Reading mode: Light Dark

Maybe I'm Blind

Go to last post
But the D100 is at a disadvantage because its Raw files are noticeably sharper and more detailed than even its best jpg output.

There has been much discussion as to why this is.

Phil says it is inadequate in camera sharpening which irrevocably destroys detail which cannot be recovered using USM; others have blamed it on a poor jpg engine. I have no idea who is correct but Phil notes that the D100 raw is marginally sharper than the D60 raw but substantially softer in jpeg...
Geir R said:
Ladehaug said:
Fundamanetally Dave Etchell's tests are broken because they are
comparing RAW output from the SD9 to in camera JPEGs from the other
cameras. Anyone who is an accomplished digital photographer, will
understand why this is not fair for comparing the sensor
technology. (If you want to credit it as a workflow advantage for
Foveo/Sigma, feel free.)

-Z-
Click to expand...
All of the files are JPG at roughly the same compression level. The
SD9 files are output from the Sigma software using Auto settings,
except where noted by Dave. So that's no real argument. Actually,
if a file is to suffer from JPG compression, then it must be the
SD9 files, since these images contain much more details within a
smaller resolution.

Geir Rune
Click to expand...
 
But the moire on the "clothesline" shot is terrible!


Geir R said:
Peter G said:
Because of the offset spacing between the colours, bayer cameras
will produce coloured moire if nothing was done. Almost every bayer
camera made has an AA filter to limit this.

Since X3 technology has full colour at each sensor it will most
likely produce moire that is the original colour, and thus the
decision to skip the AA filter. Non coloured Moire was considered
not objectional enough to remove.

Peter
Click to expand...
Remember the resolution chart? We could predict this moire to
happen based upon what we saw on the resolution chart. An AA filter
is really not necessary on the SD9, because the moire will not
change the chroma significantly (which is not the case for a Bayer
pattern color moire).

Geir Rune
Click to expand...
 
I forgot to mention that the out of focus background to this shot looks nicer in the SD9 shot. The blurred bushes in the far background look very strange on the D100 - like cardboard cutouts. The SD9 version doesn't look great but it looks a little bit less processed...
 
DMillier said:
Michael

All the editing package does is prepare the file by resampling the
file to the ideal file size required by your printer driver.

My own adjustments were just adjusting the brightness of the print
and colours so it has a close match to my monitor. If I leave the
default settings in place my printer produces a magenta cast in the
black shadows. I just compensate for that.

The 890 is the standard 6 ink A4 version of the 1280/1290 model.

The excessive colour saturation and posterisation effectI describe
is present onscreen even more clearly than in the print (at 1:1
viewing). Here's a crop:

http://www.millier.freeserve.co.uk/sigcrop.tif (230 KB)

I'dbe interested if anyone could explain this posterisation effect...
Click to expand...
I'm not sure I follow you.

Here is a jpg output of the sigcrop.tif you've post processed and made available for us to analyze:



The red channel is blown out after your post processing.

This is a crop from the original:



The red channel is over exponated in several spots. This is an issue that can be fixed if you have had the RAW file available.

Geir Rune
 
DMillier said:
But the moire on the "clothesline" shot is terrible!
Click to expand...
Yes, and that is why a couple of us have argued that even the Foveon sensors need anti-aliasing filters.

Notice something, though: since the Moire is pretty low in frequency, it means that the texture of fabric contains a pattern that is almost twice the sampling frequency of the sensor.

The lens is way sharper than what you need for the particular sensor.

Was this a center crop or an edge crop?
  • kc
 
Kok Chen said:
Notice something, though: since the Moire is pretty low in
frequency, it means that the texture of fabric contains a pattern
that is almost twice the sampling frequency of the sensor.
Click to expand...
Whoops, I mean to say twice the Nyquist frequency, which translates to the frequency of the sensor lattice, not twice. Fifty lashes for me :-)
  • kc
 
I agree. Canon is known to be best when shot in RAW, converted and post processed. The S2 review here at Phils, shows the difference between the S2, D100 and D60 jpg and RAW concerning noise levels. The D60 when shot in RAW outperforms the others in lowest noise in RAW (through ISO200). Even running through one of IR's "shootout" jpg images using Power Retouche for sharpening, it (D60 image) was greatly helped. Canon is known for its initial out of camera softness.

This is why tests are highly personal and inaccurate. Unless each is shot to its maximum and post processed to its maximum, all is mute.

Is the SD9 better out of camera in RAW? By everything I see, definitely it is, in regards to sharpness. Obviously the SD9 will require less post processing than the D60.

