Eating Crow ... continued

John Sheehy

Forum Pro
Messages
28,403
Solutions
8
Reaction score
8,818
Location
NY, USA
GeorgeML writes:
Member said:
It's naive to think that shrinking the megapixels has no effect on their quality - and how can an image be better overall if it's combined from pixels of lesser quality?
It's not naive at all. It is very realistic. That's how the real world works. You don't have to downsample, though - that's only done to make images a certain size on a coarse display.
Member said:
Don't get me wrong - I'm not saying that 3mp is all that we need. The question is, is technology ready to jump to so many megapixels.
The 7D gives the best IQ, noise-wise, of any APS-C sensor. With that fact in hand, only an irrational person would conclude that the technology isn't ready.
Member said:
Since the 5D was announced in 2005, Canon have increased the megapixels on 1.6x cameras from 8mp -> 18mp.
Member said:
At the same time, in 2009 they still do not have a 1.6x camera that produces better image quality than the 5D.
Some day, APS-C cameras may collect as many photons as the 5D, or close. The 5D is not an efficient photon collector. I don't recall anyone saying that APS-Cs would ever be better image-makers than FF with the same technology level.

For capturing subjects, with a given telephoto lens from the same distance, the 7D walks all over the 5D. The 50D did that. The 5D2 was the first FF to give better imaging in a 1.6x crop than a real APS-C, and only for noise (the 5D2 still lags in resolution). The 7D should give about the same noise as the 5D2 with a 1.6x crop at high ISOs, but have much more resolution.
Member said:
So, it's not a coincidence that some people totally dismiss the 1.6x crop factor.
There are plenty of good reasons for a crop, especially for people needing high resolution without the overhead of huge files that need to be cropped.
Member said:
Canon have not done anything to prove that 1.6x camera can have same/better image quality than FF cameras.
They shouldn't. Bigger sensors always collect more photons, with the same level of efficiency. The potential is always higher for then full image with a bigger sensor. Who are you arguing with?

--
John

 
John, I'd like to go one step further to say smaller sensors are format of the future , not full frame. Most everything can be addressed, yes even WA and DOF, in the small format in some ways. The toughest one being noise but who's going to say technology will not evolve to a point that even small sensor can satisfy most people's need. Few would still need a V8 is 6 cylinders can produce 300 or even 400+ hp.
It's naive to think that shrinking the megapixels has no effect on their quality - and how can an image be better overall if it's combined from pixels of lesser quality?
It's not naive at all. It is very realistic. That's how the real world works. You don't have to downsample, though - that's only done to make images a certain size on a coarse display.
Don't get me wrong - I'm not saying that 3mp is all that we need. The question is, is technology ready to jump to so many megapixels.
The 7D gives the best IQ, noise-wise, of any APS-C sensor. With that fact in hand, only an irrational person would conclude that the technology isn't ready.
Since the 5D was announced in 2005, Canon have increased the megapixels on 1.6x cameras from 8mp -> 18mp.
At the same time, in 2009 they still do not have a 1.6x camera that produces better image quality than the 5D.
Some day, APS-C cameras may collect as many photons as the 5D, or close. The 5D is not an efficient photon collector. I don't recall anyone saying that APS-Cs would ever be better image-makers than FF with the same technology level.

For capturing subjects, with a given telephoto lens from the same distance, the 7D walks all over the 5D. The 50D did that. The 5D2 was the first FF to give better imaging in a 1.6x crop than a real APS-C, and only for noise (the 5D2 still lags in resolution). The 7D should give about the same noise as the 5D2 with a 1.6x crop at high ISOs, but have much more resolution.
So, it's not a coincidence that some people totally dismiss the 1.6x crop factor.
There are plenty of good reasons for a crop, especially for people needing high resolution without the overhead of huge files that need to be cropped.
Canon have not done anything to prove that 1.6x camera can have same/better image quality than FF cameras.
They shouldn't. Bigger sensors always collect more photons, with the same level of efficiency. The potential is always higher for then full image with a bigger sensor. Who are you arguing with?

