5 mp CCD wrong? (long)

[SFJP]

It seems the crop of new 5 mp digicams is far from receiving an unanimous acclamation from its first users according to the posts in the various forums of dpreview. The Minolta D7, the Sony 707 and now the Olympus E20 seem to be far from meeting the expectations of many of the people who bought them. And it seems also the new Nikon 5000 gives some trouble to Nikon who is chasing the sites displaying preview samples taken with this camera to remove these samples.

What's wrong? Beyond some drawbacks specific to each model of camera body, such as D7 autofocus bad performances for instance, the CCD itself (for now all these cameras share the same 5 mp CCD produced by Sony) seems to be the main source of troubles: the D7 and E20 are said to have an excessive noise even at their lower ISO 100 or 80 setting, the 707 has artificial "electric colors" and quite visible artifacts but much less noise. What's in common in these troubles? in my view, I'd bet it is excessive noise from in the initial pictures coming from the CCD! Yes, the Sony 707 has much less noise than the two others, but only because it applies a strong in-camera processing to remove it after the image is taken, and this might well be the reason of the artifificial look of the pictures it produces (artificial lines emphasizing border between surfaces, artificial looking textures and diificulty to equilibrate colors).

This kind of problem encountered by the new 5 mp consumer level CCD was predicted more than one year ago by Fuji. Fuji argued that 2.5/3.3 mp was the maximum that can be reached from consumer level CCD that have for now a very small surface. To go beyond would be paid by an excessive amount of noise. And what happens today with the dissatisfaction encountered by the new 5 mp consumer/prosumer camera could just prove Fuji was right (while Sony 4 mp CCD seems to raise a little less criticism, but it is right that 4 mp CCD was expected to be just a transition and did not raise as much expectations as did the 5 mp CCD that was presented as the future stable standard for CCD resoution).

And if Fuji's prediction was right, for consumer level CCD with a small surface, the superCCD approach promoted by Fuji can well be the only one reasonable for a while, until CCD with larger area can be produced at reasonable price (to fit with the less than 1,000$ category).

The Fuji 3.3 mp superCCD that can be found in the Finepix 6800 and 6900, produces natively 6 mp pictures that have a resolution on par with Sony 5 mp CCD pictures (or at least very close), but Fuji pictures are rather noise free at ISO 100 and with a very acceptable level of noise (easily removed on computer, see http://www.pbase.com/sfjp/neat_image_noise_removal_tests ) at ISO 200 (ISO 400 not so good, though). And the Fuji colors are just perfect in daylight outdoor pictures. The main defect of the pictures produced by these Fuji cameras comes from the in-camera sharpening that creates visible artifacts at full 6 mp resolution: easy to bypass by using the "soft" sharpening setting instead of the "normal" one. A secondary defect is due to some difficulty to masterize the right WB setting for indoor shots, while with some effort it is possible to find the good solution in most cases (this might rather be a defect of the current Fuji WB algorithm than of the superCCD itself). For shots in dark night, the superCCD exhibits also a significant number of hot pixels for exposures > 1sec., which makes it not very adapted to this kind of use, although hot pixels can now be easily removed by post-processing (or in-camera processing with the ad hoc algorithm not available yet on Fuji cameras).

I think that most comparisons between top mega pixels cameras that conclude to the triumph of the Sony 5mp CCD over the Fuji 3.3 mp superCCD just got it wrong in practice. The 3.3 mp superCCD remains in my opinion the best solution available today for small surface CCD found in the consumer/prosumer price level and might remain for long if the ratio price/surface of CCD does not fall significantly in a short future.

So what? I believe that the Fuji 6800 and 6900 are incredibly underevaluated by the expert reviewers while they provide, except for night shots maybe, the best picture quality available today and may be for a while in the less than 4,000$ camera category (while the 6800 and 6900 are now sold around 500$ in the US!!). The 6800 and 6900 cameras themselves are not perfect and should be enhanced in the short future, with a lens that would remain without distortion in the corners, better night shot ability, a less brutal in-camera sharpening algorithm, a better autofocus and, please mr. Fuji, a better EVF on the 6900 successor. But I would be very sad if for marketing reasons and to comply with the criticismes of the expert reviewers Fuji would drop its superCCD, to follow the majority and use instead the Sony noisy 5 mp CCD, except if Fuji would introduce instead a large surface CCD of its own (or from Philips?).

Jean-Paul
http://www.pbase.com/sfjp

 

[tony gilmour]

Jean Paul, Fuji are one of only a couple of camera manufacturers to produce their own CCD so i doubt they will ever change the formulae except maybe improve on its design. Its a bit like cars......the Citroen suspension has now been licenced to Rolls Royce ! so how long will it be before we see other camera manufacturers clammering to Fuji for their CCDs.

The funniest thing I have noticed with all these people looking to buy new cameras 'they have plenty of choice in the 4 to 5 megapixel range ......so why do they keep comparing these cameras to the Fuji ?

It seems the crop of new 5 mp digicams is far from receiving an
unanimous acclamation from its first users according to the posts
in the various forums of dpreview. The Minolta D7, the Sony 707 and
now the Olympus E20 seem to be far from meeting the expectations of
many of the people who bought them. And it seems also the new Nikon
5000 gives some trouble to Nikon who is chasing the sites
displaying preview samples taken with this camera to remove these
samples.

What's wrong? Beyond some drawbacks specific to each model of
camera body, such as D7 autofocus bad performances for instance,
the CCD itself (for now all these cameras share the same 5 mp CCD
produced by Sony) seems to be the main source of troubles: the D7
and E20 are said to have an excessive noise even at their lower ISO
100 or 80 setting, the 707 has artificial "electric colors" and
quite visible artifacts but much less noise. What's in common in
these troubles? in my view, I'd bet it is excessive noise from in
the initial pictures coming from the CCD! Yes, the Sony 707 has
much less noise than the two others, but only because it applies a
strong in-camera processing to remove it after the image is taken,
and this might well be the reason of the artifificial look of the
pictures it produces (artificial lines emphasizing border between
surfaces, artificial looking textures and diificulty to equilibrate
colors).