I do look forward to seeing the Foveon technology advancements and how this will affect the Nikon and Canon future technology.
Les
Z (is real) said:
Fundamanetally Dave Etchell's tests are broken because they are
comparing RAW output from the SD9 to in camera JPEGs from the other
cameras. Anyone who is an accomplished digital photographer, will
understand why this is not fair for comparing the sensor
technology. (If you want to credit it as a workflow advantage for
Foveo/Sigma, feel free.)

-Z-
Click to expand...
 
This is very interesting! I've just compared my copy of the original to my post processed version and although you are correct in that I have exaggerated the effect slightly, my copy of the original looks quite different from yours (which doesn't display the effect).

Is there more than one file on the IR site? Mine is called YIMG00731.jpg...
Geir R said:
DMillier said:
Michael

All the editing package does is prepare the file by resampling the
file to the ideal file size required by your printer driver.

My own adjustments were just adjusting the brightness of the print
and colours so it has a close match to my monitor. If I leave the
default settings in place my printer produces a magenta cast in the
black shadows. I just compensate for that.

The 890 is the standard 6 ink A4 version of the 1280/1290 model.

The excessive colour saturation and posterisation effectI describe
is present onscreen even more clearly than in the print (at 1:1
viewing). Here's a crop:

http://www.millier.freeserve.co.uk/sigcrop.tif (230 KB)

I'dbe interested if anyone could explain this posterisation effect...
Click to expand...
I'm not sure I follow you.
Here is a jpg output of the sigcrop.tif you've post processed and
made available for us to analyze:



The red channel is blown out after your post processing.

This is a crop from the original:



The red channel is over exponated in several spots. This is an
issue that can be fixed if you have had the RAW file available.

Geir Rune
Click to expand...
 
It was about 1/3 down from the top and about 1/3 in from the left
Kok Chen said:
DMillier said:
But the moire on the "clothesline" shot is terrible!
Click to expand...
Yes, and that is why a couple of us have argued that even the
Foveon sensors need anti-aliasing filters.

Notice something, though: since the Moire is pretty low in
frequency, it means that the texture of fabric contains a pattern
that is almost twice the sampling frequency of the sensor.

The lens is way sharper than what you need for the particular sensor.

Was this a center crop or an edge crop?
  • kc
Click to expand...
 
Jimmy Chen said:
For an example about UMS, Phil already have stated earlier during
the original SD9 samples that existing PS UMS function really have
very little effect for SD9 images because the images is detailed
down to the pixel level. He indicated the UMS function works best
with existing Bayer images. This in intself already adds more
detail into what a Bayer image it dont have, but at the same time
do not enhance any detail for the X3 image.
Click to expand...
USM cannot create details that were not captured in the first place. It can only enhance existing details and edges. It does so by applying a thin line of increased contrast between a bright and a dark area.

Applying USM to an image doesn't bring the image closer to the reality. It might look more pleasing to the eye, but it is not realistic.
Jimmy Chen said:
I do agree with you that all the shots should have been done in RAW
and in-camera JPEG already kills a lot of details. In IR's forum
one user have even gone as far to suggest producting only RAWs and
using all Sigma lenses for the test as different lens brands have
different quality and the results produced are no longer just about
what the camera bodies can do.
Click to expand...
In camera processed JPGs are no different than software processed JPGs. But with RAW and software, you'll have the big advantage of having the possibility to adjuste the exposure parameters manually (with the full dynamic range data) before you output it to a JPG. The IR images: The RAW images were output to JPG using the auto setting in the software (the exceptions were noted by Dave). Therefore, the RAW - JPG question is not a relevant issue in this comparison.

Geir Rune
 
Les said:
I agree. Canon is known to be best when shot in RAW, converted and
post processed. The S2 review here at Phils, shows the difference
between the S2, D100 and D60 jpg and RAW concerning noise levels.
The D60 when shot in RAW outperforms the others in lowest noise in
RAW (through ISO200). Even running through one of IR's "shootout"
jpg images using Power Retouche for sharpening, it (D60 image) was
greatly helped. Canon is known for its initial out of camera
softness.
Click to expand...


(79Kb)

I don't believe that any manual D60 RAW processing would bring out those details that are missing in this crop. Unless, of course, there is a serious flaw in the JPG output processing from the D60.

