D100: Photo Results Summary

[Gordon Berg]

the Exif dates for the gray building comparison photos shows the same pattern 5/5/2002 for d60, d1x and 1/1/2002 for the d100. Also software rev 0.32 for the D100.

2002:01:02 06:54:01 and Colorspace=Srgb

The segull says

2002:04:29 15:15:30 and Colorspace = Srgb

THE software version for both say 0.32

But at 4/29 that was just a week ago. And no other software
update? The camera must not have gone back to Nikon for any length
of time or I think it would have been updated. Or they could be
playing the same ole game like the D1 series, make changes and
never update the rev number. Didn't they ever hear of software rev
control?

Well, it appears that the cat is out of the bag as far as images
from the D100 are concerned. However, while EXIF information is
intact on the D60 and D1X images (towards the bottom of the page),
most of the D100 EXIF information has been stripped. This indicates
that either the camera is still very pre-production or that the
images were filtered through yet another image editing program
before posting.

That said, here are the opinions I am hearing in this forum:

1. The Nikon D100 gives you less noise at most all ISO's
2. The Nikon D1X has greater color accuracy and saturation
3. The Canon D60 has the greatest resolution and detail

What are your opinions--- and what is the best tradeoff for your
printing needs?

--
Clint
http://www.pbase.com/moviebear

 

[Tau]

Movie Bear,

in most pictures the lens is set at F2.8, which is wide open or nearly wide open? Can it be that the images will become more soft then? I'd like to see some pictures with F8 or F16. Some pictures shot with F5.6 are considerable more sharp. What do you, as an expert think? I don't own a digital SLR yet.
Jack.

Tonight, I once again downloaded each of the house images for the
D60, D100 and D1X from the Japanese web site. All of these were
shot according to the lowest ISO settings. This means that the
images should show the best low noise results from each camera.

Forget Nyquist Frequencies-- forget Lines p/p per square root,
forget Quantum physics.

Loaded all images into my new PS7 at the same time. Visually
reduced each one of them to 30% so I can see them side by side on
my Apple Studio display. Before printing, I started to examine the
obvious things first. I never went above 100% in my views. There
were clearly some findings to mention that were obvious WAY before
the printing stage.

Finding #1: The D1X sensor was dirty with a spot smack in the
middle of the blue sky!

Finding #2: As I perused each square inch of the image at 100%, it
was obvious that these cameras take WAY different approaches to
imaging.

Finding #3: The D60 and D1X are FAR more similar than they are
different. The D100 image looks nothing like either the D60 or D1X.

Finding #4: Hands down (and I mean WAY DOWN) for shadow detail and
low noise at these ISO settings, the D60 is superior to the D1X and
vastly superior to the D100. Look under each ledge in the house--
look at the glass-- look at the shadows being thrown by the shrubs.
In every instance, the D100 has that "consumer cam" look to it. Try
this yourself.... it is so obvious that no amount of spec talk will
convince me otherwise.

Finding #5: For sharpness and even clarity accross the entire
photograph, the D1X and D60 are so close its' a toss up. The tile
roof is so clear on both the D1X and D60 that it is almost
palpable. Curves, contrasts, and tonal graduations on the D1X and
D60 roof tiles is spot on similar. The D100 fared very poorly
here-- a very muddied appearance on the entire roof. Its as if
there was a thin gauze over it. I repeat, do this yourself. It's
obvious.

Finding #6: For bright subjects, like the surface of the home
itself, the sharpness of all 3 cameras appears to be similar. But
that similarity ends very quickly...

Finding #7: Take a look (even at 50%) of the trees and shrubs on
the left side of the house. Both the D1X and D60 manage this level
of detail very well-- with the D1X having superior tonal control
and definition. The D100 falls flat. Complete loss of detail here
in the shrubs. Not an opinion. Just downright bad. Look at this for
yourself.

After printing 8 x 10 images of all 3 (without alteration), I
presented each photo to my rep at my local camera store. I already
knew my findings... but I wanted a blind test.

I have to tell you folks, it was no contest. The D100 photo of the
house (while OK in terms of digital standards) looked like what
could be expected from a Coolpix 6MP camera. I know that sounds
harsh, but it is so obvious. Very thinly vieled-- soft lines-- poor
green channel focus-- and muddied shadows.

Both the D60 and D1X produced photos that were breathtaking in all
areas. My rep, Dan agreed. We had a hard time telling the
difference between the D1X and D60.

I have no passion for either Nikon or Canon. I believe you'll have
to work very hard, however, to make the D100 look anything like the
output of the D1X or D60. I'll state this: All my observations and
conclusions are solely based upon the images posted at the Japanese
web site-- and I fully understand that these may not be indicative
of the final quality of this camera.