--
John

 
One word - NONSENSE! And in the face of even P&S makers like Fuji offering larger sensors and less noise.
 
I find it so hard to believe that some people insist on arguing that FF has less noise than crop sensors. Duh.... If you don't like crop sensors then DON'T BUY ONE. Coming on to a forum and trying to argue that the 7d should have been FF or that FF is better than crops only shows off your total ignorance. Well over 90% of the DSLR market is crop sensors. They are not going away in the future just because you happen to think that FF sensors are better. I shoot wildlife almost exclusively and I rarely shoot in low light. A FF sensor is a waste of money for me. But I don't feel any need to hop onto a forum and publicly blast FF sensors. Different tools for different jobs.
--
Tom
My portfolio:
http://picasaweb.google.com/SeilerBird/MyTop51Photos#
 
To be fair, he didn't really say that. He replied to the post which effectively said that crop sensors will replace FF, calling it nonsense.
 
And in the face of even P&S makers like Fuji offering larger sensors and less noise.
So? You're assuming that this tells some inherent incontrovertable truth about the state of the art now and forever.

While it might well be that Fuji are moving in that direction at the moment, it is extremely likely that they do so simply because - to deliver a target level of IQ using the technology they have available now - that's Fuji's only choice...
 
To be fair, he didn't really say that. He replied to the post which effectively said that crop sensors will replace FF, calling it nonsense.
carlk did not say crop sensors will replace FF. He said smaller sensors are the future. I don't think he's implying that crops will ever be better than FF unless development of FF does not move forward. I tend to agree that as crops get better, an even less percentage of people will feel the need to have a FF. This is not much different that what 35mm film did to medium format film, and what medium format did to large format. For all of the people that currently think that FF digital is the future, there will only be disappointment. Each new and improved crop model keeps chipping away at the FF's armor!

CR
 
Right. Crop will be the main format of the future (well it is already now) as opposed to some full frame proponents who think full frame sensor will replace crop in the future. Full frame will be relegated to the niche where MF used to occupy. It just carries too much burdens, not just cost of the sensor, and gives little in return for most people. There is no need to pay for that cost, again not just monetary cost, other than for people who really need that little extra performance, such as shooting under candle light or print larger than A2. Most people don’t.
To be fair, he didn't really say that. He replied to the post which effectively said that crop sensors will replace FF, calling it nonsense.
carlk did not say crop sensors will replace FF. He said smaller sensors are the future. I don't think he's implying that crops will ever be better than FF unless development of FF does not move forward. I tend to agree that as crops get better, an even less percentage of people will feel the need to have a FF. This is not much different that what 35mm film did to medium format film, and what medium format did to large format. For all of the people that currently think that FF digital is the future, there will only be disappointment. Each new and improved crop model keeps chipping away at the FF's armor!

CR
 
It’s not nonsense just fact. 7D already beats or matches ALL full frame sensors made 2~3 years ago in resolution and/or noise.
One word - NONSENSE! And in the face of even P&S makers like Fuji offering larger sensors and less noise.
 
Thank you so much John Sheehy for sharing with us your knowledge. I always look forward to your posts. Without you DPReview would not be worth visiting sometimes...
 
It’s not nonsense just fact. 7D already beats or matches ALL full frame sensors made 2~3 years ago in resolution and/or noise.
I very much think this remains to be seen.

It seems as if it might beat the 5D for max DR, but be a trace worse at higher ISO to maybe 1/3 stop worse (i.e. more or less the same, not better). But in many cases SNR at 18% is even more important and I'm still a little doubtful the 7D will be able to equal the 5D there. 5D is maybe 2/3 stop better or so than 50D and 5D2 maybe 1.25 better than 50D? Of course if the 7D did manage to come in 1/3 stop better than the 50D that would bring it overall pretty much in the 5D ballpark.
 