This kind of problem encountered by the new 5 mp consumer level CCD
was predicted more than one year ago by Fuji. Fuji argued that
2.5/3.3 mp was the maximum that can be reached from consumer level
CCD that have for now a very small surface. To go beyond would be
paid by an excessive amount of noise. And what happens today with
the dissatisfaction encountered by the new 5 mp consumer/prosumer
camera could just prove Fuji was right (while Sony 4 mp CCD seems
to raise a little less criticism, but it is right that 4 mp CCD was
expected to be just a transition and did not raise as much
expectations as did the 5 mp CCD that was presented as the future
stable standard for CCD resoution).

And if Fuji's prediction was right, for consumer level CCD with a
small surface, the superCCD approach promoted by Fuji can well be
the only one reasonable for a while, until CCD with larger area can
be produced at reasonable price (to fit with the less than 1,000$
category).

The Fuji 3.3 mp superCCD that can be found in the Finepix 6800 and
6900, produces natively 6 mp pictures that have a resolution on par
with Sony 5 mp CCD pictures (or at least very close), but Fuji
pictures are rather noise free at ISO 100 and with a very
acceptable level of noise (easily removed on computer, see
http://www.pbase.com/sfjp/neat_image_noise_removal_tests ) at ISO
200 (ISO 400 not so good, though). And the Fuji colors are just
perfect in daylight outdoor pictures. The main defect of the
pictures produced by these Fuji cameras comes from the in-camera
sharpening that creates visible artifacts at full 6 mp resolution:
easy to bypass by using the "soft" sharpening setting instead of
the "normal" one. A secondary defect is due to some difficulty to
masterize the right WB setting for indoor shots, while with some
effort it is possible to find the good solution in most cases (this
might rather be a defect of the current Fuji WB algorithm than of
the superCCD itself). For shots in dark night, the superCCD
exhibits also a significant number of hot pixels for exposures >
1sec., which makes it not very adapted to this kind of use,
although hot pixels can now be easily removed by post-processing
(or in-camera processing with the ad hoc algorithm not available
yet on Fuji cameras).

I think that most comparisons between top mega pixels cameras that
conclude to the triumph of the Sony 5mp CCD over the Fuji 3.3 mp
superCCD just got it wrong in practice. The 3.3 mp superCCD remains
in my opinion the best solution available today for small surface
CCD found in the consumer/prosumer price level and might remain for
long if the ratio price/surface of CCD does not fall significantly
in a short future.

So what? I believe that the Fuji 6800 and 6900 are incredibly
underevaluated by the expert reviewers while they provide, except
for night shots maybe, the best picture quality available today and
may be for a while in the less than 4,000$ camera category (while
the 6800 and 6900 are now sold around 500$ in the US!!). The 6800
and 6900 cameras themselves are not perfect and should be enhanced
in the short future, with a lens that would remain without
distortion in the corners, better night shot ability, a less brutal
in-camera sharpening algorithm, a better autofocus and, please mr.
Fuji, a better EVF on the 6900 successor. But I would be very sad
if for marketing reasons and to comply with the criticismes of the
expert reviewers Fuji would drop its superCCD, to follow the
majority and use instead the Sony noisy 5 mp CCD, except if Fuji
would introduce instead a large surface CCD of its own (or from
Philips?).

Jean-Paul
http://www.pbase.com/sfjp

 

[SFJP]

Eric,

I think this is simple while people keep comparing the new 4-5 mp CCD cameras with the FuJI 3.3 mp superCCD cameras: whatever the necessary improvements these cameras should have to become fully satsifying, the image quality of the Fuji cameras, in term of resolution/color/absenc of noise, remains just the best so far of the less than 4,000$ cameras! And they go to the forums dedicated to the manufacturers producing 4-5 mp CCD cameras (Minolta, Sony, Canon, Nikon, etc), they see all these posts of disappointed owners, then they go to our beloved Fuji forum, and they see there all these posts of delighted owners that like so much taking and showing pictures from their Fuji superCCD cameras (mainly owners of 4900/6900) that they all have public galleries now and that even some are becoming professional while they were not before.... Do you agree with my answers?

Jean-Paul
http://www.pbase.com/sfjp

The funniest thing I have noticed with all these people looking to
buy new cameras 'they have plenty of choice in the 4 to 5 megapixel
range ......so why do they keep comparing these cameras to the Fuji
?

 

[Leon Wittwer]

I have been using my E20 for something over a week and I am not disappointed. Sorry to disappoint you. I also have an E10 and have used a number of consumer digital cameras. The color of my Oly images is outstanding. The noise levels of my E20 seem a bit less than my E10 and the "look" of the E20 images seems better than the E10. It is true that noise levels of the 4 and 5 megapixel cameras are somewhat more than earlier models. The best I ever had was the Kodak DC290 which had very low noise. In most cases, the noise levels of my E10 or E20 are not an issue for me as the noise does not seem to carry though in prints or in downsampled images for the web. Noise can an issue when a shot is significantly underexposed or sometimes when I am blowing up an image. At ISO 80 and 160, it is pretty easy to suppress the noise while maintaining good image quality. Over time I have come to prefer to do my own noise processing. Agressive incamera processing, while looking nice straight out of the camera, also suppresses fine scale low contrast textures such as seen in fabrics, brickwork, etc. Be doing my own processing, I can selectively reduce the noise where I want to and to the level I want and preserve detail that is lost with most incamera processing. I realize my approach is not suitable for those who want the best final image right out of the camera or don't have the time to process images. In summary, some noise can be advantageous to the final photograph.

I think this is simple while people keep comparing the new 4-5 mp
CCD cameras with the FuJI 3.3 mp superCCD cameras: whatever the
necessary improvements these cameras should have to become fully
satsifying, the image quality of the Fuji cameras, in term of
resolution/color/absenc of noise, remains just the best so far of
the less than 4,000$ cameras! And they go to the forums dedicated
to the manufacturers producing 4-5 mp CCD cameras (Minolta, Sony,
Canon, Nikon, etc), they see all these posts of disappointed
owners, then they go to our beloved Fuji forum, and they see there
all these posts of delighted owners that like so much taking and
showing pictures from their Fuji superCCD cameras (mainly owners of
4900/6900) that they all have public galleries now and that even
some are becoming professional while they were not before.... Do
you agree with my answers?

