Pentax 645D, or 645Z: CCD vs. CMOS

[ondebanks]

My take on the difference between CCD and CMOS:

There is none. :lecture:

- Err, WHAT, Ray?

No, really. There is no inherent difference in what they do. Doped and biased silicon pixels trapping photoelectrons. That's it.

- Ah come off it, Ray! We can SEE the difference!

You can see differences alright - but they have nothing to do inherently with the underlying technology of CCD vs CMOS. Paul (wentbackward) identified many of the reasons above. Spectral response is the main one - a combination of front-of-sensor colour filtration, and photon absorption depth within the sensor. Both of these are at the whim of the designer, for both CCD and CMOS.

It so happens that some manufacturers dominate in making camera CCDs, some dominate in making CMOS, and they each have a typical way of doing these design aspects that influence colour; and therefore people have mistakenly associated one kind of colour response with CCDs and another (generally deemed inferior) kind with CMOS.

There are those who love the colour output from their Leicas (M8, M9, S, S2), or Pentax 645D, or 16/18/22/31/39/40/50 MP Hasselblads and Phase Ones; they may say it's because those cameras/backs use a CCD. Not quite! It's because they use a Kodak CCD. Same goes for all the the DALSA-chipped units. The secret sauce comes from the manufacturer, not whether it's a CCD. And then on top of that, you have how different integrators of the same sensor handle colour profiling, which is why some will prefer a Leaf back to a Phase back with the same DALSA sensor.

- But CMOS definitely has lower noise than CCDs, Ray! You often said it yourself!

Again, this is not an inherent property of the two kinds of device, but rather a consequence of engineering practicalities. Some early CMOS cameras and backs for photography, like the Leaf C-MOST and Kodak 14n/SLR-n, actually had worse readout noise than their CCD contemporaries. If you read a CCD out slowly enough, or employ EM-gain at the expense of dynamic range, it can be extremely low noise as well. Again, different engineering explains why Canons are noisy and have limited DR at low ISO (often worse on both counts than even older CCDs) while increasingly, Sonys (and Nikons, Pentaxes) have lower noise and high DR at low ISO, even though they're all CMOS.

- But...my CMOS camera gives me better shadow detail and worse highlight recovery and specular handling, than with my CCD camera. Explain that!

First, play fair: make sure you are using two cameras with very similar pixel sizes (a reasonable proxy for pixel full well capacity). If your CMOS camera is a DSLR/CSC and your CCD camera is an MFD unit, chances are that the CCD has larger pixels. And it's the area that matters - the square of the pixel width. So 5 microns CMOS might seem close enough to 6 microns CCD, but 25 square microns is a lot less than 36.

Now, underexpose your CMOS camera by about 1 stop. If you bracket a bit, you'll find an exposure where the highlights are about as good as the CCD, and the shadows have degraded but are still no worse than the CCD's.

You will probably have to add a further -1 stop underexposure to the CMOS camera for every stop above base ISO that you shoot the CCD, since most CCD units, especially MFD ones, fake higher ISO by simply underexposing, which gives naturally enough gives even better highlight recovery, since what were highlights slide down towards the mid-tones and the shadows are utterly foresaken.

- And the film-like quality and smooth highlight tonal gradations that people associate with CCDs? I suppose you have an answer for them, too?

I do. It goes back to the "high signal, high noise" nature of the typical CCD photographic capture. The typical CMOS capture OTOH has "moderate signal, low noise". (Again, I stress that what's typically engineered is not an indicator of absolute underlying performance capabilities). Plot signal to noise versus intensity (or Adams' zone scale) and the two representative curves cross over in a way which explains these visual effects. Manipulating these curves is essentially what software like DxO Filmpack does to create more film-like images from digital.

- Look, we all know that CMOS can do Live-view and Video and CCD does them poorly, if at all. That must count as a real difference?

It's only a difference in off-the-sensor routing of the signals. If you force* all the data from a CMOS sensor through a single serial pipeline like a CCD does, you'll find them no different in those regards. [* Why on earth you'd do that is not the point...I'm just saying that again it's not inherent to the sensor internals]

Ray....clicking "submit" and waiting for the :cussing:

 

[ErikKaffehr]

Ray,

A lot of good points. A few remarks:

Regarding highlight tonality/separation/DR I try to keep my exposures reasonably ETTR on both DSLR and MFD. I also frequently check my exposures in post using RawDigger, it is a learning experience.

So I try to expose to just avoid clipping, and at those exposures I see very little difference between my Sony DSLRs and my P45+ at base ISO, and base ISO is what I essentially use. (On DSLR sometimes I use high ISOs, but in those cases I am more concerned about getting the picture than maintaining highlights.)

