Re: CCD vs CMOS: A survey

After reading the Kodak's article again, I searched on DPReview to a possible advantage of CCD over CMOS to justify Kodak's position. The most obvious advantage is the dynamic range of CCD. Looking at dynamic range (RAW or JPEG) for a given ISO CCD in D1 and S1 has a higher dynamic range then CMOS.

Do Kodak's experts believe that most photographers will prefer higher dynamic range to lower noise?? What is your opinion? Would you trade off higher dynamic range for noise levels?

I personally prefer low noise images to higher dynamic range because being and amateur, I have not run into many situations that require increased dynamic range will benefit me. I would love to hear your thoughts on this subject.

-Jay

One reason Kodak and everyone else has been preferring CCD's might be that it's very possible and practical to reduce noise by using anti-noise filters. I think another thing supporting their point of view is the sharpness of the new Canon CCD cameras (G1, A10, A20, etc.) as compared to the D30. And as much as I hate to say it - since I love D30's color rendition- it seems CCD's are more accurate in color as well.
You can view Canon's own sample images on this page:
http://www.canon.co.jp/Imaging/BeBit-e.html

Jay said:

After reading the Kodak's article again, I searched on DPReview to
a possible advantage of CCD over CMOS to justify Kodak's position.
The most obvious advantage is the dynamic range of CCD. Looking at
dynamic range (RAW or JPEG) for a given ISO CCD in D1 and S1 has a
higher dynamic range then CMOS.

Do Kodak's experts believe that most photographers will prefer
higher dynamic range to lower noise?? What is your opinion? Would
you trade off higher dynamic range for noise levels?

I personally prefer low noise images to higher dynamic range
because being and amateur, I have not run into many situations that
require increased dynamic range will benefit me. I would love to
hear your thoughts on this subject.

-Jay

Click to expand...

Just how many more times does it have to be pointed out that the D30 doesn't do much in-camera sharpening as a deliberate (and IMO correct) design decision by Canon rather than anything to do with the technology?

The best use of sharpening is after all other corrections have been applied and when the image has been set to the final size for output, not while it's still in the camera.

Tony said:

Jay said:

After reading the Kodak's article again, I searched on DPReview to
a possible advantage of CCD over CMOS to justify Kodak's position.
The most obvious advantage is the dynamic range of CCD. Looking at
dynamic range (RAW or JPEG) for a given ISO CCD in D1 and S1 has a
higher dynamic range then CMOS.

Do Kodak's experts believe that most photographers will prefer
higher dynamic range to lower noise?? What is your opinion? Would
you trade off higher dynamic range for noise levels?

I personally prefer low noise images to higher dynamic range
because being and amateur, I have not run into many situations that
require increased dynamic range will benefit me. I would love to
hear your thoughts on this subject.

-Jay

Click to expand...

Click to expand...

Jay said:

After reading the Kodak's article again, I searched on DPReview to
a possible advantage of CCD over CMOS to justify Kodak's position.
The most obvious advantage is the dynamic range of CCD. Looking at
dynamic range (RAW or JPEG) for a given ISO CCD in D1 and S1 has a
higher dynamic range then CMOS.

Do Kodak's experts believe that most photographers will prefer
higher dynamic range to lower noise?? What is your opinion? Would
you trade off higher dynamic range for noise levels?

I personally prefer low noise images to higher dynamic range
because being and amateur, I have not run into many situations that
require increased dynamic range will benefit me. I would love to
hear your thoughts on this subject.

-Jay

Click to expand...

CMOS does not have an advantage when it comes to noise, either. In fact, CMOS is worse in the noise department. One thing the engineers at Canon had to do was work on the noise levels (with on chip circuitry) to make noise levels acceptable. I remember reading an article when the D30 was first released (on the net somewhere) which talked quite a bit about that problem. I dont remember where, now. Also, output level of each pixel on CMOS varies quite a bit, and on chip circuitry (a FET amplifier for each pixel) was required to equalize the output level. Because of these problems, CMOS was always at a disadvantage compared to CCD's, and the savior for CMOS is the ability to include on chip circuitry to address these inherant problems. Not that this has much interrest to the end user.

