On 2004-07-08, at 22:49, Jostein wrote:
I'm not sure about the chips being soft. Each sensor in the matrix is aThis is not rather chip. RAWs straight out of *istD when converted and shaprened in good RAW converter like Photohop CS Camera plug-in are significantly shaper and have more visible details than JPEGs made in camera. That would lead to concluson that processing engine in *istD could be less perfect when converting picture from CCD data.
discrete dot, and putting them all together should create an image where one
dot equals one pixel. This is true even after the Bayer interpolation, since
the only thing interpolation do is to calculate the exact hue and brightness
of one pixel, based on data from the neighbouring pixels.
My understanding is that images straight off the chip like that will haveAA filter seems to be quite strong in *istD, but as I wrote above, this has rather something to do with in-camera processing of RAW data.
problems with moiré and aliasing. Apparently, this is avoided by putting a
softening filter in front of the CCD in most DSLRs. I know for sure that
Kodak don't, and apparently, Nikon has done some experiments with reducing
the pre-ccd softening in the D70. That's why I asked the original question,
of whether the *istD uses pre-ccd softening only, or if something is added
in firmware.
To begin with, my questions rose because I think the *istD is _too_ soft.S2 samples from Dario's test shows some sharpening halos, while the ones from *istD don't... With proper USM it is possible to obtain good image from *istD. Realtively soft details are however mainly visible at quite flat areas in certain (large) distance from camera. Again they are not so soft when picture is converted from RAW in Photoshop CS... Strange.
Then I also remembered when Dario Bonazza tested the *istD against the Fuji
S2 Pro, he found that the *istD images would endure less sharpening before
developing halos.
The other day, I took some shots with a FA50/1.4 stopped down to f/8 of a
black metal profile against a grey sky, and played around with USM in
photoshop CS. I also persuaded Cotty to send me a couple of files from his
Canon D60, and I plan to go to Darios site and download some stuff from
there as well.
Let us know your findings!
So far, my impression is that the USM mask is less effective on slightlyIt depends on specific USM parameters, what exact did you use to sharpen your images?
soft images. It tends to place the mask a couple of pixels further away from
the contrast edge, which results in much more visible halos. If this is
correct, I will jump to the conclusion that inherently soft files can't be
_made_ sharp in software. Which is a conclusion that reinforce my initial
sentiment that the *istD is too soft to begin with.
Given that the Nikon D100 (and possibly D70) is built on the same chip, whyNo, it is rather a matter of camera. On the other hand Reichmann, who otherwise didn't like *istD worte about softness in *istD these words:
is Pentax the only "softie"? I just can't believe that Pentax lenses are
_that_ much softer than the competition; it's more likely that there's
something about the camera...
"In late October as this is being written, and during the few weeks that the *ist D has been on the market, there has been discussion on some Net discussion forums about the camera's images being "soft". Nonsense. Pentax has wisely avoided oversharpening images in the camera. When properly sharpened they leave nothing to be desired. Once again beginners and the uninformed are confusing resolution with sharpening."
And who is right??? ;-)
-- Best regards Sylwek
