CMOS vs. CCD? (Any advantages)

[bearzimages]

In my opinion, images from a CCD appear to be crisper out of the
camera. The colors also appear to have a little more pop. Cmos,
requires more Post processing. I think you can find several threads
on this if you go back to the release of the Canon 1DMKII. As far
as High ISO is concerned, currently Cmos camera appear to hold an
advantage, but 1600 ISO was and is very usable on my Canon 1D. A lot
of that has to do with the sensor design, pixel pitch, pixel
placement, and software.

Jason

Those of my European friends who use Canon seem to always &
regardless of where they are, manage to capture a most enchanting
light quality in their work. Natural light, that is. Something that
very much appeals to my sense of aesthetics

I thought perhaps possibly a contributing factor just might be the
difference in sensors?

I mean this respectfully...

The most important part of a camera is the nut behind the viewfinder.
Upgrade that part and you'll get consistently better images from any
camera. A $200 night course form a local art college, community
college, high school, etc, is a good first step.

The D80 is a fine camera.

--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed

Hey nickleback,

I couldn't agree more with that, I've always said it's ultimately the
"photographer" (person) behind the lens thatr makes photograph. I
started out shooting film over 20 years ago, so the transition to
digital in 95 really wasn't an issue. With that said I still &
unapologetically,will from time to time refer to my old "basics of
photography" reference books.

My question was really an attempt to discover which, if any of
sensors, perhaps processes light "differently" enough to be
noticeable? if the technology & processing are are relatively
similar, then logically the answer remains in the light source of the
respective geographical locations.

Ultimately of course, it's really just a matter of personal astehtics.

--
'To use a camera as a means of artistic expression, a certain quality
of spirit must be brought to aid light & air' -Bayard Wootten 1926

http://www.bearzimages.com

http://www.flickr.com/photos/bearzimages

--
Jason Stoller [email protected]
We are just Beta Testers who pay the Camera Companies to test their
new products!

I appreciate the reply Jason, & your great avatar. Reminds me of "Escape from Planet of the Apes" (which is rather appealing-these days.)

Since I shoot high ISO less than 5% of the time, it's not even a factor for me in deciding which (brand) to buy. I've been leaning towards Canon just for something different. You're right about Canon developing their own software & sensors, instead of using other manufacturers --which I happen to like.

I started out shooting a Canon TX, an all manual 35mm camera-- back in 1981. I'm curious to see what a Canon & I can create together. Jason, what exactly do you like about Canon over the other brands?
--

'To use a camera as a means of artistic expression, a certain quality of spirit must be brought to aid light & air' -Bayard Wootten 1926

http://www.bearzimages.com

 

[Timmi]

Simple,

CMOS is designed or has its design advantage when it comes to speed and High ISO.

However, nothing beats a CCD at low ISO in IQ. And thats one... if not the only reason why MF Systems are using CCDs.

And I prefer a CCD over a CMOS. The Base ISO 100 IQ is just stunning.

B

--
iThink, therefore iMac

 

[alanr0]

http://www.dalsa.com/markets/ccd_vs_cmos.asp

Posted recently by Eric Fossum (3.2 MB PDF)
http://forums.dpreview.com/forums/read.asp?forum=1000&message=28545539

http://www.ericfossum.com/Presentations/CMOS%20Image%20Sensors%20Past%20Present%20and%20Future%20Jan%202008.pdf

My take on this:
CMOS has potential advantages over CMOS, but is more difficult to get right.

The implementation (micro-lenses, CFA, amplifiers, ADC etc.) is as important as the sensor technology.

At present the two technologies seem broadly comparable for photographic applications. Eric points out that CCD is more mature by around 20 years, so there is more scope for improvement with CMOS.

Cheers.
--
Alan Robinson

 

[John Sheehy]

Simple,

CMOS is designed or has its design advantage when it comes to speed
and High ISO.

However, nothing beats a CCD at low ISO in IQ. And thats one... if
not the only reason why MF Systems are using CCDs.

And I prefer a CCD over a CMOS. The Base ISO 100 IQ is just stunning.

Most of the sensors with the lowest noise are CMOS. The best MFDBs and the best DSLRs for read noise have about the same ratio of maximum signal to minimum noise (read noise) at the pixel level. MFDBs have many more pixels, usually, hence their lower overall image read noise.

