Just for Jrista: 2014 Market Data

[sarangiman]

OK, but you left out the improvement in high ISO performance and that has something to do with sensor design. That for me has been very real and very valuable. I'm more interested in that than in low ISO dynamic range and fixing extreme underexposure. Of course I'm in favor of improved dynamic range too. But in the meantime, I'm very happy with what they are making.

That's totally fair if that's what's important to you. Personally, I think the marginal gains in ISO performance are almost nothing next to the drastic differences in base ISO DR and the types of shots they afford me. Arguing about 1/3 EV ISO performance differences when there are 3 stop DR differences is a little funny to me, but of course it all depends on your application. But seriously, if it's acceptable for people to be happy about 1/3 EV gains in ISO performance, you'd think it'd be OK for some of us to make a big fuss over 3 EV DR differences. But that's just, like, my opinion, dude.

Just FYI - improving low ISO DR does not come at the cost of ISO performance. Canon increases in low ISO DR will likely come when they start using on-chip ADCs (on the imaging sensor itself). But that requires new fab. The idea is to digitize the signal as early on as possible, so as to minimize potential interference.

I'm sure it'll come - especially b/c Canon is interested in the motion picture industry... and there most DPs clamor for more DR. More DR than what the state-of-the-art already offers there (e.g. in the Arri Alexa), which is already much more than what Canon offers. Canon tried to increase their DR performance on the Cinema EOS somewhat by pixel-binning G channels, which seemed to me to be a rather convoluted way of overcoming their high read noise. But I'm guessing it's very cost-intensive for them to actually switch to on-chip ADCs. Or something.

Interestingly, once you drastically increase low ISO DR, you can also effectively increase high ISO DR if the photographer chooses to dial back the ISO setting manually and then selectively raise the exposure - while protecting highlights - in post-processing.

It's a bit too much to get into here, but I'm sure you'll hear more people talking about it soon, if not already, on more technical forums.

 

[sarangiman]

I can appreciate that, but that doesn't mean you are not compromising by using it at f/1.4. Knowing what those compromises is important.
Your example is poor, perhaps if you had started with a different lens in a different situation you could demonstrate your point.

I knew of that compromise. What is hard to predict ahead of time - you know me not being a computer and all - is exactly where I'd run into the noise floor.

And that's what's so nice about Exmor - you don't have to worry about that. You can worry about other things, like focus, or capturing the decisive moment.

My example is not poor, it's just irrelevant to you. I've posted examples of other use-cases, and gotten answers like 'well I don't shoot that high DR scene', or 'you could've just used a GND', or 'oh you already used a GND? well you could've HDR'd it then'...

My point here is that you can always have an answer as to how you could've done it differently.

Doesn't detract from the main point: this is one less thing I have to worry about now.

These same arguments are constantly recycled. The same sentiments were thrown around during the digital vs. film debate.

As for sharing the whole file - no point. I did years ago when I showed this same problem with fixing vignetting with the 24/1.4 on my then newly acquired 5D3. And it was the same thing. Some people got it and agreed it was unfortunate, others said I should've used a flash or just accepted the vignetting, etc. And all those comments still missed the main point - I wanted available light only for that shot, and I generally don't mind vignetting but for that particular shot I didn't care for it. But I didn't have the choice to take out the vignetting, b/c of the ugly banding that ensued.

There's really no point - I just saw jrista arguing thread after thread about how he'd like to simplify his workflow by not having to resort to HDR every single time. It's the same argument here - I'd like to not be so constrained by my system when there are better options out there.

I didn't consider Nikon to be a better option until enough factors swayed me. Like not designing the grip for elf hands, for example (that one's for you, jrista).

 

[Sporgon]

I can appreciate that, but that doesn't mean you are not compromising by using it at f/1.4. Knowing what those compromises is important.
Your example is poor, perhaps if you had started with a different lens in a different situation you could demonstrate your point.