My conclusions are not wild opinions. They are readily available
for you to view and print and I hereby stand on my feelings that
this model D100 is not ready for prime time with the images that
are posted.

--

The significant problems we face can not be solved at the same level of thinking we were at when we created them.
(A.E.)

 

[Ron Parr]

A few things here. First, I saw better shadow detail perservation
from the D60, e.g. the boy's neck.

Again, we disagree. I found areas where the D100 shadow detail was
better than the D60's we can go back and forth on this both cameras

OK. Describe some.

Second, unless you know more
than we do about the transistor size in the D60 and the effective
fill rate for the D100, I don't see how you can draw any
conclusions based upon the difference in pixel size. Keep in mind
that the photodiodes are made of different materials as well, so
you'd need to know about the quantum efficiency of the different
processes. You're comparing apples and oranges and concluding that
you'll squeeze more juice from a particular apple because it's a
little bigger than a particular orange.

I never said that the D100 would be better only that it would make
sense for it to. The starting physical advantage of collecting more
light by virtue of photosites with slightly larger size would if
engineered correctly , translate into better DR performance than

No. The D100 uses an interline transfer CCD, which sacrifices a certain amount of space to the the control lines and charge transfer area (no transistors). The D60 sensor is an active pixel CMOS sensor that has control lines, and probably 3-4 transistors per pixel (no charge transfer area). Unless you know something about the interline transfer mechanism of the D100 and the transistor count and size of the D60, you simply can't make any comments about the active area of each pixel. This has nothing to do with whether one is "engineered correctly" or not.

the D60 into high ISO's. This is exactly what is being displayed
in these shots, the D100 noise remains very low and slowly rises.
As I stated else where quantification of noise levels of SD patches
taken from color charts will show exactly how good the response
is. We'll get those in a few weeks hopefully when our reviews get
production cameras to test. My statement did imply automatically
better results for the D100 by virtue of its larger photosites,
which is not necessarily the case but appears to be confirmed by
the performance in these soft beta images.

As I explained earlier, your empricial evidence is just wrong. The D100 shots looks less noisy at high ISO just because they're soft. Your a priori reasoning also has no basis.

Maybe so on the cause for the D60 moire, but I am positive that the
images on the first link page where taken with a soft lens. The
softness noted in those images is gone from the many images at the
other batch of links. Despite the percieved issues, these results
are still excellent for a beta camera.

The links on the first page used several lenses (at least 5), so your "soft lens" theory doesn't hold any water unless you think they had a camera bag full of soft lenses. The images on the other pages look soft too. It's just that you don't have side by side shots with good equipment highlight deficiencies. The long exposure shots are all also soft compared to the D60 shots.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[Tau]

Bob,

I noticed all lenses are used wide open? Can this make the images morte soft? Also i noticed that the resolution chart does not show artifacts when High sharpening is used. The d60 chart does show less sharp, has color fringing from 6 (where D100 is clean), shows stepping (where D100 is smooth), and shows odd black and white pixels.

I wonder what happen if High sharpening setting is used as it seem not to introduce artifacts. The default setting may be too soft. Also I wonder to see pictures with F8, F16, etc.
Jack.

Lets be patient and give Nikon the oppurtunity to finish their
work. They have the competition in their hands and will do you
proud.

Remember thousands of people are reading these posts and often
making purchasing decisions based on the conclusions drawn by other
members. Nikon hasn't been doing great financially and I'm sure
they know this camera needs to perform. I havent looked at the beta
images but I think the D100 is by far the best since it got the
price dropped on my D60 (now if they would just ship it to me).

Cheers

--

The significant problems we face can not be solved at the same level of thinking we were at when we created them.
(A.E.)

 

[Ron Parr]

The explanation for this and other shortcomings evident in the
images from this preproduction D100 may lie in the post below.
Seems there may have been several different sensors in different
D100 test cams, including a Coolpix sensor!

It's clear that a CoolPix sensor was not used for these shots. Such a sensor would have a 4X multiplier. Look at the reported focal lengths and the field of view - there's just no way that these could be enlargements from a CoolPix sensor.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[Ron Parr]

Finding #5: For sharpness and even clarity accross the entire
photograph, the D1X and D60 are so close its' a toss up. The tile
roof is so clear on both the D1X and D60 that it is almost
palpable. Curves, contrasts, and tonal graduations on the D1X and
D60 roof tiles is spot on similar. The D100 fared very poorly
here-- a very muddied appearance on the entire roof. Its as if
there was a thin gauze over it. I repeat, do this yourself. It's
obvious.