Derek -

No one said anything at all about sharpening. The question has to do with dynamic range versus noise, and where CCD and CMOS fall in this respect.

Derek Clarke said:

Just how many more times does it have to be pointed out that the
D30 doesn't do much in-camera sharpening as a deliberate (and IMO
correct) design decision by Canon rather than anything to do with
the technology?

The best use of sharpening is after all other corrections have been
applied and when the image has been set to the final size for
output, not while it's still in the camera.

Click to expand...

I just want to know, do the PRO's encounter situations where they would be willing to trade noise for dynamic range?

-Jay

Ulysses said:

No one said anything at all about sharpening. The question has to
do with dynamic range versus noise, and where CCD and CMOS fall in
this respect.

Derek Clarke said:

Just how many more times does it have to be pointed out that the
D30 doesn't do much in-camera sharpening as a deliberate (and IMO
correct) design decision by Canon rather than anything to do with
the technology?

The best use of sharpening is after all other corrections have been
applied and when the image has been set to the final size for
output, not while it's still in the camera.

Click to expand...

Click to expand...

CMOS also gives higher burnt out highlights.

Jay said:

After reading the Kodak's article again, I searched on DPReview to
a possible advantage of CCD over CMOS to justify Kodak's position.
The most obvious advantage is the dynamic range of CCD. Looking at
dynamic range (RAW or JPEG) for a given ISO CCD in D1 and S1 has a
higher dynamic range then CMOS.

Do Kodak's experts believe that most photographers will prefer
higher dynamic range to lower noise?? What is your opinion? Would
you trade off higher dynamic range for noise levels?

I personally prefer low noise images to higher dynamic range
because being and amateur, I have not run into many situations that
require increased dynamic range will benefit me. I would love to
hear your thoughts on this subject.

-Jay

Click to expand...

pixelsworld said:

CMOS also gives higher burnt out highlights.

Click to expand...

CCD and CMOS does not have this or that property. Different
CCD chips have different properties and different CMOS chips
have different properties. As a matter of fact, the sensor part
is very similar. All this talk abot CMOS being better/worse than
CCD is based upon today and older implementatons.

I don't see how this could possibly be an either/or choice.

The reason you have a higher dynamic range is exactly that the device has less noise.

Now for the confusing bit...

Now should you try to take advantage of the higher dynamic range by boosting the contrast in the shadows (or any other area of a picture with an equally limited brightness range) you will also boost the noise in that area. That is like having a really quiet sound recording and then turning up the volume. The original recording (the imager) has less noise but the amplification boosts the low noise level so that it seems as noisy as a normal recording - until a crescendo (in my analogy, a normally lit area) comes along.

I regularly do this with my images scanned from film. The scanner (a Polaroid Sprintscan 35+) provides 36 bit colour. It's greate dynamic range allows me to boost the shadow areas to give a more complete looking end result - but if you know what you're looking for, you can see an increase in graininess in the shadows. Fortunately film graininess is more apparent in light areas than shadows so the increase is not very noticable.

Now as for CCDs vs CMOS, it seems to me that the difference is the amount of surface area in the imager actually used to "gather" light (as opposed to "processing" the result). There's more of it in the CCD and less in the CMOS. If everything else is equal, I'd rather have more light gathering surface so I guess I prefer a CCD. Whether it makes a real difference is something I don't know. I'll let the images talk for themselves.

Gerald

( http://homepage.mac.com/gcarter )

Jay said:

....

Do Kodak's experts believe that most photographers will prefer
higher dynamic range to lower noise?? What is your opinion? Would
you trade off higher dynamic range for noise levels?

...

-Jay

Click to expand...

I can't add to the comparison between CMOS and CCD, only that in my experience with my Canon D30 it tends to white-out very easily , and as such it is necessary to underexpose in many cases to avoid whiting out the bright areas. Levels can then be brought back in Photoshop, but stretching a histogram which is mostly on the bottom half is costly in terms of dynamic range, and as such I could really use some more dynamic range in the camera. (or a way to avoid the white-outs)

See my response below:

Gerald Carter said:

I don't see how this could possibly be an either/or choice.