It’s not nonsense just fact. 7D already beats or matches ALL full frame sensors made 2~3 years ago in resolution and/or noise.
I very much think this remains to be seen.

It seems as if it might beat the 5D for max DR, but be a trace worse at higher ISO to maybe 1/3 stop worse (i.e. more or less the same, not better). But in many cases SNR at 18% is even more important and I'm still a little doubtful the 7D will be able to equal the 5D there. 5D is maybe 2/3 stop better or so than 50D and 5D2 maybe 1.25 better than 50D? Of course if the 7D did manage to come in 1/3 stop better than the 50D that would bring it overall pretty much in the 5D ballpark.
Noise is not monolithic. One camera, at the same ISO, can have more noise in the shadows than another, but less in the highlights.

That is the point I have been stressing, apparently in vain, as people continue to talk about one camera having more or less noise than another. A 7D will never outperform a FF sensor with similar quantum efficiency in every regard; the highlights will have less noise in the FF, which can be quite apparent at medium ISOs. In the deepest shadows of low ISOs, and in the shadows of medium ISO, and the full range of extremely high ISOs, an APS-C can outdo a FF. In fact, with a FF with as poor QE as the original 5D, and its mediocre read noise (by today's standards), the 7D will outperform it except in the purely shot-noise-limited ranges. And the 5D won't have a whole lot less shot noise, either, if the 7D has QE better than the 50D.

Now, talk about a crop from the original 5D the size of an APS-C sensor (relevant to telephoto shooters limited by their glass), the 7D will walk all over the original 5D in evey resect; every ISO, every range, by a good margin. Noise should be similar when compared to the 5D2 at high ISO, but 7D resolution will be much higher.

--
John

 
It’s not nonsense just fact. 7D already beats or matches ALL full frame sensors made 2~3 years ago in resolution and/or noise.
I very much think this remains to be seen.

It seems as if it might beat the 5D for max DR, but be a trace worse at higher ISO to maybe 1/3 stop worse (i.e. more or less the same, not better). But in many cases SNR at 18% is even more important and I'm still a little doubtful the 7D will be able to equal the 5D there. 5D is maybe 2/3 stop better or so than 50D and 5D2 maybe 1.25 better than 50D? Of course if the 7D did manage to come in 1/3 stop better than the 50D that would bring it overall pretty much in the 5D ballpark.
Noise is not monolithic. One camera, at the same ISO, can have more noise in the shadows than another, but less in the highlights.

That is the point I have been stressing, apparently in vain, as people continue to talk about one camera having more or less noise than another. A 7D will never outperform a FF sensor with similar quantum efficiency in every regard; the highlights will have less noise in the FF, which can be quite apparent at medium ISOs. In the deepest shadows of low ISOs, and in the shadows of medium ISO, and the full range of extremely high ISOs, an APS-C can outdo a FF. In fact, with a FF with as poor QE as the original 5D, and its mediocre read noise (by today's standards), the 7D will outperform it except in the purely shot-noise-limited ranges. And the 5D won't have a whole lot less shot noise, either, if the 7D has QE better than the 50D.

Now, talk about a crop from the original 5D the size of an APS-C sensor (relevant to telephoto shooters limited by their glass), the 7D will walk all over the original 5D in evey resect; every ISO, every range, by a good margin. Noise should be similar when compared to the 5D2 at high ISO, but 7D resolution will be much higher.

--
John

It's a big problem that noise is discussed in such an oversimplistic way. The way noise impacts on photographs is nowhere near as simple as many make out. I'm no expert on the technical aspects of noise but I've had enough experience processing photos from lots of digital cameras to know that noise isn't the simplistic one-dimensional issue that a lot of these arguments assume.
 
The issue which maybe you and others can maybe comment on is that to compare noise between camera, ones needs to understand ensure that the sensors and surrounding electronics work the same, and have the same specifications.