Something I have observed is that Capture One generally applies a "film curve" that causes the image to be overexposed, so using C1 can "lull one into" underexposing, which may help in protecting highlights.

You also seem to suggest that square pixel size is proportional to full well capacity (FWC). Is FWC not dependent on the process deployed? From sensor data I have seen it may be concluded that some pixels have very large FWC (like Nikon D4). Is FWC over pixel area constant or is it improving over time with better processes?

Something that probably plays a role is that there is a tendency to thinner design rules resulting in a better ratio of active area vs. passive components. It seems that the number of passive components is reduced in modern CMOS designs, using shared transistors for instance.

I may also feel that white balance plays a significant role in colour rendition, which may be underestimated. Getting a good white balance may be more important than small differences in CFA design.

Finally, finely tuned profiles may make a significant difference.

Best regards
Erik

My take on the difference between CCD and CMOS:

There is none. :lecture:

- Err, WHAT, Ray?

No, really. There is no inherent difference in what they do. Doped and biased silicon pixels trapping photoelectrons. That's it.

- Ah come off it, Ray! We can SEE the difference!

You can see differences alright - but they have nothing to do inherently with the underlying technology of CCD vs CMOS. Paul (wentbackward) identified many of the reasons above. Spectral response is the main one - a combination of front-of-sensor colour filtration, and photon absorption depth within the sensor. Both of these are at the whim of the designer, for both CCD and CMOS.

It so happens that some manufacturers dominate in making camera CCDs, some dominate in making CMOS, and they each have a typical way of doing these design aspects that influence colour; and therefore people have mistakenly associated one kind of colour response with CCDs and another (generally deemed inferior) kind with CMOS.

There are those who love the colour output from their Leicas (M8, M9, S, S2), or Pentax 645D, or 16/18/22/31/39/40/50 MP Hasselblads and Phase Ones; they may say it's because those cameras/backs use a CCD. Not quite! It's because they use a Kodak CCD. Same goes for all the the DALSA-chipped units. The secret sauce comes from the manufacturer, not whether it's a CCD. And then on top of that, you have how different integrators of the same sensor handle colour profiling, which is why some will prefer a Leaf back to a Phase back with the same DALSA sensor.

- But CMOS definitely has lower noise than CCDs, Ray! You often said it yourself!

Again, this is not an inherent property of the two kinds of device, but rather a consequence of engineering practicalities. Some early CMOS cameras and backs for photography, like the Leaf C-MOST and Kodak 14n/SLR-n, actually had worse readout noise than their CCD contemporaries. If you read a CCD out slowly enough, or employ EM-gain at the expense of dynamic range, it can be extremely low noise as well. Again, different engineering explains why Canons are noisy and have limited DR at low ISO (often worse on both counts than even older CCDs) while increasingly, Sonys (and Nikons, Pentaxes) have lower noise and high DR at low ISO, even though they're all CMOS.

- But...my CMOS camera gives me better shadow detail and worse highlight recovery and specular handling, than with my CCD camera. Explain that!

First, play fair: make sure you are using two cameras with very similar pixel sizes (a reasonable proxy for pixel full well capacity). If your CMOS camera is a DSLR/CSC and your CCD camera is an MFD unit, chances are that the CCD has larger pixels. And it's the area that matters - the square of the pixel width. So 5 microns CMOS might seem close enough to 6 microns CCD, but 25 square microns is a lot less than 36.

Now, underexpose your CMOS camera by about 1 stop. If you bracket a bit, you'll find an exposure where the highlights are about as good as the CCD, and the shadows have degraded but are still no worse than the CCD's.

You will probably have to add a further -1 stop underexposure to the CMOS camera for every stop above base ISO that you shoot the CCD, since most CCD units, especially MFD ones, fake higher ISO by simply underexposing, which gives naturally enough gives even better highlight recovery, since what were highlights slide down towards the mid-tones and the shadows are utterly foresaken.

- And the film-like quality and smooth highlight tonal gradations that people associate with CCDs? I suppose you have an answer for them, too?

I do. It goes back to the "high signal, high noise" nature of the typical CCD photographic capture. The typical CMOS capture OTOH has "moderate signal, low noise". (Again, I stress that what's typically engineered is not an indicator of absolute underlying performance capabilities). Plot signal to noise versus intensity (or Adams' zone scale) and the two representative curves cross over in a way which explains these visual effects. Manipulating these curves is essentially what software like DxO Filmpack does to create more film-like images from digital.