Again, soft lenses do that to images. Many of the other images on
this page and the second batch of links doesn't show this, look at
the res. charts the D100 is resolving better than the D60 shot in
every direction! How's that for a test.

They used at least 5 different lenses on that first page. Were they all soft?

IMO, the other batches are also soft.

The ultimate test will be controlled tests of noise and resolution
on color charts that will show weather or not it's you who are
hallucinating or weather it's every one else. Facts: D100 high ISO

There's a minority of people who are willfully trying to ignore how bad the D100 beta images are. Anybody can make a soft image without noise and you can always apply a Gaussian blur to the D60 or D1X images if you want that D100 look.

Ultimately, these images show that the D100 will produce some kind of output at high ISO and I expect that with tweaking the high ISO output will be usable for some applications. The question is whether you're willing to give up good ordinary pictures to get this capability. Hopefully Nikon will improve things before the final release so that people won't be forced to make such a choice.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[Gordon Berg]

Before we read too much into Nikons performance at high ISO's or saying they beat the D60 for noise at comparable ISO's I think we ought to look at the actual EV information in the Exif data and see whether Nikon's ISO ratings are accurate. In some cases we also have check the image brightness to make sure we are doing a good comparison. I did some checking of such information and it showed Nikon overrates its ISOs but there are several that I didn't check, which would give a larger sample size to go on if others did.

Below is a copy of what I posted in Hiro's original thread. (The first comparison is of the two gray buildings with buses and taxis in front and the second of fruit stand at night (bright sign says FRUTTA something), from another site.)

Looking at the photos at the bottom where the D100, D1x and D60 were compared I checked the Exif data to see how the EV numbers compared across the board.

The first thing I found was that the Nikon cameras didn't give the ISO number in the EXIF data while the Canon did - and sure enough, the D60 pic labelled ISO 1600 was properly listed in the Exif as ISO 1000.

The exposure values:
D100 F11.0 4/10000
D1x F5.6 2/10000
D60 F11.0 5/10000

So the two Nikon camera's EV's were in line with their ISO numbers - the D100 getting 1 stop less light. On the other hand you'd expect the D60 to require almost a stop more than the D100 but it was only 1/4 of a stop more and it's picture is a good half of a stop brighter!! (The two Nikon photos are very close in brightness.) So from this limited sample it looks like Nikon rates it's ISO more than a stop higher than Canon. If I get ambitious with my 56K modem I'll download further pics from that series and see if that continues to hold.

On another one of the sites where the d100 ISO 3200 & 6400 shots appeared I checked the EXIF data for the ISO 6400 versus the ISO 200 shot. The apetures were the same at 3.5 and the two times were .0040 vs. .0667 which give a time ratio of 16.7:1 - instead of the 32:1 ratio you'd expect from the ISO numbers. To make it worse the ISO 200 photo is somewhat brighter. (Somebody thought it was quite hilarious a few months ago when I said I was concerned that the Nikon marketeers set their ISO ratings instead of the Nikon engineers.) A few pics don't tell the whole story and these weren't labratory conditions so maybe we'll see data indicating they did spec the camera accurately. And it's a Beta version. I hope so - right now, it looks a bit like the D100's almost 3 stop advantage in ISO rating over the D60 is two thirds marketing hype...

 

[Gordon Berg]

You beat me to it...

Some people have also been making the point that the shots were taken in early January but if you look at the Exif data on all three cameras for the 'gray buildings with taxis and buses' you will see that the D100 Exif date is 1/1/2002 while the other two cameras show 5/5/2002 - and it's pretty clear all were shot in a short span of time. Someone else mentioned a case where the pattern of dates was 1/2/2002 vs 5/6/2002... ie same relative dates.

The explanation for this and other shortcomings evident in the
images from this preproduction D100 may lie in the post below.
Seems there may have been several different sensors in different
D100 test cams, including a Coolpix sensor!

It's clear that a CoolPix sensor was not used for these shots.
Such a sensor would have a 4X multiplier. Look at the reported
focal lengths and the field of view - there's just no way that
these could be enlargements from a CoolPix sensor.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[Clint Thayer #32755]

Movie Bear,

in most pictures the lens is set at F2.8, which is wide open or
nearly wide open? Can it be that the images will become more soft
then? I'd like to see some pictures with F8 or F16. Some pictures
shot with F5.6 are considerable more sharp. What do you, as an
expert think? I don't own a digital SLR yet.

I don't know what's going on here-- and I'm not an engineer or an expert. Yes, there are shots at the site stepped down which look considerably sharper- because of increased DOF.