I knew of that compromise. What is hard to predict ahead of time - you know me not being a computer and all - is exactly where I'd run into the noise floor.

And that's what's so nice about Exmor - you don't have to worry about that. You can worry about other things, like focus, or capturing the decisive moment.

My example is not poor, it's just irrelevant to you. I've posted examples of other use-cases, and gotten answers like 'well I don't shoot that high DR scene', or 'you could've just used a GND', or 'oh you already used a GND? well you could've HDR'd it then'...

My point here is that you can always have an answer as to how you could've done it differently.

Doesn't detract from the main point: this is one less thing I have to worry about now.

These same arguments are constantly recycled. The same sentiments were thrown around during the digital vs. film debate.

As for sharing the whole file - no point. I did years ago when I showed this same problem with fixing vignetting with the 24/1.4 on my then newly acquired 5D3. And it was the same thing. Some people got it and agreed it was unfortunate, others said I should've used a flash or just accepted the vignetting, etc. And all those comments still missed the main point - I wanted available light only for that shot, and I generally don't mind vignetting but for that particular shot I didn't care for it. But I didn't have the choice to take out the vignetting, b/c of the ugly banding that ensued.

There's really no point - I just saw jrista arguing thread after thread about how he'd like to simplify his workflow by not having to resort to HDR every single time. It's the same argument here - I'd like to not be so constrained by my system when there are better options out there.

I didn't consider Nikon to be a better option until enough factors swayed me. Like not designing the grip for elf hands, for example (that ones for you, jrista).

You can certainly talk the talk, but after all this typing and in put from others I don't see this can continue until you post the full, original image as many of us have requested. Mask the faces if necessary as suggested by PBD. You've made all sorts of allegations against the 5DIII not coping with the situation: let's see the full picture.

 

[zlatko]

@zlatko:

I understand your arguments, that other factors of the system as a whole are important, often more important. However, as everyone here so often states...Canon's system is already excellent. In many ways, it's second to none. Canon already has a stellar AF system, they now have a superb full-color meter that links into the AF system for full tilt tracking, their ergonomics and menu system are excellent.

There are only a few areas where Canon could really realize significant gains. Sensor design is one of them. Canon keeps improving the AF system...45pt to 61pt to 65pt....diminishing returns. Keeps "improving" video (DPAF is the biggest innovation there, overall video capabilities seem to have stagnated though.) Canon currently has one of, if not the, best flash system on the planet for DSLRs. They already have excellent ergonomics, button placement, menu systems. They already had an excellent 100k pixel RGB meter, how much more does a 150k pixel RGB meter really improve things?

Canon has even addressed many of the firmware concerns of many customers, such as having "proper" Auto ISO implementation (seems it's finally "done right"...although right is rather subjective here...and should finally quell complaints.) However most of the issues like that could have been addressed with firmware updates for many cameras many, many years ago, and probably shouldn't have been pressing feature concerns for new camera models.

OK, but you left out the improvement in high ISO performance and that has something to do with sensor design. That for me has been very real and very valuable. I'm more interested in that than in low ISO dynamic range and fixing extreme underexposure. Of course I'm in favor of improved dynamic range too. But in the meantime, I'm very happy with what they are making.

I use the term "sensor IQ" to be more generic than just meaning improvement to low ISO DR. Sensor IQ covers the whole thing, low ISO, high ISO, color fidelity, everything. I want it all. I don't just want two more stops at ISO 100. I also want two more stops at high ISO (as the Sony A7s proves it's possible to get two more stops of DR at ISO 51200 even, and great IQ at ISO settings even above that.)

Improvements at high ISO, for all brands, with the exception of the A7s, have been fairly marginal. For the most part, that's physics bound. There are things that can be done to improve it, for sure, but the improvement's Canon has made at high ISO have been relatively small. They could gain another two stops of DR and improved color fidelity at all ISO settings...and I'd love that. I could use every ounce of improved performance. If Ansel Adams was here, he would figure out a way to use it all as well...the only reason Ansel made the sacrifices he did was because of the limitations of the equipment he used. If anyone could show us all how to use more DR, at any ISO, it would have been Ansel.