- Look, we all know that CMOS can do Live-view and Video and CCD does them poorly, if at all. That must count as a real difference?

It's only a difference in off-the-sensor routing of the signals. If you force* all the data from a CMOS sensor through a single serial pipeline like a CCD does, you'll find them no different in those regards. [* Why on earth you'd do that is not the point...I'm just saying that again it's not inherent to the sensor internals]

Ray....clicking "submit" and waiting for the :cussing:

 

[jerome_m]

My take on the difference between CCD and CMOS:

There is none. :lecture:

- Err, WHAT, Ray?

No, really. There is no inherent difference in what they do. Doped and biased silicon pixels trapping photoelectrons. That's it.

- Ah come off it, Ray! We can SEE the difference!

You can see differences alright - but they have nothing to do inherently with the underlying technology of CCD vs CMOS. Paul (wentbackward) identified many of the reasons above. Spectral response is the main one - a combination of front-of-sensor colour filtration, and photon absorption depth within the sensor. Both of these are at the whim of the designer, for both CCD and CMOS.

It so happens that some manufacturers dominate in making camera CCDs, some dominate in making CMOS, and they each have a typical way of doing these design aspects that influence colour; and therefore people have mistakenly associated one kind of colour response with CCDs and another (generally deemed inferior) kind with CMOS.

There are those who love the colour output from their Leicas (M8, M9, S, S2), or Pentax 645D, or 16/18/22/31/39/40/50 MP Hasselblads and Phase Ones; they may say it's because those cameras/backs use a CCD. Not quite! It's because they use a Kodak CCD. Same goes for all the the DALSA-chipped units. The secret sauce comes from the manufacturer, not whether it's a CCD. And then on top of that, you have how different integrators of the same sensor handle colour profiling, which is why some will prefer a Leaf back to a Phase back with the same DALSA sensor.

- But CMOS definitely has lower noise than CCDs, Ray! You often said it yourself!

Again, this is not an inherent property of the two kinds of device, but rather a consequence of engineering practicalities. Some early CMOS cameras and backs for photography, like the Leaf C-MOST and Kodak 14n/SLR-n, actually had worse readout noise than their CCD contemporaries. If you read a CCD out slowly enough, or employ EM-gain at the expense of dynamic range, it can be extremely low noise as well. Again, different engineering explains why Canons are noisy and have limited DR at low ISO (often worse on both counts than even older CCDs) while increasingly, Sonys (and Nikons, Pentaxes) have lower noise and high DR at low ISO, even though they're all CMOS.

- But...my CMOS camera gives me better shadow detail and worse highlight recovery and specular handling, than with my CCD camera. Explain that!

First, play fair: make sure you are using two cameras with very similar pixel sizes (a reasonable proxy for pixel full well capacity). If your CMOS camera is a DSLR/CSC and your CCD camera is an MFD unit, chances are that the CCD has larger pixels. And it's the area that matters - the square of the pixel width. So 5 microns CMOS might seem close enough to 6 microns CCD, but 25 square microns is a lot less than 36.

Now, underexpose your CMOS camera by about 1 stop. If you bracket a bit, you'll find an exposure where the highlights are about as good as the CCD, and the shadows have degraded but are still no worse than the CCD's.

You will probably have to add a further -1 stop underexposure to the CMOS camera for every stop above base ISO that you shoot the CCD, since most CCD units, especially MFD ones, fake higher ISO by simply underexposing, which gives naturally enough gives even better highlight recovery, since what were highlights slide down towards the mid-tones and the shadows are utterly foresaken.

- And the film-like quality and smooth highlight tonal gradations that people associate with CCDs? I suppose you have an answer for them, too?

I do. It goes back to the "high signal, high noise" nature of the typical CCD photographic capture. The typical CMOS capture OTOH has "moderate signal, low noise". (Again, I stress that what's typically engineered is not an indicator of absolute underlying performance capabilities). Plot signal to noise versus intensity (or Adams' zone scale) and the two representative curves cross over in a way which explains these visual effects. Manipulating these curves is essentially what software like DxO Filmpack does to create more film-like images from digital.

- Look, we all know that CMOS can do Live-view and Video and CCD does them poorly, if at all. That must count as a real difference?

It's only a difference in off-the-sensor routing of the signals. If you force* all the data from a CMOS sensor through a single serial pipeline like a CCD does, you'll find them no different in those regards. [* Why on earth you'd do that is not the point...I'm just saying that again it's not inherent to the sensor internals]

Ray....clicking "submit" and waiting for the :cussing:

Exactly my take on the subject as well.