Jack.

Tonight, I once again downloaded each of the house images for the
D60, D100 and D1X from the Japanese web site. All of these were
shot according to the lowest ISO settings. This means that the
images should show the best low noise results from each camera.

Forget Nyquist Frequencies-- forget Lines p/p per square root,
forget Quantum physics.

Loaded all images into my new PS7 at the same time. Visually
reduced each one of them to 30% so I can see them side by side on
my Apple Studio display. Before printing, I started to examine the
obvious things first. I never went above 100% in my views. There
were clearly some findings to mention that were obvious WAY before
the printing stage.

Finding #1: The D1X sensor was dirty with a spot smack in the
middle of the blue sky!

Finding #2: As I perused each square inch of the image at 100%, it
was obvious that these cameras take WAY different approaches to
imaging.

Finding #3: The D60 and D1X are FAR more similar than they are
different. The D100 image looks nothing like either the D60 or D1X.

Finding #4: Hands down (and I mean WAY DOWN) for shadow detail and
low noise at these ISO settings, the D60 is superior to the D1X and
vastly superior to the D100. Look under each ledge in the house--
look at the glass-- look at the shadows being thrown by the shrubs.
In every instance, the D100 has that "consumer cam" look to it. Try
this yourself.... it is so obvious that no amount of spec talk will
convince me otherwise.

Finding #5: For sharpness and even clarity accross the entire
photograph, the D1X and D60 are so close its' a toss up. The tile
roof is so clear on both the D1X and D60 that it is almost
palpable. Curves, contrasts, and tonal graduations on the D1X and
D60 roof tiles is spot on similar. The D100 fared very poorly
here-- a very muddied appearance on the entire roof. Its as if
there was a thin gauze over it. I repeat, do this yourself. It's
obvious.

Finding #6: For bright subjects, like the surface of the home
itself, the sharpness of all 3 cameras appears to be similar. But
that similarity ends very quickly...

Finding #7: Take a look (even at 50%) of the trees and shrubs on
the left side of the house. Both the D1X and D60 manage this level
of detail very well-- with the D1X having superior tonal control
and definition. The D100 falls flat. Complete loss of detail here
in the shrubs. Not an opinion. Just downright bad. Look at this for
yourself.

After printing 8 x 10 images of all 3 (without alteration), I
presented each photo to my rep at my local camera store. I already
knew my findings... but I wanted a blind test.

I have to tell you folks, it was no contest. The D100 photo of the
house (while OK in terms of digital standards) looked like what
could be expected from a Coolpix 6MP camera. I know that sounds
harsh, but it is so obvious. Very thinly vieled-- soft lines-- poor
green channel focus-- and muddied shadows.

Both the D60 and D1X produced photos that were breathtaking in all
areas. My rep, Dan agreed. We had a hard time telling the
difference between the D1X and D60.

I have no passion for either Nikon or Canon. I believe you'll have
to work very hard, however, to make the D100 look anything like the
output of the D1X or D60. I'll state this: All my observations and
conclusions are solely based upon the images posted at the Japanese
web site-- and I fully understand that these may not be indicative
of the final quality of this camera.

My conclusions are not wild opinions. They are readily available
for you to view and print and I hereby stand on my feelings that
this model D100 is not ready for prime time with the images that
are posted.

--
The significant problems we face can not be solved at the same
level of thinking we were at when we created them.
(A.E.)

--
Clint
http://www.pbase.com/moviebear

 

[David SL]

A few things here. First, I saw better shadow detail perservation
from the D60, e.g. the boy's neck.

Again, we disagree. I found areas where the D100 shadow detail was
better than the D60's we can go back and forth on this both cameras

OK. Describe some.

The boys neck? The boy is larger in the frame and he's in a slightly different position in the D60 shot. You can't make any determination of shadow detail because of these differences. I'll rescind my statement somewhat, upon closer inspection I can't find areas where the D100 resolves more detail BUT I still see no difference between the resolved detail(which was implied by my use of "better" above) in the comparison images, they both resolve similar detail into shadows with the D60 image rendering that detail sharper but I never denied that.

I never said that the D100 would be better only that it would make
sense for it to. The starting physical advantage of collecting more
light by virtue of photosites with slightly larger size would if
engineered correctly , translate into better DR performance than

No. The D100 uses an interline transfer CCD, which sacrifices a
certain amount of space to the the control lines and charge
transfer area (no transistors). The D60 sensor is an active pixel
CMOS sensor that has control lines, and probably 3-4 transistors
per pixel (no charge transfer area). Unless you know something
about the interline transfer mechanism of the D100 and the
transistor count and size of the D60, you simply can't make any
comments about the active area of each pixel. This has nothing
to do with whether one is "engineered correctly" or not.