The reason you have a higher dynamic range is exactly that the
device has less noise.

Click to expand...

I disagree..with your above statement. What your statement implies that any new camera that comes out in future with lower noise then D30 (for example) will have lower dynamic range then D30. I don't believe that it is true because dynamic range depends upon lot of factors besides just the ability to digitize the analog signal (sort of volume control). Size of pixel, electrical characteristic of the imaging circuit, temprature, power variation all play a role in noise characteristic of the sensor. Example, compare Fuji S1 Pro with Nikon D1 at 800 and 1600 ISOs on the link below:

http://www.dpreview.com/news/0011/00111608dynamicrange.asp

Fuji's S1 pro has higher dynamic range at 800 and 1600 and lower noise then D1.

Gerald Carter said:

Now as for CCDs vs CMOS, it seems to me that the difference is the
amount of surface area in the imager actually used to "gather"
light (as opposed to "processing" the result). There's more of it
in the CCD and less in the CMOS. If everything else is equal, I'd
rather have more light gathering surface so I guess I prefer a CCD.
Whether it makes a real difference is something I don't know. I'll
let the images talk for themselves.

Click to expand...

Again I tend to disagree. The "light gathering area" or the pixel size on Nikon CCD and D30's CMOS sesor is almost the same (See http://www.dpreview.com/reviews/canond30/ ). The difference between the CMOS and CCD is their mechanism of detecting light. CCD converts light into (# of electrons) charge and CMOS converts light into voltage or current (depends on the circuit, See http://www.kodak.com/US/en/corp/researchDevelopment/technologyFeatures/cmos.shtml ).

I think that the trend of upcoming camera's (SLR or consumer) will be to reduce niose AND increase dynamic range.

-Jay

Sounds to me more like an exposure issue rather than one of dynamic range. If you try and get an exposure very close to full range, you will inevitably get some white out. Exposure is generally set at 18% or so of the area measure (unless you do it manually). This does not work for some scenes.

To get something that looks nice out of the camera, there is the tendency to overexpose. With digital, people go directly to looking at pictures on the computer. With Film, the film print process usually compensates for exposure.

Actually I find a real problem to be in Photoshop (btw their "auto compensation" is TERRIBLE and clamps out light areas). One really needs headroom "above white" when working on pictures so that you can go past the white limit without loosing the data, it would just display as white (keeping the extra bits in case future steps bring it back down). I would want 16-bits per color in photoshop with 4-bits above and 4 bits below 8-bits kept.

Karl

Peter Lindsey said:

I can't add to the comparison between CMOS and CCD, only that in my
experience with my Canon D30 it tends to white-out very easily ,
and as such it is necessary to underexpose in many cases to avoid
whiting out the bright areas. Levels can then be brought back in
Photoshop, but stretching a histogram which is mostly on the bottom
half is costly in terms of dynamic range, and as such I could
really use some more dynamic range in the camera. (or a way to
avoid the white-outs)

Click to expand...

My responses are below

Jay said:

Gerald Carter said:

I don't see how this could possibly be an either/or choice.

The reason you have a higher dynamic range is exactly that the
device has less noise.

Click to expand...

I disagree..with your above statement. What your statement implies
that any new camera that comes out in future with lower noise then
D30 (for example) will have lower dynamic range then D30. I don't

Click to expand...

We actually agree! Maybe I didn't make it clear: lower noise is a necessary condition for higher dynamic range.

Member said:

believe that it is true because dynamic range depends upon lot of
factors besides just the ability to digitize the analog signal
(sort of volume control). Size of pixel, electrical characteristic
of the imaging circuit, temprature, power variation all play a role
in noise characteristic of the sensor. Example, compare Fuji S1 Pro
with Nikon D1 at 800 and 1600 ISOs on the link below:

http://www.dpreview.com/news/0011/00111608dynamicrange.asp

Fuji's S1 pro has higher dynamic range at 800 and 1600 and lower
noise then D1.