 

[D&A]

I may be mistaken and of course can only speak for myself, but when I express to others that so far I generally much prefer the output of CCD based cameras vs. those that are CMOS based, I'm certainly not referring to just the actual sensors used in these cameras. My preferences are based on output from these cameras and whether it's simply sensor related, associated color array's or filters or even associated components used in conjuntion with a particular sensor type, is what I'm trying to express. These differences in output so far between many of the CCD based vs. CMOS based cameras is what I am seeing in output from these cameras and although post processing and other techniques can often equalize and bring some similarities to the output from both, I still find there are notable differences.

Each though I feel has some very distinct advantages (and disadvantages) and not at all saying either one is superior to the other. It depends on what positive factors one puts the most value on with regards to not only output from a given cameras, but it's ability to capture that image in a broad range of shooting circumstances and how well the camera can handle it.

Dave (D&A)

 

[fotografz]

I may be mistaken and of course can only speak for myself, but when I express to others that so far I generally much prefer the output of CCD based cameras vs. those that are CMOS based, I'm certainly not referring to just the actual sensors used in these cameras. My preferences are based on output from these cameras and whether it's simply sensor related, associated color array's or filters or even associated components used in conjuntion with a particular sensor type, is what I'm trying to express. These differences in output so far between many of the CCD based vs. CMOS based cameras is what I am seeing in output from these cameras and although post processing and other techniques can often equalize and bring some similarities to the output from both, I still find there are notable differences.

Each though I feel has some very distinct advantages (and disadvantages) and not at all saying either one is superior to the other. It depends on what positive factors one puts the most value on with regards to not only output from a given cameras, but it's ability to capture that image in a broad range of shooting circumstances and how well the camera can handle it.

Dave (D&A)

I agree Dave. I also see notable differences.

Who cares how it happens? I don't need or want to know how the sausage is made, just that it tastes good.

If you like one thing over another that's creative preference, and no amount of geekified rationale is going to change that.

- Marc

 

[Shashin]

Ray does not actually state the difference that is perceived is not real. He is simply identifying where it might come from. What I find interesting is that folks get defensive about it. Personally, if you like the "CCD" look, it might be good to try to analyze where the look is coming from. CCD cameras are not the future and at some point you will not have that choice. If you believe it is solely the sensor tech, then you are stuck. If you don't, it means there can be solutions.

 

[fotografz]

Ray does not actually state the difference that is perceived is not real. He is simply identifying where it might come from. What I find interesting is that folks get defensive about it. Personally, if you like the "CCD" look, it might be good to try to analyze where the look is coming from. CCD cameras are not the future and at some point you will not have that choice. If you believe it is solely the sensor tech, then you are stuck. If you don't, it means there can be solutions.

Clear preferences and conviction in your choices is now being "defensive"?

Don't care where the differences come from. I'll leave that to those that like that sort of speculative analization.

That CCDs are not the future is obvious, but I'll cross that bridge when I get to it just like with film. Besides, I already have plenty of CMOS cameras to practice on in the meantime.

There is enough stuff to deal with in the present without conjuring up "what ifs and maybe's".

- Marc

 

[D&A]

Clear preferences and conviction in your choices is now being "defensive"?

Don't care where the differences come from. I'll leave that to those that like that sort of speculative analization.

That CCDs are not the future is obvious, but I'll cross that bridge when I get to it just like with film. Besides, I already have plenty of CMOS cameras to practice on in the meantime.

There is enough stuff to deal with in the present without conjuring up "what ifs and maybe's".

- Marc

+1, I quite agree. The earlier (and earliest) pro level digital cameras produced images that were a far cry from what was achievable with film in terms of realistic and convincing output. Most serious photographers would agree with this. Yet here we are in 2014 and although many would say film and digital output is different, the vast majority I believe are now convinced of the superior output of digital. Of course there are exceptions, there always are.

I see possibly two routes for the future of CCD sensors in conventional photography. Either CMOS will continue to evolve and improve to the point where even diehard CCD fans will see the equalization of output from a CMOS camera being close to, equal to, or superior to that from a CCD based one. I don't believe we're there yet.

The other senario is similar to the analogue audio world vs. digital. Specialized or limited production of high quality CCD based cameras will continued to be developed and produced for those diehards that appreciate it's output and perceived image characteristic advantages and therefore a small dedicated market will grow and develop around this. Similar to high end turntable and associated analogue equipment manufacturers along with analog record producers.

It's hard to say exactly what the near future holds with regards to all this. As interesting as it would be to know the science and specifics of why I find CCD output generally more attractive, it's not on the top of my list to find out. I simply know what my eyes see with regards to esthetic cues and intrinsic value in the CCD camera's output and thus what I'm pleased to pass along, especially if its to a client.