Jean-Paul
http://www.pbase.com/sfjp

The funniest thing I have noticed with all these people looking to
buy new cameras 'they have plenty of choice in the 4 to 5 megapixel
range ......so why do they keep comparing these cameras to the Fuji
?

 

[SFJP]

I am happy (not at all disappointed) to see an owner of the E10 to be happy with his new E20, I was wondering what Olympus did so wrong in trying to enhance the E10 with the E20 to raise so much dissatisfaction in the Oly forum (such as people reselling their new E20 and keeping their old E10)... I am hoping that you are the first of a long serie of people that would use this thread to try to contradict my initial assertions, just to say how much they are happy with their Oly, Sony, Minolta, etc. My intention is not to bash other cameras that I do not own, but in the contrary to make people express their satisfaction when they are satisfyed and also to attract the attention of all on the incredible quality of the Fuji superCCD, too much neglected and underestimated by the reviewers.

Yes, noise is not so much of a concern when you target standard size prints (say less or equal to A4) with a 4-5 mp camera using an average consumer printer. If you want more you have to eliminate noise by some post processing if noise is present, but the more noise you have in your picture the more noise elimination will cost you in image quality. I am using Neat Image to clean my ISO 200 pictures (I have no noise at all in my ISO 100 images) that have some slight noise and the results are great (see http://www.pbase.com/sfjp/neat_image_noise_removal_tests ). What do you use to clean the noise of your E10/E20 pictures?

Otherwise, I fully agree with you on trying to disable all in-camera processing to get the less altered original picture and then to use instead post-processing on computer. For instance, the superCCD images taken by the Fuji 6900 are much, much better when using the "soft" sharpening setting (in fact no sharpening at all) than the "normal" sharpening setting. Because many reviewers use only the defaut settings, they might have overlooked the incredible genuine quality of the Fuji superCCD image that is considerably damaged by the too strong in-camera sharpening.

Jean-Paul
http://www.pbase.com/sfjp

I think this is simple while people keep comparing the new 4-5 mp
CCD cameras with the FuJI 3.3 mp superCCD cameras: whatever the
necessary improvements these cameras should have to become fully
satsifying, the image quality of the Fuji cameras, in term of
resolution/color/absenc of noise, remains just the best so far of
the less than 4,000$ cameras! And they go to the forums dedicated
to the manufacturers producing 4-5 mp CCD cameras (Minolta, Sony,
Canon, Nikon, etc), they see all these posts of disappointed
owners, then they go to our beloved Fuji forum, and they see there
all these posts of delighted owners that like so much taking and
showing pictures from their Fuji superCCD cameras (mainly owners of
4900/6900) that they all have public galleries now and that even
some are becoming professional while they were not before.... Do
you agree with my answers?

Jean-Paul
http://www.pbase.com/sfjp

The funniest thing I have noticed with all these people looking to
buy new cameras 'they have plenty of choice in the 4 to 5 megapixel
range ......so why do they keep comparing these cameras to the Fuji
?

 

[jacques]

Hi JP,

I expect you do have a Fuji 6900, am I right ?

Well enjoy it.

But don't write things you seem not to have any real knowledge of.

People who are buying the myth of SuperCCD, most probable will believe in the real existance of the land of Oz too.

In the Sony forum I just asked people a very simple straight forward question. Which digital cameras did you own before and how does the F707 compare to those ?

And the results of this poll, well go to the Sony forum and read.

And well the F707 isn't perfect, as well nothing is perfect, I'm not perfect, a D1X isn't perfect, perfectness we have lossed by eating an frbidden apple as a story wants us to believe.

Predictions by Fuji ! What is the value of a prediction by Fuji ?

They might have the knowledge they cann't produce a CCD with truth 5 Mpixel or above, but predictions without prove, evidents by physical laws, well are like the predictions of a gipsy reading your hand.

I don't have any objections against any of the Fuji products, well even don't have the feeling we do need more and more Mpixels, as for a fine photograph the eye behind the camera makes the difference.

And instead of writing a post as you did, I just would have been happy with the tool I owned, a tool I'm familiar with, a tool I can use with 'my eyes closed', well for the technical aspects, so I could open my eye for what does realy counts, what and how I would capture an image.

And sory, but don't believe people, who tell you they can make 1000 out of using only 500.

jacques

 

[Leon Wittwer]

I don't think that Oly did wrong. Olympus never said that the E20 would be a quantum jump over the E10 and I did not expect that. If you look at the Oly SLR forum, there was a big build up of expectations to, IMHO, unrealistic levels. I wish that Oly had speeded up the processing or had increased the buffer size but this was not remotely a deal breaker to me. The E20 does provide better images and some functions like replay are much faster. Actually, read, replay, and get the histogram is faster on the E20 since the increase in processing/write times are more than offset with the decrease in replay time. The post view functions are better particularly the magnify. Focus has been no problem. Probably some of the disappointment revolves around the fact that the E cameras are still a bit slow and are certainly not professional in that respect. I do mostly landscape panoramas (see http://pws.prserv.net/lees_pics/landscapes.htm ) and the speed issue is not very important. Regarding noise reduction, I have developed some actions in Photoshop that do a good job in most cases. In some situations, I do the job manually. The steps I typically use are:

1)upsample image about 33% in pixels per inch (makes median less agressive, varying the upsample amount controls the median function)
2)layer, duplicate layer
3)image, adjust, desaturate layer
4)filter, stylize, find edges