So, anyway...I don't disagree at all. I think some headway has been made at high ISO by Canon, but they have been leapfrogged, and quite considerably, again by Sony even at high ISO. Better sensor IQ means better sensor IQ, everywhere. I could totally use 8-9 stops of DR at high really ISO (in post-sunset light, when deer come out in throngs, I find myself using ISO 12800 more and more, and with the 5D III, that's a mere 7 1/2 stops of DR, with lower color fidelity and tonal range. The A7s...? A whopping 9 2/3 stops of DR! That would just kick ass...) I'm being tugged at very strongly by Sony's A7 line. I'll be renting at least one of them, probably the A7r to start, but I figure I should give the A7s a try as well to see how it does for my wildlife photography.

Ansel would certainly show us how to use more DR. While some of his work shows amazing efforts at capturing wide dynamic range, I think some of his work also shows creative use of limited dynamic range. I gather he was about making good photos, not necessarily copying the DR of real life. So his work includes black rocks, trees & skies that aren't black in real life. Even hopelessly blowing highlights didn't cause him to toss out an otherwise good portrait.

The A7s looks great. If I were buying one of the A7 series, that would be the one for me.

 

[zlatko]

OK, but you left out the improvement in high ISO performance and that has something to do with sensor design. That for me has been very real and very valuable. I'm more interested in that than in low ISO dynamic range and fixing extreme underexposure. Of course I'm in favor of improved dynamic range too. But in the meantime, I'm very happy with what they are making.

That's totally fair if that's what's important to you. Personally, I think the marginal gains in ISO performance are almost nothing next to the drastic differences in base ISO DR and the types of shots they afford me. Arguing about 1/3 EV ISO performance differences when there are 3 stop DR differences is a little funny to me, but of course it all depends on your application. But seriously, if it's acceptable for people to be happy about 1/3 EV gains in ISO performance, you'd think it'd be OK for some of us to make a big fuss over 3 EV DR differences. But that's just, like, my opinion, dude.

Just FYI - improving low ISO DR does not come at the cost of ISO performance. Canon increases in low ISO DR will likely come when they start using on-chip ADCs (on the imaging sensor itself). But that requires new fab. The idea is to digitize the signal as early on as possible, so as to minimize potential interference.

I'm sure it'll come - especially b/c Canon is interested in the motion picture industry... and there most DPs clamor for more DR. More DR than what the state-of-the-art already offers there (e.g. in the Arri Alexa), which is already much more than what Canon offers. Canon tried to increase their DR performance on the Cinema EOS somewhat by pixel-binning G channels, which seemed to me to be a rather convoluted way of overcoming their high read noise. But I'm guessing it's very cost-intensive for them to actually switch to on-chip ADCs. Or something.

Interestingly, once you drastically increase low ISO DR, you can also effectively increase high ISO DR if the photographer chooses to dial back the ISO setting manually and then selectively raise the exposure - while protecting highlights - in post-processing.

It's a bit too much to get into here, but I'm sure you'll hear more people talking about it soon, if not already, on more technical forums.

Where do you get 1/3EV gains in high ISO performance? I don't know which cameras you're comparing or what time frame. In my experience, each generation of cameras has offered about one extra stop of high ISO usability. The current 1DX at ISO 102,400 looks about like the original 1D at ISO 3200. That's about 5 stops in 11 years. The 5D > 5D2 > 5D3 progressed about 3 stops in 6 years. That's not marginal to me.

On-chip ADCs sound great. In the meantime, I'm not missing them.

 

[sarangiman]

I'm talking about normalized Raw performance, which has seen very, very small gains. Save for at the highest ISOs (e.g. 25.6k and above), where lowering sensor-level (upstream) read noise actually affects image quality, since you're amplifying the raw signal off the sensor so much (b/c the signal is so, so small).