Oh I get it, so the selection of impedance characteristics, line width and signal processing used on the CCD/CMOS chips has nothing to do with engineering to increase overall S/N ? It makes no sense to you that these parameters would be adjusted to maximize the S/N ratio ? As an electronics engineer the statement you made is laughable. Listen to what you are saying Ron, stop trying to split hairs.

the D60 into high ISO's. This is exactly what is being displayed
in these shots, the D100 noise remains very low and slowly rises.
As I stated else where quantification of noise levels of SD patches
taken from color charts will show exactly how good the response
is. We'll get those in a few weeks hopefully when our reviews get
production cameras to test. My statement did imply automatically
better results for the D100 by virtue of its larger photosites,
which is not necessarily the case but appears to be confirmed by
the performance in these soft beta images.

As I explained earlier, your empricial evidence is just wrong. The
D100 shots looks less noisy at high ISO just because they're soft.
Your a priori reasoning also has no basis.

Softness of a lens doesn't affect noise appearance in a final image, remember the sensor forms the digital image from what ever incident light it gets from the lens. Noise is a sensor artifact. I don't know of any cameras that have been suspected or critisized for softening images outside of the affects of the lens attached, yet this is what you are asserting for the D100 images here and that is beyond wrong its rediculous.

Maybe so on the cause for the D60 moire, but I am positive that the
images on the first link page where taken with a soft lens. The
softness noted in those images is gone from the many images at the
other batch of links. Despite the percieved issues, these results
are still excellent for a beta camera.

The links on the first page used several lenses (at least 5), so
your "soft lens" theory doesn't hold any water unless you think
they had a camera bag full of soft lenses. The images on the other
pages look soft too. It's just that you don't have side by side
shots with good equipment highlight deficiencies. The long
exposure shots are all also soft compared to the D60 shots.

Your perception of those pages versus mine, I think the other pages where quite exceptably sharp. Let's leave it at that. As I said before, the production camera under empirical tests will bring the truth to light.

Regards,

--

 

[Jim K53618]

Gordon, while I appreciate your attempt to investigate the exposures more closesly, I'm affraid your findings are inconclusive and can not be used to prove any point. Remember that lighting conditions can change dramatically out side in just a few seconds. To really see what any differences are from these cameras you must first see how each camera's meter is compared to a remote (high quality) hand held meter, there may be variations in the way they meter. For example under certain conditions the D60 purposely underexposes by -0.05EV to maintain highlights without blowing them out. Also the cameras need to be compared inside under identical conditions with equal lenses to make any conclusive statements. I don't mean to insult your simple tests, but must remind everybody that there is a lot more to comparisons then downloading images from web pages and checking the EXF headers. Also we are looking at a preproduction model. I would personally hold off from any serious comparisons or tests until we have samples from a production model, and even then I would be careful in comparisions because there are a lot of factors to consider and unless you have the camera's in your posession and can duplicate the identical conditions and shots for each camera, comparisons have little true value.

Regards,
Jim K

 

[Ron Parr]

The boys neck? The boy is larger in the frame and he's in a
slightly different position in the D60 shot. You can't make any
determination of shadow detail because of these differences. I'll

It's true that the illumination could be different, but I find the dropoff here to be typical of the harsh shadow dropoff in other D100 shots as well, e.g., the house. I suspect that there it little hope of us seeing eye to eye on this point...

No. The D100 uses an interline transfer CCD, which sacrifices a
certain amount of space to the the control lines and charge
transfer area (no transistors). The D60 sensor is an active pixel
CMOS sensor that has control lines, and probably 3-4 transistors
per pixel (no charge transfer area). Unless you know something
about the interline transfer mechanism of the D100 and the
transistor count and size of the D60, you simply can't make any
comments about the active area of each pixel. This has nothing
to do with whether one is "engineered correctly" or not.

Oh I get it, so the selection of impedance characteristics, line
width and signal processing used on the CCD/CMOS chips has nothing
to do with engineering to increase overall S/N ? It makes no sense
to you that these parameters would be adjusted to maximize the S/N
ratio ? As an electronics engineer the statement you made is
laughable. Listen to what you are saying Ron, stop trying to split
hairs.

Split hairs? I'm saying that you know nothing about the size of the photsensitive area of each chip and should refrain from making conclusions based only upon the cell size until you have more information. I explained all of the reasons why there can be large differences in the photoreceptive area for different types of sensors. You ignore all this, wave around your "electronics engineer" status and say all of the factors you're failing to consider are laughable? Oh please... The only thing that's laughable here is that you expect me or anybody else to be impressed by this.