5)curves tool, set upper end of line to (191,255) and do another point at (127,63)
6)layer, duplicate layer
7)filter, noise, median
8)select, color range, highlights
9)select, inverse
10)store selection in channel palette
11)delete last layer in layer palette
12)select, color range, highlights
13)select, inverse
14)add selection in channel palette to current selection
15)delete layer in layer palette
16)select, inverse
17)filter, noise, median
18)remove current selection
19)noise, blur, gaussian blur(radius=0.3)
20)resize image back to original size
21)delete selection stored in channel palette
At this point, you should have a slightly soft denoised image. I usually apply

unsharp mask(47,1,0) and then more unsharp if necessary. At this point, the highpass filtering sharpening method also works well. Steps 6 through 15 select the image regions with gradients that I don't want to apply the median. If there are additional regions that I don't want to affect much, I add them to the selection after step 14. This could be fabric textures, etc. Hope this is useful. Finally, and I am not trying to start a battle, but I would say that my 2560x1920 image does capture more data to work with than the superCCD. I do signal processing for a living and you must have the info to get the result. I agree that Fuji does a lot with the pixels with incamera processing but even Fuji can't do magic. Leon

for the web. Noise can an issue when a shot is significantly
underexposed or sometimes when I am blowing up an image. At ISO 80
and 160, it is pretty easy to suppress the noise while maintaining
good image quality. Over time I have come to prefer to do my own
noise processing. Agressive incamera processing, while looking
nice straight out of the camera, also suppresses fine scale low
contrast textures such as seen in fabrics, brickwork, etc. Be
doing my own processing, I can selectively reduce the noise where I
want to and to the level I want and preserve detail that is lost
with most incamera processing. I realize my approach is not
suitable for those who want the best final image right out of the
camera or don't have the time to process images. In summary, some
noise can be advantageous to the final photograph.

I think this is simple while people keep comparing the new 4-5 mp
CCD cameras with the FuJI 3.3 mp superCCD cameras: whatever the
necessary improvements these cameras should have to become fully
satsifying, the image quality of the Fuji cameras, in term of
resolution/color/absenc of noise, remains just the best so far of
the less than 4,000$ cameras! And they go to the forums dedicated
to the manufacturers producing 4-5 mp CCD cameras (Minolta, Sony,
Canon, Nikon, etc), they see all these posts of disappointed
owners, then they go to our beloved Fuji forum, and they see there
all these posts of delighted owners that like so much taking and
showing pictures from their Fuji superCCD cameras (mainly owners of
4900/6900) that they all have public galleries now and that even
some are becoming professional while they were not before.... Do
you agree with my answers?

Jean-Paul
http://www.pbase.com/sfjp

The funniest thing I have noticed with all these people looking to
buy new cameras 'they have plenty of choice in the 4 to 5 megapixel
range ......so why do they keep comparing these cameras to the Fuji
?

 

[SFJP]

Well, what to tell you? I am not a priori a believer of anything, but my experience with Fuji 6900 showed me there is an amazing potential in the superCCD to get reasonably priced cameras able to produce "professional" quality pictures, potential that most people who did not seriously experiment with it overlook. It does not make 1000 out of 500 but, say, 800-900 out of 500 and without noise or artifacts, which is its main interest! And there is some simple technical explanation of why it is possible (look at Fuji sites).

Jacques, my intention was not to bash Sony, Minolta, Olympus, Canon or Nikon, but simply to state in a bit provocative way something that seems important to most of the photographers who accept presently to be the guinea pigs of the nascent digicam industry: what's our expectations in our quest to get better pictures out of digicams. Camera body and lenses ergonomy and reliability is an important aspect, but, otherwise, the nature of the sensorr and its related processing is certainly also very important: after all the very truth of digicams is just to try to replace conventional film by an electronic senso, so sensor technology should somewhat matter! No?

We see today a race toward packing more and more pixels in a small surface, which may provide better resolution but also a loss of image purity due to noise. After all. many people are willing to spend more than 4,000$ on a D30 and its addional lenses just to get cleaner images, while D30 resolution, 3.3 mp, is no longer at the head of the mp race. So there is certainly an open question on what should be the next wanted enhancements: to force the way to packing yet another bunch of pixels in the same surface and go to the 10 mp with more and more artificial looking images because of noise removal processing or to think to good solutions to have really clean "professional" images?

In this perspective, I think that there are two possible ways: the one opened by Canon with its large size CMOS, which is to use more surface, which allows very clean noise-free images, and the way opened by Fuji with its superCCD, which is to pack less "hardware" pixels with a particular honeycomb geometry in a small surface allowing to get more "software" pixels without artifacts nor noise after some clever extraction. Mt experience with the Fuji 6900, is that really this last approach is worth more coonsideration that it actually deserves among the experts and reviewers, while to get lasrgte surface CCD is certainly what would be the best if this can be done at a resonable pricing (and why not thinking to a mix of both approaches: more surface + superCCD approach. May be it is the road to have quickly affordable medium format digital equivalent, which would require something around 20 mp to 50 mp).

As I said to Leon, I am happy if you are happy with your 707 or your E10, etc. and I would myself be glad if I had on my Fuji 6900 the much better Sony 707 EVF and autofocusing, but I still think that the Fuji produces in general better quality images than the 707 (which is a small part in what constitute a good picture: the photographer talent is 80 %). My goal is not to have people claiming their camera is worth to trash but to see how the digicam industry can improve in a sensible way and not just by marketing arguments based on mp hastily packed in a way that may damage overall image quality.

Jean-Paul

Hi JP,

I expect you do have a Fuji 6900, am I right ?

Well enjoy it.

But don't write things you seem not to have any real knowledge of.

People who are buying the myth of SuperCCD, most probable will
believe in the real existance of the land of Oz too.

In the Sony forum I just asked people a very simple straight
forward question. Which digital cameras did you own before and how
does the F707 compare to those ?

And the results of this poll, well go to the Sony forum and read.

And well the F707 isn't perfect, as well nothing is perfect, I'm
not perfect, a D1X isn't perfect, perfectness we have lossed by
eating an frbidden apple as a story wants us to believe.

Predictions by Fuji ! What is the value of a prediction by Fuji ?

They might have the knowledge they cann't produce a CCD with truth
5 Mpixel or above, but predictions without prove, evidents by
physical laws, well are like the predictions of a gipsy reading
your hand.

I don't have any objections against any of the Fuji products, well
even don't have the feeling we do need more and more Mpixels, as
for a fine photograph the eye behind the camera makes the
difference.

And instead of writing a post as you did, I just would have been
happy with the tool I owned, a tool I'm familiar with, a tool I can
use with 'my eyes closed', well for the technical aspects, so I
could open my eye for what does realy counts, what and how I would
capture an image.