Otherwise, ISO performance is largely dictated by sensor size these days.

 

[Loswr]

You can certainly talk the talk, but after all this typing and in put from others I don't see this can continue until you post the full, original image as many of us have requested. Mask the faces if necessary as suggested by PBD. You've made all sorts of allegations against the 5DIII not coping with the situation: let's see the full picture.

What, and risk having his claims invalidated? Why would he do that? :

 

[Tugela]

You asked for info regarding the latest interchangeable lens camera market. Thom Hogan has done the job for you:

http://www.dslrbodies.com/newsviews/interchangeable-lens-market.html

"So I went back to Canon and Nikon’s recent published numbers and the CIPA shipment numbers for 2014 so far and plugged them into a spreadsheet with Sony’s claims and an assumption about Sony’s DSLR sales (which they didn’t disclose). What I came up with was a range of values that went like this:

Canon — 38 to 40%
Nikon — 33 to 35%
Sony — 12 to 14%

...But, overall, it looks to me that we’re still locked into the same basic pattern for unit volume in interchangeable lens cameras that we’ve had since the early 90’s: Canon leading, Nikon following, and Sony (originally Minolta) trailing. At the moment, there’s nothing appearing from any of the three that seems to be going to alter that any time soon."

What they overlook is the fact that interchangeable lens cameras are part of a gear commitment, which means that once someone has a certain level of investment in gear for a particular body, they are less inclined to upgrade to a different manufacturer.

If you want to know where the sales are heading, look at the figures for young buyers, or those who are buying their first camera. Those are the customers who are going to be locked into the manufacturer systems they bought into when they expand their gear set. My guess is that the profile of sales in that demographic is quite different to overall sales.

Bluntly put, the impact of innovation in cameras today are going to show their effect on sales in 10 or 15 years from now. Canon may sell more now, but that is a result of the cameras they produced a decade ago, not the result of what they are producing now.

Things which are important to youngers buyers today, such as a camera's ability to integrate with other electronic systems buyers have (cell phones for example) are going to be the features which will have the greatest long term impact, not stuff like DR or pixel count. Samsung potentially have a great opportunity here, since they can leverage their cameras off their cell phone expertise and goodwill.

 

[Sporgon]

I'm talking about normalized Raw performance, which has seen very, very small gains. Save for at the highest ISOs (e.g. 25.6k and above), where lowering sensor-level (upstream) read noise actually affects image quality, since you're amplifying the raw signal off the sensor so much (b/c the signal is so, so small).

Otherwise, ISO performance is largely dictated by sensor size these days.

If you are referring to low ISO I wouldn't agree. The 5DII had substantially more latitude that's the original 5D at the highlight end, even if the actually DR wasn't greatly different. Likewise the mkIII has substantially more latitude than the II, and the 6D a tad more again. Tonal graduation has also improved, as well as highlight headroom in terms of clipping to white, as well as at the other end in graduation to black.

OK so this might only be noticeable to the discerning user.

 

[AmselAdans]

aaaand another thread sucked into the black hole named "DR issues".

Oh boy...

 

[takesome1]

I can appreciate that, but that doesn't mean you are not compromising by using it at f/1.4. Knowing what those compromises is important.
Your example is poor, perhaps if you had started with a different lens in a different situation you could demonstrate your point.

I knew of that compromise. What is hard to predict ahead of time - you know me not being a computer and all - is exactly where I'd run into the noise floor.

And that's what's so nice about Exmor - you don't have to worry about that. You can worry about other things, like focus, or capturing the decisive moment.

My example is not poor, it's just irrelevant to you. I've posted examples of other use-cases, and gotten answers like 'well I don't shoot that high DR scene', or 'you could've just used a GND', or 'oh you already used a GND? well you could've HDR'd it then'...

My point here is that you can always have an answer as to how you could've done it differently.

Doesn't detract from the main point: this is one less thing I have to worry about now.