With that said, I respect that others may have a very different opinion and their own personal preferences.

Dave (D&A)

 

[tsjanik]

My take on the difference between CCD and CMOS:

......Doped and biased silicon pixels trapping photoelectrons. That's it.

...............:Ray....clicking "submit" and waiting for the :cussing:

OK Ray, I'll bite and play Devil's Advocate :angel:.

Silver halide crystals are simply photon counters and yet no one would state that all films are the same. You argue that all the perceived difference in CMOS vs. CCD occurs downstream of the photo event (aside from pixel size). I have no expertise in sensors, but if that is the case, why hasn't someone produced a CMOS sensor with the qualities some of us find in a CCD sensor?

Tom

 

[Shashin]

OK Ray, I'll bite and play Devil's Advocate :angel:.

Silver halide crystals are simply photon counters and yet no one would state that all films are the same. You argue that all the perceived difference in CMOS vs. CCD occurs downstream of the photo event (aside from pixel size). I have no expertise in sensors, but if that is the case, why hasn't someone produced a CMOS sensor with the qualities some of us find in a CCD sensor?

Tom

Tom, if you don't mind, I will chip in. Sensors have a linear response and need no pigments to reproduce color. Film has a non-linear response (they don't make good photon counters) and require dyes to reproduce color. The dye layers are also stacked, which complicates the process. The color is created through an indirect process of dye couplers with silver which needs to be removed--this get into things like the efficiency of the dyes and whether each dye layer has a similar non-linear response, and when you realize that equal color density in the dye layers don't actually produce a neutral tone, film get really messy. When going to print, you have the added complexity of the print material is also non-linear and matching tone curves become very complex.

I don't think Ray state everything was down stream. There is the spectral response of the sensor and the color filter array.

I like both my p25+ and my 645D. I also shoot Sony CMOS. I might be able to show a different result, but both results would still be good, neither of which could be said to be better.

Now, I am not a big fan of magic when it comes to imaging. I do not believe things have mystical qualities. Now, I might have a preference to certain cameras, sensors, or films, but I also know there are too many variables in the process to equate any of them as having special qualities--most likely I have simply done the work with them and learnt a process to get the result I like.

So, can you or anyone else show the qualities that are in a CCD vs. a CMOS? I am really interested. This is why in my initial post in this thread I did not recommend one sensor over the other because I am really hard pressed to show a difference--my personal feelings are really not enough.

 

[Shashin]

Clear preferences and conviction in your choices is now being "defensive"?

Don't care where the differences come from. I'll leave that to those that like that sort of speculative analization.

That CCDs are not the future is obvious, but I'll cross that bridge when I get to it just like with film. Besides, I already have plenty of CMOS cameras to practice on in the meantime.

There is enough stuff to deal with in the present without conjuring up "what ifs and maybe's".

- Marc

Marc, if you don't know and don't care where the perceived difference between CCD and CMOS sensors come from, how can you have a conviction that there is actually a difference in the sensor technology?

I certainly don't mind personal preference. I have them too. But before I want someone else to suffer my convictions, I would want to know if they are willing to do the time. Maybe defensive was the wrong word, but there seems to be very strong feelings here. And not just positive for CCD, but negative for CMOS. I guess I just don't understand--I was always a Kodak film guy, which my Fuji friends could not fathom. The more things change...

 

[fotografz]

Shashin, I agree that one shouldn't recommend one sensor over another based just on personal preferences. There are far too many "subjective taste variables," "imaging chain variables", or "specific response needs" in that equation.

On the other hand, why the need to scientifically quantify the differences of those same variables in one part of the imaging chain?

In my case, my eye does the "quantifying". For all I know, it is simply coincidence that CCD cameras had just the right read-out, lens performance, and processing response to the prevailing software of the time that collectively appealed to my eye, where CMOS did not appeal as much … and still doesn't.

In short, it doesn't matter why.

My own trek is what has formed my considered "creative" opinions. It isn't some argument just for the sake of arguing.

Despite all the short comings of the Contax N Digital camera, I preferred the images from this camera with its' FF 6 meg Phillips CCD sensor to any Canon or Nikon of the day, and even afterwards (many of which I also owned).

Likewise, the Leica DMR with its' 1/37x crop frame 10 meg Kodak CCD and Imacon electronics, delivered images I vastly preferred to my Canon 5D with a FF 12.7 meg CMOS sensor using the same adapted R, or Zeiss CY and N lenses. Many DMR owners persist using that camera despite all the trials and tribulations of an abandoned platform.