And sory, but don't believe people, who tell you they can make 1000
out of using only 500.

jacques

 

[SFJP]

Thank you for your very detailed recipe of your noise removal action. I will try it soon.

I agree with you that Fuji can't make magic but that it does already a lot fromm the initial 3.3 mps on its superCCD, say it provides a resolution close to a 5 mp in its 6 mp resulting pictures. And it is my experience that at ISO 100 Fuji superCCD pictures at 6mp do not exhibit any noise when taken with a correct exposure. With Neat Image I get from my ISO 200 pictures that have a slight noise very clean images that can compete in cleanness with waht can produce a Canon D30 (but are better in resolution of course as the Canon is a simple 3 mp). I am eagger to try your action to see if it can provide me with similar results but more quickly than Neat Image on my ISO 200 images. Thank you,

Jean-Paul

 

[bhenning]

Hi,

Chalk me up as another satisfied E-10 user... I got mine only two weeks ago, after deciding it was more suitable for my uses than an E-20 (what I really wanted was a Nikon D1x but could not justify the cost). Noise (so far) has not been an issue for me, and I've been VERY happy with the 8x10's I've made so far; and the one 12.5x17" print I've done (upsampled with PS) is also quite good.

I did not think I would gain much with the 33% more pixels than my CP990, but I do appear to notice a bit more fine detail even in the 8x10's.

As far as Fuji goes, I seriously considered buying an S1Pro; but the lack of a RAW mode and "drive" mode in anything but the "sports" mode combined with costing more than twice as much as an E-10 by the time you bought a Nikkor F2.8 28-105mm lens steered me to the E-10.

Due to consulting commitments I have not has as much time to shoot with my E-10 as I would have liked so far, but I have a fashion portfolio shoot tomorrow - and I'll finally see how the E-10 does for model shoots.

Best Regards,

Bill

I am happy (not at all disappointed) to see an owner of the E10 to
be happy with his new E20, I was wondering what Olympus did so
wrong in trying to enhance the E10 with the E20 to raise so much
dissatisfaction in the Oly forum (such as people reselling their
new E20 and keeping their old E10)... I am hoping that you are the
first of a long serie of people that would use this thread to try
to contradict my initial assertions, just to say how much they are
happy with their Oly, Sony, Minolta, etc. My intention is not to
bash other cameras that I do not own, but in the contrary to make
people express their satisfaction when they are satisfyed and also
to attract the attention of all on the incredible quality of the
Fuji superCCD, too much neglected and underestimated by the
reviewers.

 

[Peter Brogden]

It seems to me that the issue that has been overlooked in all of these posts is that of lens quality.

I have a Fuji6800 and was very pleased with it until I did a direct comparison with a Fuji S1 The resolution of the S1 was far superior which I put down to the lens, not the CCD.

In most tests I have read, SLR lenses can resolve around 100-120 lines per millimetre. The large CCD in a S1 requires a lens with a resolving power of around 90 lpm. With most consumer digital cameras having 2/3 inch CCD's, the resolving power of the lens needs to be around 400 lpm to take full adavatage of the CCD. Are lenses on these cameras capable of this type of resolving power? - I don't think so, which is why my 6800 compares so badly with the S1.

I am not especially a technical person when it comes to photography, but to my mind it does not matter whether Olympus put a 4MP or a 5Mp CCD in their E10/E20 cameras, the lens cannot take advantage of the number of pixels anyway. All this argument about noise is a less important issue.

Isn't it?

Peter

Hi JP,

I expect you do have a Fuji 6900, am I right ?

Well enjoy it.

But don't write things you seem not to have any real knowledge of.

People who are buying the myth of SuperCCD, most probable will
believe in the real existance of the land of Oz too.

In the Sony forum I just asked people a very simple straight
forward question. Which digital cameras did you own before and how
does the F707 compare to those ?

And the results of this poll, well go to the Sony forum and read.

And well the F707 isn't perfect, as well nothing is perfect, I'm
not perfect, a D1X isn't perfect, perfectness we have lossed by
eating an frbidden apple as a story wants us to believe.

Predictions by Fuji ! What is the value of a prediction by Fuji ?

They might have the knowledge they cann't produce a CCD with truth
5 Mpixel or above, but predictions without prove, evidents by
physical laws, well are like the predictions of a gipsy reading
your hand.

I don't have any objections against any of the Fuji products, well
even don't have the feeling we do need more and more Mpixels, as
for a fine photograph the eye behind the camera makes the
difference.

And instead of writing a post as you did, I just would have been
happy with the tool I owned, a tool I'm familiar with, a tool I can
use with 'my eyes closed', well for the technical aspects, so I
could open my eye for what does realy counts, what and how I would
capture an image.

And sory, but don't believe people, who tell you they can make 1000
out of using only 500.

jacques

 

[SFJP]

You are right, lens is a very big factor alone. After all, when judging film systems, where there is no difference in sensor or in-camera processing (!), the great debate is about lens quality (Canon versus Nikon versus Leica versus Zeiss, etc), and many comparison tests show that Leitz lenses are those producing visibly the bestsharp images while in lab they show only an average resolution not better than consumer level low price lenses!! However, the choice of film emulsion, the way it is treated in laboratory, is nevertheless of most importance. I remember when I was young and taking pictures with film cameras, the debate was: should I choose Kodachrome or Ektachrome, without forgetting the outsider of this time, Fujichrome... I guess now the debate is about other names of emulsions, and ISO sensitivity. You can use a Leica system, but if you put inside a bad emulsion producing dirty grainy results, you might not be satisfied with the overall quality of the combination camera + film emulsion.

The Fuji 4900, 6900, in a less extent the 6800, the Canon G1, G2, the Sony 85, 707 have all excellent lenses for their range of price. Especially the Fuji 6900 and Sony 707 if you consider the zoom range they provide (we will see what will the price of the new Panasonic with a Leica lens and if this lens is on-par with what could be expected of the Leica name). Now, if you put a 2,000$ top Nikon zoom lens on a 2,500$ Fuji S1 body, you might expect a better image quality than what can produce a 500-1,000$ complete system like a Fuji 6900 or Sony 707, would'nt you? And, as you mentioned, the bigger the surface of the CCD, the less line/mm resolution is needed for a lens, so in any case, professional cameras with bigger surface CCD should win easily even if using lenses that are no better than those offered in the consumer/prosumer range of cameras.