Softness of a lens doesn't affect noise appearance in a final
image, remember the sensor forms the digital image from what ever
incident light it gets from the lens. Noise is a sensor artifact.

It's obvious that softness is not from the lenses since they used a lot of different lenses.

I don't know of any cameras that have been suspected or critisized
for softening images outside of the affects of the lens attached,
yet this is what you are asserting for the D100 images here and
that is beyond wrong its rediculous.

It's just a fact. Open your eyes! BTW, image processing and sensor issues have a big effect on the overall sharpness and quality of an image. Ever wonder why so many cameras with seemingly identical lenses have different performance? Compare the G2 with the S85. The G2 has less noise, but lower resolution chart performance. The tradeoff between noise and sharpness isn't unusual; it's at the core of what digital imaging is all about. Phil notes how the Sony F707 images get softer at high ISO:

http://www.dpreview.com/reviews/sonydscf707/page12.asp

What's unusal is how soft the D100 images are even at low ISO.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[Erik Magnuson]

I'm affraid your findings are inconclusive
and can not be used to prove any point.

Of course they are inconclusive. The only "point" he is trying to prove is one that Phil is beginning to address in some of his reviews: namely that there does not seem to be a great deal of consistency in "ISO equivalence" between vendors even under "controlled" conditions. So merely comparing specs here is perhaps just as meaningless as comparing EXIF exposure values.

For example under certain conditions the D60 purposely
underexposes by -0.05EV to maintain highlights without blowing them
out.

You keep repeating this. Don't you have a decimal place loose somewhere? (e.g. -0.5EV?) Neither the shutter nor the aperture control of a D60 is capable of anything finer than 1/3 stop.

--
Erik
Free Windows JPEG comment editor
http://home.cfl.rr.com/maderik/edjpgcom

 

[David SL]

Jim,

Thanks for bringing some sanity to this discussion, it's obvious that some of the images(the soft ones) aren't perfect but they are far from the horrible showing that some people are making them out to be. As you, I am waiting for controlled tests on res. and color charts on a production camera before I make any conclusions, but IMHO in their present state they've already fostered close examination and debate to determine weather or not they are comparable to the best on the market. I think that in itself is a very good sign that Nikon has the competitions attention.

Just a few more weeks.

Gordon, while I appreciate your attempt to investigate the
exposures more closesly, I'm affraid your findings are inconclusive
and can not be used to prove any point. Remember that lighting
conditions can change dramatically out side in just a few seconds.
To really see what any differences are from these cameras you must
first see how each camera's meter is compared to a remote (high
quality) hand held meter, there may be variations in the way they
meter. For example under certain conditions the D60 purposely
underexposes by -0.05EV to maintain highlights without blowing them
out. Also the cameras need to be compared inside under identical
conditions with equal lenses to make any conclusive statements. I
don't mean to insult your simple tests, but must remind everybody
that there is a lot more to comparisons then downloading images
from web pages and checking the EXF headers. Also we are looking at
a preproduction model. I would personally hold off from any serious
comparisons or tests until we have samples from a production model,
and even then I would be careful in comparisions because there are
a lot of factors to consider and unless you have the camera's in
your posession and can duplicate the identical conditions and shots
for each camera, comparisons have little true value.

Regards,
Jim K

--

 

[David SL]

Split hairs? I'm saying that you know nothing about the size of
the photsensitive area of each chip and should refrain from making
conclusions based only upon the cell size until you have more
information. I explained all of the reasons why there can be large
differences in the photoreceptive area for different types of
sensors. You ignore all this, wave around your "electronics
engineer" status and say all of the factors you're failing to
consider are laughable? Oh please... The only thing that's
laughable here is that you expect me or anybody else to be
impressed by this.

I didn't conclude anything other than saying that there was a good likely hood it's performance at highier ISO's could be better, both CCD and CMOS sensors specify active area's to gather light regardless of type of sensor each uses. For both, the max. S/N is directly correlated to the gathering area, from that basis and knowing the gathering area of each photosite, one can get an idea (assuming good engineering retains maximum signal from the gathered area) of which will most likely perform better into high ISO's. If you dont understand that I can't help you. I didn't make a conclusion, just pointing out the relevant likelyhoods. Your reasons didn't add anything to the argument against my point, that's why they were ignored.

Softness of a lens doesn't affect noise appearance in a final
image, remember the sensor forms the digital image from what ever
incident light it gets from the lens. Noise is a sensor artifact.

