks of what you said. Ouch again.

 

Putting L glass on a 10D is like putting $100 diamond in a $10,000 mount.

 

Putting Sigma glass on a Sigma camera is like putting a dog poo on a cow
pat.

 

http://www.hyperdictionary.com/dictionary/empirical

Interesting that a synonym of empirical is "falsifiable". Rather like the
SD9s false sharpness.

 

Wrong again.
I was talking about YOUR images.

 

No, you said only an idiot would think crop factor is not optical zoom.

 

The whole wolrd is against you...

http://www.google.co.kr/search?q=10.2MP&ie=UTF-8&oe=UTF-8&hl=ko&btnG=%EA%B5%
AC%EA%B8%80+%EA%B2%80%EC%83%89&lr=

 

Sweaty Betty, clean up your act...

 

For some reason you say that it is an absolute truth that
Foveon images are sharp and that Bayer images are not.

So ... I assume that the same kind of absolute truth
must exist for photos taken on film. Do you know which?


Roland

 

Not me - but I may not be a part of the wolrd.


Roland

 

In message <br9bhc$iol$>,

Nonsense:

1) You don't buy an 'L' lens just because of the optics. They are
sealed better from moisture and dust; they manually focus smoothly
without play.

2) Every lens varies in sharpness at different apertures, and different
focal lengths if it is a zoom. L lenses tend to have a larger
percentage of their range sharp enough for the sensor.

3) Many L lenses can resolve far more than than current DSLR sensor can,
but their higher sharpness means more contrast in the highest
frequencies recorded, and less artifacts should you decide to boost the
high-frequency contrast for final output.

4) Many L lenses have almost 0 chromatic abberation. My 300mm f4L, for
example, has so little aberration that no focusing compensation is
necessary for IR.
--

 

In message <bra7tc$7l8$>,

Ooh, whoopdy do! Opinions that you would never respect unless they
agreed with you can parrot press releases made by dishonest companies!
WHat will they think of next?
--

 

In message <br9aoi$ifq$>,

Sharp with *what*? Sharp with illegitimate sampling artifacts, or sharp
with real image detail, consistent up to a certain frequency?

You can downsize any DSLR image to a 60*40 pixel image with nearest
neighbor, and it will be very sharp. Sharp with *what*?
--

 

In message <>,

And how often does *that* happen, outside of benchmarks? All I see in
most 10D bayer images, when zoomed in, is a limitation in detail. I
don't see surprise colors, I don't see false textures, I just see an
image wisely to filtered to what most scenes need.

The bayer artifacts from DSLRs are not noticeable in real scenes. The
SD9 artifacts are glaring in *any* instance where a sharp lens is used,
and is in sharp focus.
--

 

No I didn't, I said look and it and decide if it is sharp. Yes, Roland, you
are capable of deciding when a picture is sharp, you don't need a commercial
entity to tell you what you see.

 

That would be a sharp, 2400P non-interplated image. The "non-interpolated"
part is really only an approximation of geniune Foveon quality, as it is
accomplished by averaging previous interpolations. IOWs, averaging 4
guesses that were based on 25% real data. While this doesn't really get to
a probability of 1, and is therefore is obviously never as good as simply
knowing the proper color through direct image sensing, it is a reasonable
standard for the purposes of this discussion.

Most want at least 1 sharp, non-interpolated MP for mid-range amatuer work
(not to be confused with 4MP-interpolated, which is the default Bayer output
for 1MP non-interpolated). Even the soft, sensor-poor Canon 10D can output
a 1.58MP non-interpolated image with decent sharpness, roughly comparable to
a non-interpolated 3.43MP Foveon image. 3.5MP non-interpolated
(14MP-interpolated) is about the cutoff for absolute bare minimum pro level
results with Bayer (using 35mm color film equivalent as the "pro" standard,
that is using color, not B&W targets) . Foveon does exceed that standard
substantially with only 3.43MP non-interpolated, but that is due to a major
reduction in artifacts compared to all-over-the-map Bayers, and consistent
performance through all color channels.

 

Allow me to provide a shoretned explanation.

"I am George. I can talk bollocks. Blah Blah Blah Blah Blah sharp blah blah
blah orhpaned blah blah blah 10.4 blah blah 1.58 blah blah. By the way, I am
colourblind and have limited cognitive function".