--
John

 

[John Sheehy]

Wow John do you really believe that?

Believe WHAT ? Remember, I wasn't there in your head when you read something that you thought you took exception to; I have no idea which of the several things I said you are referring to. Do not assume telepathic abilities on the part of other posters.

The reason the D3 Sensor(sony) helped the D3's image is because
finally Nikon, after many years of saying they were not going to go
FF, finally produced a FF camera. One of the reason's it improved
the image and the ISO capablity is because they had more real estate
to space the pixels, and room for the larger size pixel pitch. I am
sure you are aware that the Canon 5D has 3 year old technology inside
it but was the benchmark for noise control at High ISO's for a long
time.

The 1Ds2 has less noise and more resolution than the 5D. Of course, you won't see it viewing both at 100% pixel view, which is what an ignorant public is prone to do.

And the 20D, 30D, 40D and 450D all outperform a 1.6x crop from the 5D, in terms of both noise and resolution. It is an inferior imager per unit of area, but happens to have a lot more area. Its pixel size is of no real benefit; the pixels from any of those other cameras, filling an entire frame, would be superior.

Again it is full frame with a larger Pixel pitch which does
make a difference. While we do not know for sure and can only hope,
a 5D replacement may be in the works. I am sure we will find out more
in the next few months.

Since Sony now is in the DSLR game and attempting to become a major
player, I suspect that their future FF camera's sensor will end up
equal to or surpass what Nikon offers. I doubt seriously that Canon
is going to sit back after leading the Digital market in noise
control for years, and allow Nikon, Sony, or anyone else to take that
away from them.

OK, now I see what it is that you found unbelievable.

Well, what if Canon runs out of ideas? Canon has been really short on ideas since the revolutionary low high-ISO noise of the 10D, and then the next generation with the 20D, 5D, and 1Dmk2. I can not rule out the possibility of Nikon taking the lead in high-ISO noise in crop cameras, too, and high-density full-frames.

To blindly assume that Canon will always be better is just fanboyism.

--
John

 

[mailman88]

CMOS sensors haven't changed much since 2001. What did change?....NOISE SUPPRESSION. Who ever has the best noise suppression on their cheap CMOS sensor is the winner. The reviews of cameras with CMOS and CCD sensors always mention noise suppression in their sensors when a new camera comes out. All the rave in the camera business is the Fuji S5 and its class-leading dynamic range. Its a great wedding camera, but offers nothing else. I guess CMOS's speed trumps CCD's low-noise and dynamic range. It boils down to money and cheap sunglasses.

 

[John Sheehy]

CMOS sensors haven't changed much since 2001. What did
change?....NOISE SUPPRESSION. Who ever has the best noise suppression
on their cheap CMOS sensor is the winner. The reviews of cameras with
CMOS and CCD sensors always mention noise suppression in their
sensors when a new camera comes out. All the rave in the camera
business is the Fuji S5 and its class-leading dynamic range. Its a
great wedding camera, but offers nothing else. I guess CMOS's speed
trumps CCD's low-noise and dynamic range. It boils down to money and
cheap sunglasses.

And what is your point, exactly? Noise and DR are the only variables in which the potential differences in CCD and CMOS can manifest themselves. There are no other qualities. If current CMOS is better at DR and noise sensitivity, there is nothing left for CCD to be better at! Everything is covered.

--
John

 

[Jason Stoller]

I started out shooting a Canon TX, an all manual 35mm camera-- back
in 1981. I'm curious to see what a Canon & I can create together.
Jason, what exactly do you like about Canon over the other brands?

At the time I bought the 1D, I was choosing between it and Nikon. I have shot several different camera brands over the years, but after Olympus dragged their feet making available a DSLR that was comparable to Nikon or Canon, both in capability, as well as accurate focus and speed, I like many others, who date back to the E-10, E-20, E-100RS days, jumped ship. At the time I was grabbed by local paper to shoot sports for them. I needed a camera body with accurate focus and good low light capablity.

Nikon and Canon were pretty much the only players in the game that produced a camera body that would meet my needs at the time. The Nikon body would have been less expensive to buy. Since no one offered full frame in a 35mm body, and the Canon 1D had less of a crop factor, I thought it would be an easier transition to my style of shooting. I also liked IS, and at the time Nikon just did not offer what I needed in that area.