It's obvious that softness is not from the lenses since they used a
lot of different lenses.

I'll just say that it's interesting the same softness bug attacked the D1x shots. Now as far as I know that camera never had an issue with soft images did it? Could it be they used the same lenses for the two Nikon cameras on this comparison page? hmmmmm....Also, your theory fails to explain the other pages having many very sharp images. (Dolls shots, Res. Chart shots, Nightscene and Boat shots. Garden shot, Clown Shots..etc.)

I don't know of any cameras that have been suspected or critisized
for softening images outside of the affects of the lens attached,
yet this is what you are asserting for the D100 images here and
that is beyond wrong its rediculous.

The tradeoff between noise and sharpness isn't
unusual;

Obvious. But I never argued otherwise. You said they were less noisy because of softness but if there are examples(sited above) of sharp images at all ISO's how can the problem be the sensor?

What's unusal is how soft the D100 images are even at low ISO.

Only on a few of the images on the first page to me. Did you look at the resolution test charts? Explain how this soft sensor can resolve equal or better detail in every direction as shown clearly in the charts? Where the D60 shows jaggies and moire, the D100 is smooth.(Note: block diagonals near bottom, and center circle lines) The D60 tends to start artifacts way before the D100 when compared to any of its 3charts. Even if the "softness" you see is not a factor of the lens, the D100 at least on this res. chart test is displaying generally greater resolving capability than the D60 at lowest ISO of 200 compared to 100 for the D60. (Where your theory says it should be sharpest.) If a beta result can be this impressive the production result will truly be remarkable maybe then your eyes will open and finally see it.

Regards,

--

 

[Jim K53618]

I'm affraid your findings are inconclusive
and can not be used to prove any point.

Of course they are inconclusive. The only "point" he is trying to
prove is one that Phil is beginning to address in some of his
reviews: namely that there does not seem to be a great deal of
consistency in "ISO equivalence" between vendors even under
"controlled" conditions. So merely comparing specs here is perhaps
just as meaningless as comparing EXIF exposure values.

I though that was what I was saying. And why I mentioned the need to check the metering and how it meters. The poster seemed to be making premature conclusions to me based on EXIF values. Perhaps I missed his point.

For example under certain conditions the D60 purposely
underexposes by -0.05EV to maintain highlights without blowing them
out.

You keep repeating this. Don't you have a decimal place loose
somewhere? (e.g. -0.5EV?) Neither the shutter nor the aperture
control of a D60 is capable of anything finer than 1/3 stop.

Yes you are correct, I mean to say -0.5 EV. How is repeating this a bad thing? It demonstrates how some camera's purposely control exposure to prevent blown out highlights, some have mistaken this as a "bug" or malfunction.

Regards,
Jim K

 

[Erik Magnuson]

I though that was what I was saying. And why I mentioned the need
to check the metering and how it meters. The poster seemed to be
making premature conclusions to me based on EXIF values. Perhaps I
missed his point.

I think the point is people are drooling over "ISO 6400". But note the specs only call it HI-2 ( ISO 6400), e.g. not even Nikon is claiming a true 6400. Between the EXIF evidence and the known issues with comparing ISO values between vendors, we should take these numbers with a grain of salt (as opposed to salt-sized grain!) Why am I reminded of that scene in Spinal Tap where the amp is louder because it goes all the way to 11?

Yes you are correct, I mean to say -0.5 EV. How is repeating this a
bad thing?

Repeating the typo was a confusing thing. The first time I saw it, I just assumed it was a typo. When I saw it repeated here (I think this was the 3rd time), I thought you were serious and wanted to know how you got that improbable number.

--
Erik
Free Windows JPEG comment editor
http://home.cfl.rr.com/maderik/edjpgcom

 

[Ron Parr]

I didn't conclude anything other than saying that there was a good
likely hood it's performance at highier ISO's could be better, both
CCD and CMOS sensors specify active area's to gather light

No they don't! This is what I keep trying to tell you! In the case of interline transfer CCDs up to 50% of each photocell can be consumed by the charge transfer area. In the case of CMOS, up to 50% can be consumed by transistors. This is not reported.

regardless of type of sensor each uses. For both, the max. S/N is
directly correlated to the gathering area, from that basis and
knowing the gathering area of each photosite, one can get an idea
(assuming good engineering retains maximum signal from the gathered
area) of which will most likely perform better into high ISO's. If
you dont understand that I can't help you. I didn't make a
conclusion, just pointing out the relevant likelyhoods. Your
reasons didn't add anything to the argument against my point,
that's why they were ignored.