Here is the same crop sample as seen above, saved at JPG compression level 3 (LOW):



(17Kb)
Les #39998 said:
This is why tests are highly personal and inaccurate. Unless each
is shot to its maximum and post processed to its maximum, all is
mute.
Click to expand...
I'm sure Dave at IR is being quite neutral in this question.
Les #39998 said:
Is the SD9 better out of camera in RAW? By everything I see,
definitely it is, in regards to sharpness. Obviously the SD9 will
require less post processing than the D60.
I do look forward to seeing the Foveon technology advancements and
how this will affect the Nikon and Canon future technology.
Les
Click to expand...
It's all about capturing an image that represents the real world as accurately as possible. One important factor is the ability to render the details correctly in relation to its resolution, and this ability should not vary with the chroma. Bayer interpolation and USM doesn't bring the picture closer to reality.

Geir Rune
 
DMillier said:
This is very interesting! I've just compared my copy of the
original to my post processed version and although you are correct
in that I have exaggerated the effect slightly, my copy of the
original looks quite different from yours (which doesn't display
the effect).

Is there more than one file on the IR site? Mine is called
YIMG00731.jpg...
Click to expand...
Here is the link to the page I've found my version:

http://www.imaging-resource.com/PRODS/SD9/FULLRES/YIMG00731.HTM

Geir Rune
 
Just buy one and a couple of lenses (dedicated lenses) and trow some couple of thousands of Euro or Dollars in the drain.

jacques.
 
Yes, I hope you are right, but he will not be so stupid and buy it now,

jacques.
 
JPEG artifacts aren't my concern, highlight and shadow detail is more the issue. A more controlled experiment would have done the same RAW to JPEG processing for the other cameras. (Desktop software can be somewhat more sophisitcated than in-camera software.)

But if you are really interested in image quality, you shoot RAW and then manually adjust. This can have a huge effect on how great an image looks. It is clear that higher quality RAW processing is a need in existing software. That is why the PhaseOne and Adobe announcements are interesting.

-Z-
 
Hi Karl:

Lateral chromatic aberration (e.g. color fringing) is not affected by aperture. You could stop down to f/22 and it won't change. In fact, at around f/8 and beyond color fringing is normally the only significant aberration left, along with distortion which is also unaffected by aperture.

I'm not familiar with Sigma lenses, but I have found that several Nikon zoom lenses change the sign of chromatic aberration during zooming. In other words, it switches from magenta-green to green-magenta. Normally in the middle of the zoom range there is a position where there is hardly any color fringing at all. It would be interesting to examine images taken at the extreme focal length positions of wide-tele zooms. Clearly, if the color fringing flips sign it is a lens effect, not a sensor effect.

Brian
Karl Guttag said:
Jimmy Chen said:
Karl Guttag said:
Without more evidence and controlled tests, it is hard to
isolate/prove whether the chroma aberations in the I-Review
pictures are due to the Sensor or the Lens.
Click to expand...
So, you are saying you cannot come up with ANY conclusions
because you cannot isolate the issues ...
Click to expand...
That is not what I said, I said that I cannot absolutely prove it.
Karl Guttag said:
I think it is at
Karl Guttag said:
least in part (and more likely majorly) caused by the Sensor.
Click to expand...
And right after the above statement, you concluded it is the sensor
Click to expand...
I have said that it is because these pictures are taken at F8 which
helps reduce chroma aberations and that the aberations are getting
pretty bad even at 1/2 the distance from the center of a 35mm Frame
and that I doubt they would make/sell a lens this bad if it is this
bad at 1/2 a 35mm frame at F8. What is so hard to get about this
logic? We have similar shots with the other Cameras that don't
show such horrible aberations.
Karl Guttag said:
...

And I wonder just how did you manage and derived to that conclusion.
Click to expand...
It is a lot easier for me to derive my conconclusion that the
conclusion that the SD9 images have more "depth." Every SD9 image
I have looked at in detail in the IR comparision has significant
chroma aberations. This would suggest that it is more likely that
the aberations, if anything are creating an illusion of depth (like
a stereoscopic image image that is printed in one picture that you
where glasses to decode -- if you take two images and slide the
colors a bit, you will get an auto-stereoscopic effect).

Karl
Karl Guttag said:
--
jc
Sony F707
http://www.reefkeepers.org/gallery/f707
http://www.reeftec.com/gallery
Click to expand...
--
Karl
Click to expand...
--
J. Brian Caldwell
http://www.caldwellphotographic.com
 
Michael OHara said:
What other factors?
Click to expand...
Size and amount of sharpening.
Michael OHara said:
Why are you altering the images as posted?
Click to expand...
Because you cannot easily compare images of different sizes? Different sources use may different settings for the amount of sharpening. Are you suggesting that I just ignore this?
Michael OHara said:
Don't you ever trust your
eyes? I must admit that people like you are alien to me.
Click to expand...
What people who try and be objective and reduce the number of uncontrolled variables?

--
Erik
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top