These same arguments are constantly recycled. The same sentiments were thrown around during the digital vs. film debate.

As for sharing the whole file - no point. I did years ago when I showed this same problem with fixing vignetting with the 24/1.4 on my then newly acquired 5D3. And it was the same thing. Some people got it and agreed it was unfortunate, others said I should've used a flash or just accepted the vignetting, etc. And all those comments still missed the main point - I wanted available light only for that shot, and I generally don't mind vignetting but for that particular shot I didn't care for it. But I didn't have the choice to take out the vignetting, b/c of the ugly banding that ensued.

There's really no point - I just saw jrista arguing thread after thread about how he'd like to simplify his workflow by not having to resort to HDR every single time. It's the same argument here - I'd like to not be so constrained by my system when there are better options out there.

I didn't consider Nikon to be a better option until enough factors swayed me. Like not designing the grip for elf hands, for example (that one's for you, jrista).

All equipment has limits, if you push it past those limits then it will fail.
This is true of other bodies and sensors as well.
No matter which body or manufacture you are using you should always take the DR in to account, none have enough range.
The best claim you could make here is that you do not have to discard as many shots with the other camera due to your poor craftsmanship.

But you give a bad example, it is bad because you pushed the lens and body beyond its limits. It was the combination of the body and lens not just the sensor.

A good example is a comparison of two identical shots taken with the same exact lens in the exact same situation.
Posts an example like this and we have something to discuss other than your skill level with a shot that you failed.

 

[Loswr]

What they overlook is the fact that interchangeable lens cameras are part of a gear commitment, which means that once someone has a certain level of investment in gear for a particular body, they are less inclined to upgrade to a different manufacturer.

If you want to know where the sales are heading, look at the figures for young buyers, or those who are buying their first camera. Those are the customers who are going to be locked into the manufacturer systems they bought into when they expand their gear set. My guess is that the profile of sales in that demographic is quite different to overall sales.

Bluntly put, the impact of innovation in cameras today are going to show their effect on sales in 10 or 15 years from now. Canon may sell more now, but that is a result of the cameras they produced a decade ago, not the result of what they are producing now.

What you overlook is that the majority of dSLR buyers buy the camera with 1-2 kit lenses, maybe a cheap 50/1.8, and they're done. The people who are 'locked in' and who 'expand their gear set' are a relatively small minority. That's evident from Canon's milestones of number of bodies vs. lenses produced.

Bluntly put, most of those young buyers and those buying their first camera will not buy another camera, nor any more lenses, for several years if at all. If/when they do, there's a good chance it'll be the same brand, not because of 'lock in' due to lenses/flashes/etc., but due to familiarity.

 

[AmselAdans]

What they overlook is the fact that interchangeable lens cameras are part of a gear commitment, which means that once someone has a certain level of investment in gear for a particular body, they are less inclined to upgrade to a different manufacturer.

If you want to know where the sales are heading, look at the figures for young buyers, or those who are buying their first camera. Those are the customers who are going to be locked into the manufacturer systems they bought into when they expand their gear set. My guess is that the profile of sales in that demographic is quite different to overall sales.

Bluntly put, the impact of innovation in cameras today are going to show their effect on sales in 10 or 15 years from now. Canon may sell more now, but that is a result of the cameras they produced a decade ago, not the result of what they are producing now.

What you overlook is that the majority of dSLR buyers buy the camera with 1-2 kit lenses, maybe a cheap 50/1.8, and they're done. The people who are 'locked in' and who 'expand their gear set' are a relatively small minority. That's evident from Canon's milestones of number of bodies vs. lenses produced.

Bluntly put, most of those young buyers and those buying their first camera will not buy another camera, nor any more lenses, for several years if at all. If/when they do, there's a good chance it'll be the same brand, not because of 'lock in' due to lenses/flashes/etc., but due to familiarity.