So I took a risk because the 1D was just released at the time, sold my E-10, E-20, Nikon Film gear, and later my E-100RS, and all the Olympus lenses and accessories to raise the $5000 I needed at that time to get the 1D. The dealer I had at the time(who sells both Nikon and Canon) also advised me to get the 1D and offered me a money back 30 day trial period. Even though it was a lot of money for me at the time, the money back deal is was pushed me over the edge.

When I received the camera, even though it was heavy, I immediately fell in love with the body just as I did with the E-10 when I first picked it up. The focus speed and balance was amazing. I really learned to appreciate balance in a camera build.

Olympus does a very good job in the build of the E-1 and E-3 as well. I still have my E-1 but decided the E-3 copy I first purchased had some issues. I worked with Olympus to resolve them, but I still discovered that the E-3 just did not offer me what I needed at the time, so I let that body go. When I let it go, I purchased a 5D for IQ and the noise control at high ISO. Of course the build does not match the E-3. I also still have my E-1.

I personally am not adverse to the other brands out there. Each one of them has their pro's and con's. I have done my fair share of chasing gear. I even considered buying a D3 at one point, but changing systems can be an expensive proposition if you already own high dollar glass. Flash systems are not cheap either.

I will say that Nikon has a better warranty on their lenses than Canon. Unless things have changed, Nikon has a 5 year warranty where Canon only offers you 1 year.

Prior to the D3 and based on my needs, I would have always considered Nikon and other brands as a second choice on my list. Of course I would want some time shooting a D3 before I would make a final decision. I did not jump for the 1DMIII because of the focus issues, and the 1DMKII I own appears to still be performing well for me.

In closing, probably a couple of the main reason's I decided to buy Canon or Nikon is that you can go into just about any major city in the US and rent a body or lens. This is just not true for other brands.

You can lose your butt on the bodies, but both the lenses for Canon and Nikon are readily available both in the new and used markets, and seem to hold their values well.

--
Jason Stoller [email protected]

We are just Beta Testers who pay the Camera Companies to test their new products!

 

[Jason Stoller]

John, I did not say that I would blindly assume that Canon would always be better. What I will say is that while many other camera companies can be very innovative with their products, Canon has shown by their track record, that they are not asleep at the switch, and do not play second fiddle very long. Nikon on the other hand has been playing catch up for years. Now you are welcome to deny that fact but the larger force in the Digital Market place is Canon. Thats not fanboy talk, that is reality. A major factor to consider is who is making their own sensors. Here is an incomplete list, Canon, Panasonic, Samsung, Sony, Foveron. I am sure there may be a few others but Nikon is not on that list, so at the current time they are dependent on another company(Sony, an upcoming and direct competitor) to supply their sensors. Canon on the other hand has a very different situation.

Wow John do you really believe that?

Believe WHAT ? Remember, I wasn't there in your head when you read
something that you thought you took exception to; I have no idea
which of the several things I said you are referring to. Do not
assume telepathic abilities on the part of other posters.

The reason the D3 Sensor(sony) helped the D3's image is because
finally Nikon, after many years of saying they were not going to go
FF, finally produced a FF camera. One of the reason's it improved
the image and the ISO capablity is because they had more real estate
to space the pixels, and room for the larger size pixel pitch. I am
sure you are aware that the Canon 5D has 3 year old technology inside
it but was the benchmark for noise control at High ISO's for a long
time.

The 1Ds2 has less noise and more resolution than the 5D. Of course,
you won't see it viewing both at 100% pixel view, which is what an
ignorant public is prone to do.

And the 20D, 30D, 40D and 450D all outperform a 1.6x crop from the
5D, in terms of both noise and resolution. It is an inferior imager
per unit of area, but happens to have a lot more area. Its pixel
size is of no real benefit; the pixels from any of those other
cameras, filling an entire frame, would be superior.

I have no idea what you actually shoot, but it has been shown over and over that the Pixel Pitch and Size is a very important factor. If not then people have been wasting their money on higher end camera's for years. Why on earth would Nikon, Canon, Kodak, and even MF manufacturers bother using larger Pixels and sensors if your theory holds true?

Again it is full frame with a larger Pixel pitch which does
make a difference. While we do not know for sure and can only hope,
a 5D replacement may be in the works. I am sure we will find out more
in the next few months.