This is really the central point of the debate I want to see develop about the future of prosumer digicams. What is the best way to get better, i.e. sharper and cleaner images at an affordable price. Packing each year more and more pixels in the same small surfaces of CCD might not go very far beyond what we have, and may be all the disatisfaction about the 5 mp cameras, disatsifaction I think related to noise problem, show that going further than 3-4 mp is not indicated with small surface consumer CCD. The best technical solution is of course to seek for the surface of CCD becoming larger, but there is the question of the price. So come the superCCD solution, that is a compromise to get more pixels with an acceptable noise on small surface sensors. A mix of both approaches, bigger CCD with superCCD geometry might provide the best compromise to get quickier better resolution with sharp and clean images at an affordable price. But, to get the top notch lenses, you'll have to pay a very high price anyway.

Jean-Paul

It seems to me that the issue that has been overlooked in all of
these posts is that of lens quality.

I have a Fuji6800 and was very pleased with it until I did a direct
comparison with a Fuji S1 The resolution of the S1 was far
superior which I put down to the lens, not the CCD.

In most tests I have read, SLR lenses can resolve around 100-120
lines per millimetre. The large CCD in a S1 requires a lens with a
resolving power of around 90 lpm. With most consumer digital
cameras having 2/3 inch CCD's, the resolving power of the lens
needs to be around 400 lpm to take full adavatage of the CCD. Are
lenses on these cameras capable of this type of resolving power? -
I don't think so, which is why my 6800 compares so badly with the
S1.

I am not especially a technical person when it comes to
photography, but to my mind it does not matter whether Olympus put
a 4MP or a 5Mp CCD in their E10/E20 cameras, the lens cannot take
advantage of the number of pixels anyway. All this argument about
noise is a less important issue.

Isn't it?

Peter

 

[Leon Wittwer]

Regarding where things may be going, the best solution would be to have more bits of info (dynamic range) per pixel out of the ccd, faster processing, and other such improvements. Unfortunately, it may not be all that easy. The next level cameras have physically bigger pixels to get more dynamic range. Perhaps this will filter down to less expensive cameras. Kodak makes ccd that have very high performancs but they cost a great deal and it is questionably whether we will see that technology in one or two thousand dollar cameras for some time (if ever). At one time I would have said that mass markets, mass production would have brought the higher technologies to more folks but now I am not so sure. It may be that the mass market will be happy with 2-4 megapixel cameras which will match or be slightly better than they are now used to in their consumer film cameras. The current ccd technology seems to support these size cameras pretty well. Thus there may not be enough demand from us folks midway between consumer and pro to pull the more expensive technologies into true mass production. Well, time will tell. Leon

The Fuji 4900, 6900, in a less extent the 6800, the Canon G1, G2,
the Sony 85, 707 have all excellent lenses for their range of
price. Especially the Fuji 6900 and Sony 707 if you consider the
zoom range they provide (we will see what will the price of the new
Panasonic with a Leica lens and if this lens is on-par with what
could be expected of the Leica name). Now, if you put a 2,000$ top
Nikon zoom lens on a 2,500$ Fuji S1 body, you might expect a better
image quality than what can produce a 500-1,000$ complete system
like a Fuji 6900 or Sony 707, would'nt you? And, as you mentioned,
the bigger the surface of the CCD, the less line/mm resolution is
needed for a lens, so in any case, professional cameras with bigger
surface CCD should win easily even if using lenses that are no
better than those offered in the consumer/prosumer range of cameras.

This is really the central point of the debate I want to see
develop about the future of prosumer digicams. What is the best way
to get better, i.e. sharper and cleaner images at an affordable
price. Packing each year more and more pixels in the same small
surfaces of CCD might not go very far beyond what we have, and may
be all the disatisfaction about the 5 mp cameras, disatsifaction I
think related to noise problem, show that going further than 3-4 mp
is not indicated with small surface consumer CCD. The best
technical solution is of course to seek for the surface of CCD
becoming larger, but there is the question of the price. So come
the superCCD solution, that is a compromise to get more pixels with
an acceptable noise on small surface sensors. A mix of both
approaches, bigger CCD with superCCD geometry might provide the
best compromise to get quickier better resolution with sharp and
clean images at an affordable price. But, to get the top notch
lenses, you'll have to pay a very high price anyway.

Jean-Paul

It seems to me that the issue that has been overlooked in all of
these posts is that of lens quality.

I have a Fuji6800 and was very pleased with it until I did a direct
comparison with a Fuji S1 The resolution of the S1 was far
superior which I put down to the lens, not the CCD.

In most tests I have read, SLR lenses can resolve around 100-120
lines per millimetre. The large CCD in a S1 requires a lens with a
resolving power of around 90 lpm. With most consumer digital
cameras having 2/3 inch CCD's, the resolving power of the lens
needs to be around 400 lpm to take full adavatage of the CCD. Are
lenses on these cameras capable of this type of resolving power? -
I don't think so, which is why my 6800 compares so badly with the
S1.

I am not especially a technical person when it comes to
photography, but to my mind it does not matter whether Olympus put
a 4MP or a 5Mp CCD in their E10/E20 cameras, the lens cannot take
advantage of the number of pixels anyway. All this argument about
noise is a less important issue.

Isn't it?

Peter

 

[Peter Brogden]

I agree

Most digital cameras are glorified compacts IMHO. My 6800 is a great snapshot camera, but is never going to replace my Canon EOS. I guess I will be buying a D30 or a S1 in order to enjoy the creativity my EOS provides. I have looked at the E10/E20 and Dimage 7, but non interchangable lenses and awful viewfinders won't persuade me that these are the 'prosumer' cameras for me.

Pro Digital SLR's are all film cameras that have been tinkered with.
They are inevitably a compromise.

I think at some point, the manufacturers will have to bite the bullet and introduce a new format - smaller than 35mm to suit the APS-sized CCD's used now. But this means a new lens mount, new lenses and much expense. I wonder who will be the first?

.