Reminds me of a colleague, who had a Rebel with its kit lens. He was horrified by the idea of putting another lens onto the camera because of all the things that might go wrong (dust on the sensor, dropping either the lens or the camera or both, deforming the lens mount, accidentally evoking the dark lord Cthulu, ...).
Indeed a very good marketing that achieved to sell interchangeable lens systems to users who never intend to actually change them. (they hired the guys who sold refrigerators to the eskimos, I suppose)

 

[J.R.]

All equipment has limits, if you push it past those limits then it will fail.
This is true of other bodies and sensors as well.
No matter which body or manufacture you are using you should always take the DR in to account, none have enough range.
The best claim you could make here is that you do not have to discard as many shots with the other camera due to your poor craftsmanship.

But you give a bad example, it is bad because you pushed the lens and body beyond its limits. It was the combination of the body and lens not just the sensor.

A good example is a comparison of two identical shots taken with the same exact lens in the exact same situation.
Posts an example like this and we have something to discuss other than your skill level with a shot that you failed.

This is all too true.

On an aside, why would some continue to shoot with Canon gear if they hate it so much, if not hate it, find it too limiting. If they have switched to Nikon, you've obviously moved on, why complain about it on a Canon site?

 

[zlatko]

I'm talking about normalized Raw performance, which has seen very, very small gains. Save for at the highest ISOs (e.g. 25.6k and above), where lowering sensor-level (upstream) read noise actually affects image quality, since you're amplifying the raw signal off the sensor so much (b/c the signal is so, so small).

Otherwise, ISO performance is largely dictated by sensor size these days.

OK, you have your own definition of improvement in high ISO performance, not one that matches practical experience with these cameras. And so you conclude there's only been a 1/3 stop improvement between some unspecified cameras over some unspecified number of years, which you mock as "marginal".

You've set up a personal rule that you know Canon will fail. Your rule is that in order to be a "good sensor", a sensor has to be able to withstand extreme pushing of underexposure without showing any noise even in the darkest parts of an image. And so you posted a crop of a shot that failed to meet that test, not a crop from the shot that you actually used.

In the same way, each photographer can set up their own personal rules to make a sensor or system fail. If I set up a personal rule that a "good sensor" has to produce color exactly like Canon's, then Nikon, Sony, etc. would fail every time. If I set up a personal rule that a "good system" has to match Canon's ergonomics, lenses and flashes, then Nikon, Sony, etc. would fail every time. Likewise, if I set up a personal rule that a "good system" has to be full-frame or medium format, then Olympus and Panasonic would fail every time.

On-chip ADCs sound great, but there are photographers everywhere and at every level who don't seem to find the absence of on-chip ADCs limiting. Salgado produced his magnificent Genesis landscape book partly using Canon cameras -- despite the lack of on-chip ADCs. Photographers are producing award-winning photojournalism, advertising and editorial work with Canon despite the lack of on-chip ADCs. Somehow they are managing not to produce crummy noisy photos at ISO 100 or any ISO. Knowing all about on-chip ADCs, sensor fab, read noise, amplification, etc., is important if you're an engineer, but I'm not sure it helps one to do photography.

 

[jrista]

Where do you get 1/3EV gains in high ISO performance? I don't know which cameras you're comparing or what time frame. In my experience, each generation of cameras has offered about one extra stop of high ISO usability. The current 1DX at ISO 102,400 looks about like the original 1D at ISO 3200. That's about 5 stops in 11 years. The 5D > 5D2 > 5D3 progressed about 3 stops in 6 years. That's not marginal to me.

On-chip ADCs sound great. In the meantime, I'm not missing them.

From an actual noise performance standpoint, successive generations of cameras from most manufacturers have not actually realized a true 1-stop improvement. What we have seen is a one-stop increase in the highest user selectable ISO settings, but the performance of that top setting (or even lower settings) has not always improved, from a noise standpoint, by one stop every generation.