Since Sony now is in the DSLR game and attempting to become a major
player, I suspect that their future FF camera's sensor will end up
equal to or surpass what Nikon offers. I doubt seriously that Canon
is going to sit back after leading the Digital market in noise
control for years, and allow Nikon, Sony, or anyone else to take that
away from them.

OK, now I see what it is that you found unbelievable.

Well, what if Canon runs out of ideas? Canon has been really short
on ideas since the revolutionary low high-ISO noise of the 10D, and
then the next generation with the 20D, 5D, and 1Dmk2. I can not rule
out the possibility of Nikon taking the lead in high-ISO noise in
crop cameras, too, and high-density full-frames.

To blindly assume that Canon will always be better is just fanboyism.

A lot of 10D users made the switch to the 1D. If you do a search in the Canon forum, you might see the reason's why. Pixel density is not everything, and even in DPR's own statement on this site. Below is a quote from the article that appeared on the front page.

"Pixel Density is a calculation of the number of pixels on a sensor, divided by the imaging area of that sensor. It can be used to understand how closely packed a sensor is and helps when comparing two cameras with different sensor sizes or numbers of photosites (pixels). Because the light collecting area and efficiency of each photosite will vary between technologies and manufacturers, Pixel Density should not be used as an absolute metric for camera quality but instead to get an impression for how tightly packed the imaging chip is."

This quote says it all.

--
John

--
Jason Stoller [email protected]

We are just Beta Testers who pay the Camera Companies to test their new products!

 

[LJSmak]

and the ability to have support transistors at the
photosites.

Huh? That's like the "ability" of my car to use 4 wheels... that's just the way CMOS works and actually it is one of the drawbacks of CMOS, that limits the fill-factor compared to CCD. People work day and night to get the number of needed support transistors down.

Lourens

 

[nickleback]

CMOS sensors haven't changed much since 2001.

If nothing has changed since 2001, why did it take until 2008 for CMOS to dominate?

cheap CMOS sensor

Name a cheap APS-C or larger CMOS sensor.

The only article you linked to which was worth a damn:

http://www.dalsa.com/markets/ccd_vs_cmos.asp

says "Costs are similar at the chip level."

I guess CMOS's speed trumps CCD's low-noise and dynamic range.

CMOS at DSLR sizes is typically lower noise than CCD. Dynamic range is now comparable. That's why all the major DSLR manufacturers have cameras with CMOS sensors, including all top-of-the-line cameras.

It boils down to money and cheap sunglasses.

Again, got a reference to cheap CMOS? Earlier on it was thought that CMOS could be less expensive to make than CCD, but that it isn't the case.

--
Seen in a fortune cookie:
Fear is the darkroom where negatives are developed

 

[ejmartin]

Noise and DR are the only variables
in which the potential differences in CCD and CMOS can manifest
themselves. There are no other qualities. If current CMOS is better
at DR and noise sensitivity, there is nothing left for CCD to be
better at! Everything is covered.

What about pattern noises? I've not investigated CCD's in this regard, so I'm just asking; the readout methodology is different, so I would think one should be better than the other. Which is it?

--
emil
--



http://theory.uchicago.edu/~ejm/pix/20d/

 

[Jason Stoller]

One of the differences was Cmos used to be a lower power consuming chip. I do not know if that is still true. Another was heat. Maybe someone else can shed on light on these two factors.

Jason
--
Jason Stoller [email protected]

We are just Beta Testers who pay the Camera Companies to test their new products!

 

[John Sheehy]

Noise and DR are the only variables
in which the potential differences in CCD and CMOS can manifest
themselves. There are no other qualities. If current CMOS is better
at DR and noise sensitivity, there is nothing left for CCD to be
better at! Everything is covered.

What about pattern noises? I've not investigated CCD's in this
regard, so I'm just asking; the readout methodology is different, so
I would think one should be better than the other. Which is it?

I haven't isolated the line noises in recent CCD cameras, but I do remember isolating them in D200 RAWs, and the actual intensity of the line noises were greater than they were in the Canon 20D, even though you couldn't see them because of the higher 2-D noise.

--
John

 

[John Sheehy]

I have no idea what you actually shoot, but it has been shown over
and over that the Pixel Pitch and Size is a very important factor.

It is important. The denser the pixels, the better, with only one exception; currently, you can get slightly lower image-level read noise at ISO 1600 and above with multi-transistor CMOS photosites than you can with 2 micron CCD or nMOS photosites. At low ISOs, the latter have far less image read noise.