During this trek I even noted the differences with-in MF CCD "look and feel" … Kodak 39 meg CCD Hasselblad verses Dalsa 33 meg CCD Mamiya for example … both using adapted Zeiss CFi/CFE optics.

My last foray into "quantify with the eye" was recently testing a Leica M(240), which is a much improved platform over the M9P I owned … the M(240) with a CMOSIS sensor produced image characteristics I did not like. I spent two solid weeks experimenting with post techniques, options and profiles provided by users to no avail. Still do not like the image characteristics from this camera.

So far, I see nothing compelling from the 50c sensor, and as Bob said, meh. I remain open to its' unexplored possibilities … and I fully understand that for many, the quantifiable operational advantages like LV and higher ISO performance far outweigh any subjective reactions to image qualities … however, that is not my personal criteria, nor do I feel that my current choices limit my creative options due to what and how I make photographs.

- Marc

 

[D&A]

If plagiarism was a positive accepted norm in our society and sincerest form flattery, I'd simply repost Marc's post above as my own. Photography is a visual art form, so is painting and sculpturing and a good deal of what one sees and likes is both visceral and highly subjective. Sure to the trained eye, professional or otherwise, certain defined aspects are often part of what makes a picture or painting deemed successful or not. Yet there is also the personal subjective part, the visceral part, that's very individualistic, which appeals to one's senses. They can try to describe what they see or hear, but it's often not something that one can fit neatly into a shoebox or prove with a mathematical formula.

This doesn't just apply to the technical aspects of the work of art and whether it meets some standard or appeals to some based on its form and content, but also includes elements that simply appeal to the senses. In fact in some cases from a purely technical standpoint, aspects of the picture, painting etc. may simply fall short. Maybe colors aren't perfectly realistic, dynamic range too constricted, or heaven forbid, one cannot put their finger on it but on all accounts it should be inferior, but its not. Simply put one or some combination of factors that are attractive are working to make this given image or painting very appealing.

Sure it would be great if one could measure and duplicate it, or developed a simply receipe that can be passed on to the others, but as is often the case, the resulting works of art are simply facsimiles.

At this time I'm not sure what's responsible for my preference of output from CCD sensor based cameras vs. CMOS ones, but I know what my eyes and senses tell me. Same thing happened to me as Marc experienced when I compared files from the Leica M240 vs. the M9 or worked with the Contact N digital camera. There was an immediate positive response to much of what I was seeing. Technically some of those cameras often fell short and in the case of the M240, it's operational prowess ran circles around it predisessor. Yet when it came down to it, image output took precedence over everything else. Same thing in my opinion will occur when I compare the Pentax 645D vs. the new 645Z. At this point I have no idea what that comparison will turn out to be like, but first and foremost it will be on the image qualities that appeal to me.

All this is in many respects like the often cited high end audio analogy many compare photography to. Listening to recorded music is highly subjective and visceral responses to various links in the reproduction chain is often opened to debate. Technical and scientific measurements only account for part of the story and what people hear and like from say a given speaker system or audio amplifier for example. This is much akin to what ones sees in image output from a digital file....there is a measured degree of subjectivity that enters into the equation, that simply can't be explained away simply because the numbers don't add up or the science behind it say there should be no difference.

Dave (D&A)

 

[Shashin]

Marc and Dave, Ray does not state there is no difference from a camera with a CCD and CMOS. There is a difference. However, the reason is not simply because it is a CCD or CMOS architecture.

However, neither of you seem to know or even care about the science. Since the science can clearly point to reasons why you are seeing a difference, why are you dismissing it? Don't you think the science that can bring you these marvelous machines could also analyze them? In fact, it can. Both Ray and I have indicated where the difference may come from.

As far as observer bias, science covers that very well. There are many factors why someone prefers something over another. Quite well documented as far as I have found.

So, if Ray and I are essentially agreeing with you, except for the root cause for which you cannot point to, what is your argument? BTW, I am not saying you are not seeing a difference nor that your preference is invalid.

As far as why I am interested in the science/details is simply because I find it very interesting. Also, if I know the factors involved, it gives me greater insight into the photographic process as well as greater control.

This kinds of reminds me about the Kodak and Dalsa sensor debates. Both CCD, but both giving different results. There has to be something more than the architecture...

 

[fotografz]

Marc and Dave, Ray does not state there is no difference from a camera with a CCD and CMOS. There is a difference. However, the reason is not simply because it is a CCD or CMOS architecture.

However, neither of you seem to know or even care about the science. Since the science can clearly point to reasons why you are seeing a difference, why are you dismissing it? Don't you think the science that can bring you these marvelous machines could also analyze them? In fact, it can. Both Ray and I have indicated where the difference may come from.