Measurable noise from generation after generation of many DSLR cameras from many manufacturers have improved marginally for the most part. The 1D X was one of very few cameras that actually realized a true one-stop advantage over it's predecessor. The 1D IV got about 9.5 stops of DR at ISO 3200, and the 1D X gets about 9.7 stops at ISO 6400. Read noise levels did not change much, but saturation point did as the FWC increased. The 5D III realized no benefit in high ISO DR over the 1D IV or 5D III...less than 1/3rd of a stop at best, and no difference in read noise.

That has been the case for a while, and is generally expected barring some kind of significant technological improvement. High ISO is ultimately physics bound, and as Canon's sensor tech hasn't changed for a decade or so, one shouldn't expect a real difference in high ISO performance. The A7s is a rather intriguing camera in that it realized more significant gains at even higher ISO settings than the 1D X than the 1D X did itself. The 1D X brought native ISO 51200 to the table, at 6.6 stops of DR...the A7s improved that to 8.8 stops of DR. All while concurrently having quite a bit less low ISO read noise (~21e- vs. ~38e-) and more low ISO DR. I don't suspect another major improvement in high ISO performance will occur until another radical technological improvement is made...such as multi-bucket reads or layered photodiodes or some kind of deep charge photodiode capable of holding more electrons than a normal photodiode is introduced. WHEN such technology is introduced, I'd expect high ISO performance to take off again.

In general, though, for a given technology, all we have really seen over the years is an increase in the highest selectable ISO settings...not an actual concurrent reduction in noise that would make that higher stop of ISO perform as well as a lower stop of a prior generation. Only a couple cameras have actually achieved that lately...the 1D X and A7s, and maybe the D600/700/800 series (although they may have simply started out with around an extra stop relative to Canon sensors at ISO 3200/6400 thanks to using Exmor.)

 

[zlatko]

Where do you get 1/3EV gains in high ISO performance? I don't know which cameras you're comparing or what time frame. In my experience, each generation of cameras has offered about one extra stop of high ISO usability. The current 1DX at ISO 102,400 looks about like the original 1D at ISO 3200. That's about 5 stops in 11 years. The 5D > 5D2 > 5D3 progressed about 3 stops in 6 years. That's not marginal to me.

On-chip ADCs sound great. In the meantime, I'm not missing them.

From an actual noise performance standpoint, successive generations of cameras from most manufacturers have not actually realized a true 1-stop improvement. What we have seen is a one-stop increase in the highest user selectable ISO settings, but the performance of that top setting (or even lower settings) has not always improved, from a noise standpoint, by one stop every generation.

Measurable noise from generation after generation of many DSLR cameras from many manufacturers have improved marginally for the most part. The 1D X was one of very few cameras that actually realized a true one-stop advantage over it's predecessor. The 1D IV got about 9.5 stops of DR at ISO 3200, and the 1D X gets about 9.7 stops at ISO 6400. Read noise levels did not change much, but saturation point did as the FWC increased. The 5D III realized no benefit in high ISO DR over the 1D IV or 5D III...less than 1/3rd of a stop at best, and no difference in read noise.

That has been the case for a while, and is generally expected barring some kind of significant technological improvement. High ISO is ultimately physics bound, and as Canon's sensor tech hasn't changed for a decade or so, one shouldn't expect a real difference in high ISO performance. The A7s is a rather intriguing camera in that it realized more significant gains at even higher ISO settings than the 1D X than the 1D X did itself. The 1D X brought native ISO 51200 to the table, at 6.6 stops of DR...the A7s improved that to 8.8 stops of DR. All while concurrently having quite a bit less low ISO read noise (~21e- vs. ~38e-) and more low ISO DR. I don't suspect another major improvement in high ISO performance will occur until another radical technological improvement is made...such as multi-bucket reads or layered photodiodes or some kind of deep charge photodiode capable of holding more electrons than a normal photodiode is introduced. WHEN such technology is introduced, I'd expect high ISO performance to take off again.