If
not then people have been wasting their money on higher end camera's
for years.

Nonsense. They bought what was best AVAILABLE . The optimum cameras have not been made available. Sometimes, they aren't the best. To get the most from a big-pixel, big-sensor camera, you have to use the whole frame. The 450D will outperform any big-pixel, big-sensor DSLR if you take a 1.6x crop from the latter.

Why on earth would Nikon, Canon, Kodak, and even MF
manufacturers bother using larger Pixels and sensors if your theory
holds true?

Whoahh, stop right there. I have never, ever said that big sensors aren't beneficial for noise. What I have said is that big complex CMOS pixels limit DR and have extra noise at low ISOs over small pixels. Big CCD pixels have no IQ benefits, whatsoever, all in the context of filling a given sensor size.

There are any number of a number of reasons why we don't have big sensors with small pixels now:

The manufacturers are ignorant and don't see the value.

The manufacturers see the value and are waiting to use this as another carrot to dangle for future upgrades.

The readout speed would become slower or the parallelization would need more power, or the systems would get more expensive.

The demand isn't there, because an ignorant public believes all this nonsense about the superiority of big pixels

and any of many other reasons.

OK, now I see what it is that you found unbelievable.

Well, what if Canon runs out of ideas? Canon has been really short
on ideas since the revolutionary low high-ISO noise of the 10D, and
then the next generation with the 20D, 5D, and 1Dmk2. I can not rule
out the possibility of Nikon taking the lead in high-ISO noise in
crop cameras, too, and high-density full-frames.

To blindly assume that Canon will always be better is just fanboyism.

A lot of 10D users made the switch to the 1D. If you do a search in
the Canon forum, you might see the reason's why.

I haven't profiled the 1D's shot noise, but being 1.7x the sensor size over the 1D, that's about 1.7x as many photons for the same exposure, so there is likely to be less shot noise. I have measured the 1D's read noise, because someone gave me a set of blackframes, and the read noise at the pixel level is clearly higher than the 10D's, especially at the higher ISOs, and all ISOs even higher at the image level because of the lower pixel density. Other than slightly less shot noise, what most people probably liked about the 1D was that it made almost no demands on the lenses, and was able to modulate to the maximum allowed by the AA filter even a little bit outside the DOF. I have heard many comments like this myself. One guy, who currently shoots a 1Dmk2 for birds says that he misses the 1D because he could blow the images up more. Another 1Dmk2 user told me he misses his Sigma SD9 for the same reason. Yet, when I take images from the 1D and the SD9, and interpolate them so that they match the higher pixel densities of current APS-C DSLRs, they look like soft junk, in comparison, or pixelated, if you choose the sharper look. I can't figure out what these people saw in their previous cameras. And maybe they are not quite so sure, either, because they stopped using them!

Pixel density is
not everything, and even in DPR's own statement on this site. Below
is a quote from the article that appeared on the front page.

"Pixel Density is a calculation of the number of pixels on a sensor,
divided by the imaging area of that sensor. It can be used to
understand how closely packed a sensor is and helps when comparing
two cameras with different sensor sizes or numbers of photosites
(pixels). Because the light collecting area and efficiency of each
photosite will vary between technologies and manufacturers, Pixel
Density should not be used as an absolute metric for camera quality
but instead to get an impression for how tightly packed the imaging
chip is."

This quote says it all.

It doesn't say it all, it says that pixel density is not a direct indication of camera quality, which is true. It leaves out the issue of whether higher pixel density is good or bad. And, from what I have seen of the opinions of the administrators here on the subject, I would have to disagree if they did make that judgement.

--
John

 

[John Sheehy]

and the ability to have support transistors at the
photosites.

Huh? That's like the "ability" of my car to use 4 wheels... that's
just the way CMOS works and actually it is one of the drawbacks of
CMOS, that limits the fill-factor compared to CCD. People work day
and night to get the number of needed support transistors down.

There is value in having these transistors, otherwise, they wouldn't be used. Tiny, simple CCD and nMOS pixels binned in software, not even in hardware, give less read noise than big, complex CMOS pixels at low ISOs. At high ISOs, the latter are still superior by up to a stop. The read noises we're seeing for P&S cameras with 2 micron pixel pitches are the product of inexpensive cameras; there are no big-budget cameras with large sensors full of tiny pixels, and if readout of such were properly parallelized, and more money was spent on quality, maybe read noise in those tiny pixels could be even lower than they are now.