As far as observer bias, science covers that very well. There are many factors why someone prefers something over another. Quite well documented as far as I have found.

So, if Ray and I are essentially agreeing with you, except for the root cause for which you cannot point to, what is your argument? BTW, I am not saying you are not seeing a difference nor that your preference is invalid.

As far as why I am interested in the science/details is simply because I find it very interesting. Also, if I know the factors involved, it gives me greater insight into the photographic process as well as greater control.

This kinds of reminds me about the Kodak and Dalsa sensor debates. Both CCD, but both giving different results. There has to be something more than the architecture...

Thanks for the post.

I DO care about the science of photography, but only to the degree I have to in order to make something work. I'm not an engineer nor have any plans to be one. My education, training and background is in art … drawing, painting, design, and photography. Photography is just another medium to express myself with.

I do not discount those who find it all fascinating, after all, it is scientists and engineers who brought us these tools.

Yet, in the same manner, I do not need to see the schematic of the machine that made my brushes, or how a chemist formulated my paints.

I simply don't have the inclination or the interest in all the detailed reasoning and rationale' as to why, or how, any given camera delivers that which I prefer, just that it does.

So, it is not a matter of dismissing it, instead more the matter of how much do I need to know to make images that please my eye? For me, I only need to see the differences that I prefer, not know all the complex reason why … because knowing why, isn't going to change anything.

When and if it changes, I won't care why either, just that it did change, and I'll move to that if it helps me.

We all go at this in our own individualistic way … and thankfully, we have choices in the way we think, create, and the tools we use.

- Marc

BTW, I also noted the Kodak/Dalsa discussions, and experienced the two first hand. In either case, I preferred the results to CMOS based images.

 

[D&A]

Hi Will,

I too very much appreciate your response and instead of repeating much of what Marc just wrote, I'll add a few brief thoughts.

I also have a keen interest on the science on how and why things work. If I didn't I wouldn't have extensive education and training in the sciences not would it be part of my life's endeavors. When the time comes when more investigation is done into the how's and why's CCD output seems to be favored by a substantial group of individuals vs. CMOS based cameras, I would be glad to be involved in the discussion.

Right now those differences are for the most part conjecture and so debating why CCD output for some is superior to CMOS is simply different theoretical points of view Look how long it's been since digital recorded music has been available to the masses and still no one has a clear explanation to the science as to why some find analogue recorded music preferable to digital. Same discussions goes for tube amps, preamps, and other similar equipment vs. solid state designs. Science has given us lots of theories and so me good evidence (sampling rates etc.) but nothing definitive.

So believe me, both science and art is my life's work, but there is so much energy and time I can devote to various sub disciplines to gain a full appreciation and understanding...especially when no one has made a direct correlation between many of the how's and why's one sensor or camera's output is favored over another. All interesting stuff but what I like and see in my imagery takes precedence.

Dave (D&A)

 

[tsjanik]

Tom, if you don't mind, I will chip in. Sensors have a linear response and need no pigments to reproduce color. Film has a non-linear response (they don't make good photon counters) and require dyes to reproduce color. The dye layers are also stacked, which complicates the process. The color is created through an indirect process of dye couplers with silver which needs to be removed--this get into things like the efficiency of the dyes and whether each dye layer has a similar non-linear response, and when you realize that equal color density in the dye layers don't actually produce a neutral tone, film get really messy. When going to print, you have the added complexity of the print material is also non-linear and matching tone curves become very complex.

I don't think Ray state everything was down stream. There is the spectral response of the sensor and the color filter array.

I like both my p25+ and my 645D. I also shoot Sony CMOS. I might be able to show a different result, but both results would still be good, neither of which could be said to be better.

Now, I am not a big fan of magic when it comes to imaging. I do not believe things have mystical qualities. Now, I might have a preference to certain cameras, sensors, or films, but I also know there are too many variables in the process to equate any of them as having special qualities--most likely I have simply done the work with them and learnt a process to get the result I like.

So, can you or anyone else show the qualities that are in a CCD vs. a CMOS? I am really interested. This is why in my initial post in this thread I did not recommend one sensor over the other because I am really hard pressed to show a difference--my personal feelings are really not enough.

Hi Will,
I understand the chemistry of film fairly well; the electronics of digital much less so.
I am not saying that CCD and CMOS are or aren't inherently different in how they render an image, I am simply saying that I, and many others, find the existing output different. My point is that if the rendering is a consequence of engineering choices, why not design a CMOS that produces the same output of a CCD; it would make a lot of customers very happy. I certainly didn't say anything about magic or mystical properties; the output is different, the question is why.