Member said:

Now as for CCDs vs CMOS, it seems to me that the difference is the
amount of surface area in the imager actually used to "gather"
light (as opposed to "processing" the result). There's more of it
in the CCD and less in the CMOS. If everything else is equal, I'd
rather have more light gathering surface so I guess I prefer a CCD.
Whether it makes a real difference is something I don't know. I'll
let the images talk for themselves.

Click to expand...

Again I tend to disagree. The "light gathering area" or the pixel
size on Nikon CCD and D30's CMOS sesor is almost the same (See
http://www.dpreview.com/reviews/canond30/ ). The difference between
the CMOS and CCD is their mechanism of detecting light. CCD
converts light into (# of electrons) charge and CMOS converts light
into voltage or current (depends on the circuit, See

http://www.kodak.com/US/en/corp/researchDevelopment/technologyFeatures/cmos.shtml ).

Click to expand...

Both the CCD and the CMOS ultimately use similar photodiodes to "convert" light to charge. The CMOS has the charge reading electronics adjacent to the photodiodes whereas the CCD shifts the charge to reading electronics on the periphery of the chip (it can even be off the chip). Comparitively the photodiodes on the CMOS are smaller than that of the CCD simply because the CMOS has more surface area dedicated to processing electronics. Look at the diagrams in the Kodak article.

Phil's calculations did not give the area of the photodiode. He merely divided the area of the chip by the resolution. Since the chips are roughly the same size but the Canon has a slightly higher pixel count, Phil calculated a slightly lower area per pixel. If you make him a donation of a scanning electron microscope (I can't afford to do this myself!), I'm sure you can convince him to measure the area of the photodiodes.

Member said:

I think that the trend of upcoming camera's (SLR or consumer) will
be to reduce niose AND increase dynamic range.

-Jay

Click to expand...

By the way, I've previously pointed out that the dynamic range calculations presented here have been inaccurate, see http://www.dpreview.com/forums/read.asp?forum=1022&message=875476

Gerald

( http://homepage.mac.com/gcarter )

That was his first point.

His second said that the CCD cameras were sharper than the D30, and I pointed out that is nothing to do with the senor technology.

Ulysses said:

No one said anything at all about sharpening. The question has to
do with dynamic range versus noise, and where CCD and CMOS fall in
this respect.

Derek Clarke said:

Just how many more times does it have to be pointed out that the
D30 doesn't do much in-camera sharpening as a deliberate (and IMO
correct) design decision by Canon rather than anything to do with
the technology?

The best use of sharpening is after all other corrections have been
applied and when the image has been set to the final size for
output, not while it's still in the camera.

Click to expand...

Click to expand...

Indeed, CMOS sensors are higher noise than CCD sensors, all other things being equal.

But now that snag is well offset by the ability to add extra processing on-chip, so after that processing the end result has less noise yet can still span a reasonable ISO range.

Ron Shaw said:

Jay said:

After reading the Kodak's article again, I searched on DPReview to
a possible advantage of CCD over CMOS to justify Kodak's position.
The most obvious advantage is the dynamic range of CCD. Looking at
dynamic range (RAW or JPEG) for a given ISO CCD in D1 and S1 has a
higher dynamic range then CMOS.

Do Kodak's experts believe that most photographers will prefer
higher dynamic range to lower noise?? What is your opinion? Would
you trade off higher dynamic range for noise levels?

I personally prefer low noise images to higher dynamic range
because being and amateur, I have not run into many situations that
require increased dynamic range will benefit me. I would love to
hear your thoughts on this subject.

-Jay

Click to expand...

CMOS does not have an advantage when it comes to noise, either. In
fact, CMOS is worse in the noise department. One thing the
engineers at Canon had to do was work on the noise levels (with on
chip circuitry) to make noise levels acceptable. I remember reading
an article when the D30 was first released (on the net somewhere)
which talked quite a bit about that problem. I dont remember where,
now. Also, output level of each pixel on CMOS varies quite a bit,
and on chip circuitry (a FET amplifier for each pixel) was required
to equalize the output level. Because of these problems, CMOS was
always at a disadvantage compared to CCD's, and the savior for CMOS
is the ability to include on chip circuitry to address these
inherant problems. Not that this has much interrest to the end user.

Click to expand...