--
John

 

[bearzimages]

My take on this:
CMOS has potential advantages over CMOS, but is more difficult to get
right.
The implementation (micro-lenses, CFA, amplifiers, ADC etc.) is as
important as the sensor technology.

At present the two technologies seem broadly comparable for
photographic applications. Eric points out that CCD is more mature
by around 20 years, so there is more scope for improvement with CMOS.

Cheers.
--
Alan Robinson

Hello Allan,

This is interesting & I appreciate "your take" on the subject. Now then would it be correct to understand, that the CMOS sensor, requires more in-camera processing (micro-lenses, CFA, amplifiers, ADC etc.) to record an image --than a CCD sensor?

--

'To use a camera as a means of artistic expression, a certain quality of spirit must be brought to aid light & air' -Bayard Wootten 1926
http://www.bearzimages.com

 

[alanr0]

My take on this:
CMOS has potential advantages over CMOS, but is more difficult to get
right.
The implementation (micro-lenses, CFA, amplifiers, ADC etc.) is as
important as the sensor technology.

At present the two technologies seem broadly comparable for
photographic applications. Eric points out that CCD is more mature
by around 20 years, so there is more scope for improvement with CMOS.

This is interesting & I appreciate "your take" on the subject. Now
then would it be correct to understand, that the CMOS sensor,
requires more in-camera processing (micro-lenses, CFA, amplifiers,
ADC etc.) to record an image --than a CCD sensor?

Depends what you mean by 'more'. For micro-lenses and CFA, there is little difference related to the silicon technology.

Most CMOS and CCD sensors use micro-lenses to increase light collection efficiency. Only part of the chip area is photosensitive, and the micro-lenses concentrate light at each pixel onto the sensitive area. In each case you need a micro-lens at each pixel. http://www.dpreview.com/learn/?/Glossary/Camera_System/microlenses_01.htm

CFA stands for colour filter array. Most cameras (with the notable exception of Sigma) use a Bayer array of colour filters so that each pixel responds mostly to red, green or blue light. Again, the requirements are similar for CMOS and CCD sensors.
http://www.dpreview.com/learn/?/Glossary/Camera_System/color_filter_array_01.htm
http://www.dpreview.com/learn/?/Glossary/Camera_System/sensors_01.htm

Both CCD and CMOS devices convert light (photons) into electric charge which is initially stored at each pixel. Both have a close to linear response, so that any difference in colour rendition or tone curve is due to the CFA implementation and on subsequent signal processing in the camera's firmware, rather than the silicon process.

CCD and CMOS devices differ in the way information is extracted from each pixel. In a CCD device, the charge is transferred by a sequence of 'electron wells' - sometimes described as a 'bucket-brigade'. At the edge of the chip the charge is converted to a voltage signal, which is (optionally) amplified, and read by an analog to digital converter (ADC).

CMOS devices convert the charge to voltage at each pixel. They are made by the same processes used to fabricate transistors and integrated circuits, and can incorporate active transistors to process the output at each pixel. The output from each pixel is switched in turn to the edge of the chip, and read by an ADC. In CMOS, the ADC can be fabricated on the same chip as the sensor, though I don't know how common this is for photographic applications.

The active processing at each pixel allows CMOS sensors to correct some of the sources of noise in the image. This gives them a significant advantage at high ISO (low light levels or short shutter speeds). Until recently, the noise reduction was less effective at lower ISO, so if you needed wide dynamic range at higher exposure values, there was less difference, with a possible advantage to CCD sensors.

As Emil pointed out earlier, Nikon's new D300 CMOS sensor has excellent DR at low ISO, and comparable high ISO noise to the better Canon CMOS offerings.

Getting back to your original question, CMOS does not need any additional in-camera hardware and image processing compared to CCD. However it does have the capability to integrate additional processing onto the sensor chip.

Something I expect to see more of in future is parallel processing. Noise can be reduced by giving the read-out circuitry more time to measure the output of each pixel. With a single ADC this means slower frame rates, but it is becoming viable to include multiple converters, and read-out data from several parts of the sensor in parallel. This has been done for both CMOS and CCD sensors, but CMOS makes it easier to include multiple ADC units on the sensor chip.