 

[David SL]

Jean Paul,

Thanks for a well thought out analysis, here are a few observations:

It's true that the noise present on 5mp ccd's at the 2/3" dimension is theoretically harder to remove than in the lower resolution cameras. However, two factors serve to mitigate the appearance of this noise in print. 1) Increased pixel count means smaller photosites, which means any noise that does exist even if more numerous is visibly not as apparent at standard print sizes ( 4 x 6, 5 x 7 , 8 x 10) only when you zoom in to greater than 100% view does the increased noise become apparent. 2) Noise reduction features (previously uncommon to prosumer cameras) reduces further any existing noise, depending on the treatement by the particular manufacturer this can result in an improvement over current 3 and 4 mp camera's noise performance.

I think that using the MD7 , Sony 707 and the Nikon 5000 as examples highlights how different the experience of the manufacturer and their treatment of noise issues can go into circumventing the limitations of the common sony ccd. Minolta's first attempt was good, but their lack of experience in the DC market contributed to the results that we've seen from its implementation. Sony being the manufacturer was probably able to take advantage of design knowledge that Minolta wasn't privy too, and also had time to tweak away the issues that cropped up from Minolta's experience with the MD7. They were better able to accentuate the ccd's positives (high sensitivity) and detract from its weaknessesses (NR features to reduce noise). Nikon brings to the table even more experience tweaking the last drop of performance out of ccd's in their dc's, as well as high end noise reduction and image processing algorithms, from what I've seen from both the Sony 707 and the pre prod. shots of the 5000, I would say both have succeeded in yielding a visible improvement in noise performance over lower mp cameras using the 5mp consumer ccd. As time goes by and technology improves the prosumer landscape will be changed as larger ccd dimensions fall within the price range of prosumer cameras. Fuji's bet was for the most part accurate, but weather their window of opportunity for proving the superiority of their SuperCCD technology is still unknown. I think that this time next year consumer level ccd's will be a bit larger and cheaper than Fuji might have anticipated thanks to the advancement of technology.

We shall see how things proceed in time to come, until then enjoy your cameras and learn to get the best out of the solution provided by your cameras manufacturer.

Dsl

SFJP wrote:

 

[SFJP]

David,

I mostly agree on your remarks except on one point related to Sony 707. I was expecting that Sony being the manufacturer of the CCD would get better results from it than others. Based on the samples I have downloaded from various sites, I'd say that in my view Sony 707 did succeed in removing noise at the expense of the overall photographic quality of the pictures and that eventually the 707 images are the worst of the breed. In all the full size 1:1 samples I have seen so far the surface textures, the overemphasized surface boundaries and the colors are terrible. And there I am not biased by any support of Fuji because the superior quality of the EVF provided by the 707 had almost convinced me to buy it.

For the Nikon 5000, I cannot say anything, each time I have seen a link to samples these samples had been removed when I tried the link, which I find a strange strategy from Nikon for a camera that is arriving on the shelves...

Now, for what is needed for average amateur prints with average amateur printers at 8x10 or less, all this discussion is out of scope because a 2.5 mp is quite enough to cover the quality needs, 4 or 5 mp is quite above what is needed and noise is rather not visible. This is not true if you are using a high quality printer, which reveal all defects of the original picture. So, liiking at 4-5 mp cameras and above for the future, is not a concern of average amateur but of consumer that wants professional quality, and then noise level matters.

Jean-Paul
http://www.pbase.com/sfjp

Jean Paul,

Thanks for a well thought out analysis, here are a few observations:

It's true that the noise present on 5mp ccd's at the 2/3" dimension
is theoretically harder to remove than in the lower resolution
cameras. However, two factors serve to mitigate the appearance of
this noise in print. 1) Increased pixel count means smaller
photosites, which means any noise that does exist even if more
numerous is visibly not as apparent at standard print sizes ( 4 x
6, 5 x 7 , 8 x 10) only when you zoom in to greater than 100% view
does the increased noise become apparent. 2) Noise reduction
features (previously uncommon to prosumer cameras) reduces further
any existing noise, depending on the treatement by the particular
manufacturer this can result in an improvement over current 3 and 4
mp camera's noise performance.

I think that using the MD7 , Sony 707 and the Nikon 5000 as
examples highlights how different the experience of the
manufacturer and their treatment of noise issues can go into
circumventing the limitations of the common sony ccd. Minolta's
first attempt was good, but their lack of experience in the DC
market contributed to the results that we've seen from its
implementation. Sony being the manufacturer was probably able to
take advantage of design knowledge that Minolta wasn't privy too,
and also had time to tweak away the issues that cropped up from
Minolta's experience with the MD7. They were better able to
accentuate the ccd's positives (high sensitivity) and detract from
its weaknessesses (NR features to reduce noise). Nikon brings to
the table even more experience tweaking the last drop of
performance out of ccd's in their dc's, as well as high end noise
reduction and image processing algorithms, from what I've seen from
both the Sony 707 and the pre prod. shots of the 5000, I would say
both have succeeded in yielding a visible improvement in noise
performance over lower mp cameras using the 5mp consumer ccd. As
time goes by and technology improves the prosumer landscape will be
changed as larger ccd dimensions fall within the price range of
prosumer cameras. Fuji's bet was for the most part accurate, but
weather their window of opportunity for proving the superiority of
their SuperCCD technology is still unknown. I think that this time
next year consumer level ccd's will be a bit larger and cheaper
than Fuji might have anticipated thanks to the advancement of
technology.

We shall see how things proceed in time to come, until then enjoy
your cameras and learn to get the best out of the solution provided
by your cameras manufacturer.

Dsl

SFJP wrote:

 

[Anand67599]

some facts:
(mostly from various news source on dpreview)

The dynamic range of s1 is about twice that of D1 and thrice of D30 at iso800. Normally it increases in same proportion with sensor size, but superccd seems to defies it totally. Given same number of pixels(information), superccd position them more strategically to win interpolation battle also, so much so, it seems to multiply the resolution to double, (however, by noaccident its only in 'soft' mode). 'moire' and other artifact are bound to exhibit as about 1/3 of the image is just a guess work (for any ccd sensor, tho guess' intelligence differs). Not only this, 4 ccds together produces 1 pixel equivalent, implies about 1/2-4th truthfulness about postion of the color. There is still a long way to go in comparison to conventional film. To put a better perspective, best 6mp phillips ccd sensor's pixel size is also thrice as big as 35mm fine grain size. Which is already said to crammed ccd maximamly. (I am curious if 'lines' in recent > 4mp cameras' results are due to masking more rows etc.(in square grid, info of entire row completely lost then) to find 'black level' in relatively noisy sensor.)