The 7D and 5DMkII sensors are both CMOS and although the both use the "DIGIC 4" processors but that is wher eit ends. The CMOS sensors are DIFFERENT and the 7D has TWO processors so the logic and the way and amount of light "no of photons" the accumulated light is converted to a charge, stored, “read out” moved and converted cannot be exactly the same, because the surrounding logic and circuits are different. Although I stand to be correct, but only by an informed Canon electronic engineer who has the circuit diagrams and component specs in front of him.

Thus comparing the noise characteristics is a bit silly, as they will be different resulting in different image noise characteristics, and to say which one is better or worse is again only possible at a “high” level when one compares the nosie by using ones subjective judgement or by the use noise measuring algorithms. Which again are themselves partially subjective in terms of the fact the algorithms will “favour” one “system” or over another. Why because thr CMOS sensors geometries etc are different.

Note it is the overall “system” that needs to be considered and this is not trivial. Not at all.
 
There are some boundaries, physical limits. One is the size of human body, the second one is the wavelength of visible light. Ergonomy and optics, in other words.

Ergonomy - making a DSLR body much smaller than xxD line does not make sense. Rebels are already just over the edge for some. And 5D shows that FF sensor fits well into xxD size body. No advantage for crop here.

Optics - the task would be to make an EF-S lens that would cater for approx. 20 MP crop sensor as well as a good EF lens does for approx. 20 MP FF sensor. To achieve the same narrow DOF, the EF-S lens must be faster (mitigating size and weight advantage) and you have to squeeze production tolerances tight to achieve the resolution - and that costs money!

Take 7D + 17-55/2.8 vs. 5DII + 24-105/4. The FF lens is both wider and longer, and delivers narrower DOF. Moreover, its mechanical built and sealing should be superior. Size and weight are almost equal, the crop lens costs less (not much though) and its only advantage is shorter exposure time in extreme conditions thanks to its lower f-number. All this assuming that both sensors' hi-ISO performance is similar as for noise and detail.

I expect production costs of both crop and FF DSLR sensors to decrease in the future (just as the large plasma and LCD TV's), unlike the costs of fine optics and mechanical parts of camera lenses and bodies. And the comparison of lenses shows that a very good FF lens does not have to be larger, heavier, not even more expensive than an EF-S lens matching its performance.

So while APS-C might be "the format of the future" in telephoto and macro (well, it already is...), for all other applications FF will be the future choice IMHO.
John, I'd like to go one step further to say smaller sensors are format of the future , not full frame. Most everything can be addressed, yes even WA and DOF, in the small format in some ways. The toughest one being noise but who's going to say technology will not evolve to a point that even small sensor can satisfy most people's need. Few would still need a V8 is 6 cylinders can produce 300 or even 400+ hp.
--
Marek
http://galerie.kolas.cz/
 
This issue of eating crow is silly. There are two issues: one is DPR saying that 50D's IQ was not good. Another is they saying that's due to higher pixel density. The second is unrelated to Canon and is a mistake on their part.

The fact that the 50D and the 5DII had some problems with pattern noise and shadow noise is not related to pixel density, obviously, since both have quite different values for that. I think Bob2n had it right, and I think you also wrote similarly: it's the electronics and algorithms used, not at same level as for the 1D series. The 7D is likely proof of that, it's showing better noise in all departments, and an abscence of patter noise which was a common issue with the APS-C Canons (and with 5DII), totally unrelated to pixel density.

So, the part they have to reconsider is the one relating pixel density and IQ, w/o checking technology involved.

The "anti-Canon bias" fanboi cry is not warranted, IMO.

--
Renato.
http://www.flickr.com/photos/rhlpedrosa/
OnExposure member
http://www.onexposure.net/

Good shooting and good luck
(after Ed Murrow)
 

Keyboard shortcuts

Back
Top