I guess a lot of people define the 'diffraction limit' as the f-stop when the resolution falls X% below its peak, where X is a number that results in a clearly visible difference. And there is some merit in defining such an f-stop, it gives a range where your lens performs close to optimally.
Do you mean this web site?:
Funny you should mention that..
The problem is that its net effect is to misinform. When people quote this diffraction limit tripe, they refer to McHugh's site as evidence, and as a whole it is riddled with errors. This is actually one of the better pages, but careless words cost lives (or at least misinform)
The sensor was newer technology. Had the 10MP sensor been made a 14MP one, the improvements would've been there.
When viewed at 100%? OK, resize the 5MP image to 10MP and re-assess.
This isn't a popularity contest. It doesn't matter what the "big public" says. It really amounts to squat. A common ignorant notion, luckily, doesn't change any of the facts.
Moreover, most didn't agree that the G11 had better IQ. What DPR said in the G11 review:
I'm not understanding this. If you were talking about the lens in isolation, I can understand your point. I'm not understanding it when what is shown is limited by the sensor resolution, print size, and viewing distance.
Well, with this you define 'the effect of diffraction' something that is noticeable when viewed at maximum resolution. But coming back to Bob's graph:
Where in this graph would say is the diffraction limit? And what criteria would define this? As I said above, one could postulate this as when the resolution falls, eg, 30% below its peak. One technically more correct definition would the f-stop from which on resolution follows a straight line (and this f-stop is largely sensor-resolution-independent).
At f/5.6 or a little above, but that doesn't mean a particular sensor/print/viewing distance will show its effect that early.Where in this graph would say is the diffraction limit?
The whole sensor/print/viewing distance does not have any practical meaning.At f/5.6 or a little above, but that doesn't mean a particular sensor/print/viewing distance will show its effect that early.Where in this graph would say is the diffraction limit?
That is a totally synthetic and absurd way to look at this issue.
Steve, I'm not sure what you mean by 'work better'. It will allow your $100 lens to resolve more - even taking diffraction into account, because the loss of detail due to diffraction is still less than the loss of detail due to larger pixels.
.
When I want to print very large, I'll probably be downsampling to 24 or 16MP, and printing from there. When I want smaller prints, or even something very small like web display, I'll be downsampling to 12MP, perhaps smaller.
I'm excited to try my very best lenses, as well. But every lens in my collection just got better!
Only at 100%, not at display sizes.
Again, I suspect you're thinking in terms of 100 or 200% viewing...
Why are the slopes different (right side of the graph)?
Because the system is not diffraction limited even to f/16. That is the camera resolution is still playing a significant part in defining the system resolution. It's always like that. Here's another two cameras, same lens.
These graphs are not the 'lens in isolation', they are a system of lens and camera, and if you change the camera you change the resolution at all f-numbers.
The graphs say otherwise. The point is that the 'effects of diffraction' are not things that are invisible and suddenly become apparent when you pass some threshold. They are always there, every f-number, every viewing distance, every image size. Like all blur, its less visible for small images and at big distances. And note, when you say 'show the effects of diffraction' you are quoting almost directly for Sean McHugh, as if his website has been responsible for putting this idea in you head that diffraction suddenly pops out at you, and now you can't get rid of it, and it's stopping you seeing the rather simpler truth. That is the problem, falsehoods told with nice computer graphics and interactive widgets can be quite compelling, but they are still false.
As I said, "extreme terms". Exagerated to try and make a point.
My point, exactly.
If you can't explain it simply, you don't understand it well enough.
I get the impression that you are hung up on the technical definition. As I tired to say earlier, I am interested in the practical application, and that is what I think
I get the impression you don't belive him - you should try and would be surprised ...
It is not very generalized, and it indeed is backed up with plenty of tests and user experiences. Try for yourself, you would be surprised.I disagree with all of that, it's far too generalised and is not backed up by any amount of tests or user experiences.Sorry for the short response...I am unwell at the moment and don't really have the energy for a lengthy reply.
It isn't gobbledigook...basically you compared small sensor with big and blamed pixel size, when you should have blamed sensor size. In that example two things changed, the sensor size and the pixel size. Some how it's always the pixels fault....unfairly
If you made a 2 different sensors sizes using the exact same pixel in each...the bigger sensor would be better.
or another direction. For two same size sensors identical in every single way bar megapixel count...the higher megapixel count one will be better (IQ wise....not speed wise or small file size).
That is largely because they switched from one CCD technology to another, the one in G11 is newer and better (mainly on video which was teh main drive behind switching), but less evolved in terms of pixel size.
I am not aware of those two models, but from your description they both sound fairly old. You cannot just make single comparisons, and draw general conclusions from them. Overall, pixel sizes has shrunk and still we get better DR, better high iso performance and overall better image quality. Look at any series of models or sensor sizes over a period of ten years, and you will see the pattern. Look at all those big pixel 6-8 MP DSLR cameras we had some 6-8 years ago, and see if you can find any single one of them hos is not soundly beaten by any 14-16-18 MP DSLR camera of today, no matter how you look at the images - DR, noise, color, the works. And yet the pixels are smaller. You can make teh same comparisons with compacts - they have all improved in spite of those small pesky pixels having multimplied and gotten smaller ...