Kodak has developed 16mp ccdsensor, has about twice the resolution of conventional film. Thats amazing. breaking fuji and other's claim by far. However, costwise rest assured, it can't be affordable in forseable future. But again forseeing 'future' against man and technology is dumb. Foveon has iched 70million !! cmos transistors(thanks to exploding and hence costeffective semiconductor technology, it would cost just several dollars). It would produce 16.8mp. We may expect this number not to grow because of finite wavelength of visible light. 4-5 mp is about right(3 is lil less) informationwise to produce a unobjectionable image to our eyes.

Interesting! This CMOS may be little short in image resolution from ccds but energy efficient(big concern) and can have gamma curve like film as oppose in ccd which results in digital artifact looking ugly. Being simpler and intelligent(each foveon's pixel is madeup of 7! transisters, which allows onboard logic). It leaves enormous possibily for creative technocrats to develop image-inhancement tool(such as digitally programmed shutter, selective area loupe).

My opinion; digcam still has lots to go in area ,Before, the damn photon counting sensors. Like taking care of dead area in sensor, optical zoom range( Food for imagination and thought: if you compensate for lens distortion by shaping the sensor in appropriate curve or developing appropriate algorithm (as a function of zoom maybe), one can get rid of many correction lens decreasing lens weight and increase zoom reach considerably.

 

[Homer40784]

David,

I mostly agree on your remarks except on one point related to Sony
707. I was expecting that Sony being the manufacturer of the CCD
would get better results from it than others. Based on the samples
I have downloaded from various sites, I'd say that in my view Sony
707 did succeed in removing noise at the expense of the overall
photographic quality of the pictures and that eventually the 707
images are the worst of the breed. In all the full size 1:1 samples
I have seen so far the surface textures, the overemphasized surface
boundaries and the colors are terrible. And there I am not biased
by any support of Fuji because the superior quality of the EVF
provided by the 707 had almost convinced me to buy it.

Well you are diametrically opposed to every reviewer that I have read on this issue. (And my own opinion.....but I give that smaller weight since I don't make my living evaluating Digicams)

Of course, these ("Photographic" quality) are subjective opinions and you are allowed yours. I just find it odd that not a single reviewer that I have read agrees with you.

That of course does not make you wrong....but definitly different than these pros.

Homer

 

[Homer40784]

Well, not sure if I agree with everything but for the most part I agree.
And I found particularly interesting........

( Food for imagination and thought: if you

compensate for lens distortion by shaping the sensor in appropriate
curve or developing appropriate algorithm (as a function of zoom
maybe), one can get rid of many correction lens decreasing lens
weight and increase zoom reach considerably.

The concept is not new.

I remember a similiar (tho not as complex) situation with a Sony Stereo some years ago.

It was basicallya reciever with a Turntable built into the top.......I;m going to challenge my memory and say it was a Model HP500.......
With two bookshelf speakers.
It had incredible sound for something like $300USD in 1974 or so.

What they did was build a response curve of the speakers and then built a "mirror" amplifier.

Where the speakers rolled off, the amp peaked. Where the speakers had peaks, the amp had valleys.

the result was a nearly flat frequency response sonically...... but not electronically.

Of course it was a "system". Put "good" speakers on this little system and it sounded terrible ;> )

Would it work with CCDs and lens? well the concept works.......

Homer

 

[SFJP]

Homer,

You are right, I seem very isolated in this appreciation of the artificial look of Sony 707 images.Users of Sony 707 are mainly complaining about colors that many find unacceptable, but I have not seen any message about the surface texture and surface boundaries alteration by artifacts due to in-camera noise reduction and sharpening (except Phil in its review who noticed the overemphasis of the boudaries by added lines due to in-camera processing).

Well, in the Fuji camp I see the same kind of situation: I seem rather alone to claim that the in-camera sharpening of the 6900 alters significantly the overall quality of the image in adding artifiacts that are very very visible for me. But the truth is that I take 6 mp images with my Fuji while most people use the 3 mp reduction following the reviewers that say it is only a 3.3 mp camera. Well, in my experience 6 mp pictures coming from the Fuji 6900 reveal much more the defects of your images because they provide much more details than a simple interpolarlation from 3 mp would, and artifacts due to sharpening are quite visible to me. I guess this is the same for the Sony 707, people are not looking carefully to full size 1:1 5 mp pictures on a screen, they just look at prints in standard size with standard consumer printers.

So may be I am just obsessed by details that noone see and should go to consult a psy or maybe the thruth is that most people don't care to such details because they do not look carefully at full size 1:1 full resolution their images, and in any case are happy just if their 5 X 7 or 8 x 10 prints prinyted with average consumer printers are OK. Well why not? The goal is that people can be happy. But in this case, 5 or 6 mp cameras are not needed, 2.5-3.3 mp cameras would be far enough in most cases.

Jean-Paul

David,

I mostly agree on your remarks except on one point related to Sony
707. I was expecting that Sony being the manufacturer of the CCD
would get better results from it than others. Based on the samples
I have downloaded from various sites, I'd say that in my view Sony
707 did succeed in removing noise at the expense of the overall
photographic quality of the pictures and that eventually the 707
images are the worst of the breed. In all the full size 1:1 samples
I have seen so far the surface textures, the overemphasized surface
boundaries and the colors are terrible. And there I am not biased
by any support of Fuji because the superior quality of the EVF
provided by the 707 had almost convinced me to buy it.

Well you are diametrically opposed to every reviewer that I have
read on this issue. (And my own opinion.....but I give that smaller
weight since I don't make my living evaluating Digicams)
Of course, these ("Photographic" quality) are subjective opinions
and you are allowed yours. I just find it odd that not a single
reviewer that I have read agrees with you.
That of course does not make you wrong....but definitly different
than these pros.

Homer