Tom

 

[jerome_m]

Thanks for the post.

I DO care about the science of photography, but only to the degree I have to in order to make something work. I'm not an engineer nor have any plans to be one. My education, training and background is in art … drawing, painting, design, and photography. Photography is just another medium to express myself with.

I do not discount those who find it all fascinating, after all, it is scientists and engineers who brought us these tools.

Yet, in the same manner, I do not need to see the schematic of the machine that made my brushes, or how a chemist formulated my paints.

I simply don't have the inclination or the interest in all the detailed reasoning and rationale' as to why, or how, any given camera delivers that which I prefer, just that it does.

So, it is not a matter of dismissing it, instead more the matter of how much do I need to know to make images that please my eye? For me, I only need to see the differences that I prefer, not know all the complex reason why … because knowing why, isn't going to change anything.

If I may voice an opinion: I think that this is an important point on which I do not agree. Contrary to you, I believe that finding out the complex scientific reason(s) why a particular camera gives a pleasing output and another less so is going to change a few things.

At least that is how it works for me. Understanding the reasons why a given camera gives a given output often helped me to produce the look I wanted, and not always necessarily with that camera.

 

[fotografz]

If I may voice an opinion: I think that this is an important point on which I do not agree. Contrary to you, I believe that finding out the complex scientific reason(s) why a particular camera gives a pleasing output and another less so is going to change a few things.

At least that is how it works for me. Understanding the reasons why a given camera gives a given output often helped me to produce the look I wanted, and not always necessarily with that camera.

Personally, I think that is a very valid perspective. Especially if you are trying to create a body of work that has a signature look and feel but use different cameras.

The question that then arises is just how much does each person want to immerse themselves into a regimen involving a lot of variables? How consistant can we make something look like something else?

For example, the Leica S and M9 have a synergistic look and feel … but try as I might, I could not get the M(240) files in the same area. They are simply different.

- Marc

 

[Shashin]

Hi Will,
I understand the chemistry of film fairly well; the electronics of digital much less so.
I am not saying that CCD and CMOS are or aren't inherently different in how they render an image, I am simply saying that I, and many others, find the existing output different. My point is that if the rendering is a consequence of engineering choices, why not design a CMOS that produces the same output of a CCD; it would make a lot of customers very happy. I certainly didn't say anything about magic or mystical properties; the output is different, the question is why.

Tom

And that is the tough question is they are unknown in the fact we have no idea what manufacturers are doing. There are also things that change our behavior, like DR.

So, for example, it has been proposed that one way manufacturers improve sensitivity is that they make the color filter array weaker. That give more cross channel bleed which needs to be dealt with somewhere in the pipeline. If that is the case, and photographers love their sensitivity, then I don't see manufacturers going back to better, in terms of color reproduction, CFA. Kodak was known to be fanatical about color and so hang onto that 645D.

Then how are manufacturers building the de-Bayer profile? All cameras from other companies are different. The secret sauce.

Some of it might be the photographer's fault. Certainly DR has been improving and that makes subtle differences in the way a photographer may process--simply by processing for the shadows and highlights will change contrast and saturation. I know photographers love the idea of more DR, but be careful what you wish for.

As for observer bias, there is the phenomenon of perception where we have been using a particular camera for sometime and have learnt the how to process to our liking. A new camera comes along with a new sensor, we need to start the process all over again. Since we forgot about the learning curve of the old camera, the new camera seems to be a pain. Especially if there are expectations the sensor is "better"--more DR, less noise, etc--we find it disappointing. It isn't, we just not have developed the process to get the look we like. It took me awhile to figure out my RX-1. No my RX-1 and 645D are not the same, but there are too many factors masking the results for me to see a CCD/CMOS thing--different formats, Kodak/Pentax vs. Sony color and profiles.

If you look at all the variable in a photograph from the original scene, the optics, the DR/noise of the sensor, to exposure, to processing, and then display. And the skill of the photographer is most likely the largest influence here. The sensor architecture is pretty obscure. You, Ed, and I regularly post 645D images here. All consistently high in quality. Yet, our photographs show a distinct style of the photographer--I am sure no one would confuse them. And not just in subject matter, but the tonal and color qualities. Where is the CCD look?

I am a firm believer in the fact the 645D and my Phase p25+ makes great images. I also know how easy they can be to process. They make rich images. I rather think it is chip size and pixel area that is contributing to that than they are CCDs. I also think Kodak and Delsa and Pentax and Phase did great jobs with the CFA/profiles.