In general, though, for a given technology, all we have really seen over the years is an increase in the highest selectable ISO settings...not an actual concurrent reduction in noise that would make that higher stop of ISO perform as well as a lower stop of a prior generation. Only a couple cameras have actually achieved that lately...the 1D X and A7s, and maybe the D600/700/800 series (although they may have simply started out with around an extra stop relative to Canon sensors at ISO 3200/6400 thanks to using Exmor.)

That doesn't match my experience with actual photos. It's not just what is "user-selectable" but what photos I can use with those settings. When talking about actual usable photos, I've seen approximately one stop gain with each generation of camera. The 5D2 gave me one stop over the 5D. The 5D3 gave me one stop over the 5D2. So that's two stops gain in going from the 5D to the 5D3. The 6D gives me just a little more than the 5D3.

I should probably say "each generation of sensor" instead of "each generation of camera" as some cameras have come out with the same generation of sensor. So the 1D series has gone through 5 generations of sensor, although actually more than 5 cameras if you count every variation.

 

[privatebydesign]

That doesn't match my experience with actual photos. It's not just what is "user-selectable" but what photos I can use with those settings. When talking about actual usable photos, I've seen approximately one stop gain with each generation of camera. The 5D2 gave me one stop over the 5D. The 5D3 gave me one stop over the 5D2. So that's two stops gain in going from the 5D to the 5D3. The 6D gives me just a little more than the 5D3.

Well that is where you are going wrong zlatko, we don't allow actual empirical results to affect our opinions here, we have sensor readings and maths to tell us better what we should be able to see. /sarcasm

 

[Tugela]

What they overlook is the fact that interchangeable lens cameras are part of a gear commitment, which means that once someone has a certain level of investment in gear for a particular body, they are less inclined to upgrade to a different manufacturer.

If you want to know where the sales are heading, look at the figures for young buyers, or those who are buying their first camera. Those are the customers who are going to be locked into the manufacturer systems they bought into when they expand their gear set. My guess is that the profile of sales in that demographic is quite different to overall sales.

Bluntly put, the impact of innovation in cameras today are going to show their effect on sales in 10 or 15 years from now. Canon may sell more now, but that is a result of the cameras they produced a decade ago, not the result of what they are producing now.

What you overlook is that the majority of dSLR buyers buy the camera with 1-2 kit lenses, maybe a cheap 50/1.8, and they're done. The people who are 'locked in' and who 'expand their gear set' are a relatively small minority. That's evident from Canon's milestones of number of bodies vs. lenses produced.

Bluntly put, most of those young buyers and those buying their first camera will not buy another camera, nor any more lenses, for several years if at all. If/when they do, there's a good chance it'll be the same brand, not because of 'lock in' due to lenses/flashes/etc., but due to familiarity.

Incorrect, they will buy a body for the lenses and flash units they already have (even if it is only a few).

People who buy self-contained cameras have little brand loyalty since it is the abilities of the specific camera that are important, not stuff they already have, and that market is dominated by companies such as Sony - Canon is a Sad Sack in that market. That is what gives a more realistic indication of where the future is heading for photography in general. Right now Canon are living off past glories, but unless they get off their backsides all their glory will be in the past.

 

[Loswr]

That doesn't match my experience with actual photos. It's not just what is "user-selectable" but what photos I can use with those settings. When talking about actual usable photos, I've seen approximately one stop gain with each generation of camera. The 5D2 gave me one stop over the 5D. The 5D3 gave me one stop over the 5D2. So that's two stops gain in going from the 5D to the 5D3. The 6D gives me just a little more than the 5D3.

Well that is where you are going wrong zlatko, we don't allow actual empirical results to affect our opinions here, we have sensor readings and maths to tell us better what we should be able to see. /sarcasm

Shots with all of those cameras taken with the lens cap on then pushed 5 stops in post aren't significantly different, although the 6D might show a very marginal improvement. DxO Scores have not gone up much at all. Any perceived improvement in actual pictures is subjective and irrelevant.

: : :