HTH
--
Alan Robinson

 

[William Carson]

Both CCD and CMOS devices convert light (photons) into electric
charge which is initially stored at each pixel. Both have a close to
linear response, so that any difference in colour rendition or tone
curve is due to the CFA implementation and on subsequent signal
processing in the camera's firmware, rather than the silicon process.

CCD and CMOS devices differ in the way information is extracted from
each pixel. In a CCD device, the charge is transferred by a sequence
of 'electron wells' - sometimes described as a 'bucket-brigade'. At
the edge of the chip the charge is converted to a voltage signal,
which is (optionally) amplified, and read by an analog to digital
converter (ADC).

CMOS devices convert the charge to voltage at each pixel. They are
made by the same processes used to fabricate transistors and
integrated circuits, and can incorporate active transistors to
process the output at each pixel. The output from each pixel is
switched in turn to the edge of the chip, and read by an ADC. In
CMOS, the ADC can be fabricated on the same chip as the sensor,
though I don't know how common this is for photographic applications.

The active processing at each pixel allows CMOS sensors to correct
some of the sources of noise in the image. This gives them a
significant advantage at high ISO (low light levels or short shutter
speeds). Until recently, the noise reduction was less effective at
lower ISO, so if you needed wide dynamic range at higher exposure
values, there was less difference, with a possible advantage to CCD
sensors.

As Emil pointed out earlier, Nikon's new D300 CMOS sensor has
excellent DR at low ISO, and comparable high ISO noise to the better
Canon CMOS offerings.

Getting back to your original question, CMOS does not need any
additional in-camera hardware and image processing compared to CCD.
However it does have the capability to integrate additional
processing onto the sensor chip.

Something I expect to see more of in future is parallel processing.
Noise can be reduced by giving the read-out circuitry more time to
measure the output of each pixel. With a single ADC this means
slower frame rates, but it is becoming viable to include multiple
converters, and read-out data from several parts of the sensor in
parallel. This has been done for both CMOS and CCD sensors, but CMOS
makes it easier to include multiple ADC units on the sensor chip.
Alan Robinson

I think the question I have and have had each time CMOS vs CCD discussions come up is: are the advances in CMOS technology you are talking about above able to compensate for the inherent difference in well depth of I think 2 to 1 of the CCD over CMOS in signal to noise. When you look at images from cameras with pixels of the same pitch, the noise is discernably larger in the CMOS sensor cameras over the CCD sensor cameras. There are some clear advantages of CMOS sensors in that it can use an electronic shutter thus avoiding rolling shutter time exposure in very high shutter speed, but the noise issue is unclear to me.
Will

 

[alanr0]

alanr0 wrote:
I think the question I have and have had each time CMOS vs CCD
discussions come up is: are the advances in CMOS technology you are
talking about above able to compensate for the inherent difference in
well depth of I think 2 to 1 of the CCD over CMOS in signal to noise.
When you look at images from cameras with pixels of the same pitch,
the noise is discernably larger in the CMOS sensor cameras over the
CCD sensor cameras. There are some clear advantages of CMOS sensors
in that it can use an electronic shutter thus avoiding rolling
shutter time exposure in very high shutter speed, but the noise issue
is unclear to me.
Will

First a quick discaimer - I am not fully up to date on specifics of current sensors. It would be good if the someone more clued-up could chip in (e.g, John Sheehy or Emil Martinec from this thread).

Next, it's not a one-dimensional comparison. There are significant differences between different CCD implementations, and between different CMOS implementations at the same pixel pitch.

I take it you are interested in low ISO performance. According to the data on Roger Clark's site http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/index.html#data

The Nikon D200 (CCD) had better dynamic range than the Canon 40D (CMOS) at low ISO, in spite of lower well capacity in the CCD sensor. However, the CMOS read noise was much lower at high ISO.

The newer Nikon D300 CMOS sensor has comparable high ISO noise and well capacity to the 40D, and better low ISO noise than both D200 and 40D.

The Samsung CMOS sensor used in the GX20 and Pentax K20D and is significantly noisier than sensors in the top end Nikon and Canon cameras, but I would not be surprised to see a more competitive offering next time around from Samsung.

I find it difficult to make out a consistent trend in well capacity, except that well capacity per square micron has improved over time for both CCD and CMOS sensors.

Cheers,
--
Alan Robinson