You're just ignoring what you don't understand, David. Drop the attitude and learn something about sensors.

I'll just say that it's interesting the same softness bug attacked
the D1x shots. Now as far as I know that camera never had an issue
with soft images did it? Could it be they used the same lenses for
the two Nikon cameras on this comparison page? hmmmmm....Also, your

On the contrary, the remarkable sharpness of the D1x shots in comparison to the D100, despite the D1x's interpolation, really proves the point that the D100 is having major problems. Look at the plants in the D1x shots. They're clear and sharp. In the D100 shots, they're fuzz balls.

theory fails to explain the other pages having many very sharp
images. (Dolls shots, Res. Chart shots, Nightscene and Boat shots.
Garden shot, Clown Shots..etc.)

I'm not impressed with these shots. I suspect you find them impressive because you haven't seen good shots side by side.

The resolution chart shots don't impress me either. The D100 has no more resolution, although it does do a better job of avoiding Bayer artifiacts, further evidence of heavy anti-aliasing filter.

Obvious. But I never argued otherwise. You said they were less
noisy because of softness but if there are examples(sited above) of
sharp images at all ISO's how can the problem be the sensor?

There are no examples of sharp images at any ISO for the D100.

Only on a few of the images on the first page to me. Did you look
at the resolution test charts? Explain how this soft sensor can

Yes.

resolve equal or better detail in every direction as shown clearly
in the charts? Where the D60 shows jaggies and moire, the D100 is

It's not equal or better. At the limit, the D100 just gets muddy, while the the D60 shows artifacts. If you're actually counting lines, you can go further wit the D60.

The D60 has more artifacts because it is probably using a less agressive anti-aliasing filter and a different Bayer interpolation algorithm. The sharpening level for the D60 shot is also greater, which will tend to emphasize and even create artifacts.

smooth.(Note: block diagonals near bottom, and center circle lines)
The D60 tends to start artifacts way before the D100 when compared
to any of its 3charts. Even if the "softness" you see is not a
factor of the lens, the D100 at least on this res. chart test is
displaying generally greater resolving capability than the D60 at

No it isn't. The D100 is trading fuzziness for avoidance of color artifacts.

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/

 

[David SL]

Ron,

Man you last longer than that pink rabbit on those TV commercials!

There's only one thing that I have the breath (or is it finger flexibility) to respond to. The light gathering of CCD and CMOS, as I said before both produce more signal when the sensors (photodiodes) have greater receptive area. Both rely on the photo electric affect, and that affect produces more photocurrent depending on area of incident light, this can't be denied. If the CMOS doesn't work this way then how does it convert the light to electrical energy?

Whatever the percieved deficiencies of the D100 images, the fact remains that those were beta images and improvement is almost sure to follow.

--

 

[Ron Parr]

There's only one thing that I have the breath (or is it finger
flexibility) to respond to. The light gathering of CCD and CMOS, as
I said before both produce more signal when the sensors
(photodiodes) have greater receptive area. Both rely on the photo
electric affect, and that affect produces more photocurrent
depending on area of incident light, this can't be denied. If the
CMOS doesn't work this way then how does it convert the light to
electrical energy?

David, the charge transfer area in an interline transfer CCD is not photoreceptive. In fact, it is designed to be shielded from light. This is a significant portion of the area of each photocell. Phil's description of this is somewhat flawed, but if you look at his diagram for an interline transfer CCD, you should get some idea:

http://www.dpreview.com/learn/Glossary/Camera_System/Sensor_01.htm

Active pixel CMOS sensors, like the D60's, need 3-4 transistors per pixel in addition to the photodiode. The transistors are not light sensitive, but take up a significant portion of the area of each cell. This is why active pixel CMOS sensors were not practical until transistor sizes got down to 0.5 microns - the transistor would use up so much space that there would not space left for the photodiode.

All of this is why I'm telling you that unless you know a lot more about this than you've shown evidence of so far, you have no business comparing the pixel sizes of active pixel CMOS sensors and interline transfer CCD sensors and making any kind of suppositions about their relative efficiency.

When you figure out the transistor size and count per pixel of Canon's sensor, figure out the photreceptive area of the D100's pixels, and come up with some a comparison of the microlens designs from the different manufacturers, then you'll be qualified to make judgements about which one ought to be more sensitive than the other.

Here are some slides on sensors that can get you started:

http://www.stanford.edu/class/ee392b/handouts/sensors.pdf

--
Ron Parr
FAQ: http://www.cs.duke.edu/~parr/photography/faq.html
Gallery: http://www.pbase.com/parr/