George.ML wrote:
The problem with the 15mpx on the 50D is actually not the sensor but the current Canon lens lineup.
If all Canon lenses (~60 of them) were sharp enough, there would be no problem with 15+ megapixels on 1.6x crop sensors (not counting the ISO/noise issues).
The thing is, out of the ~60 Canon lenses, maybe only 2-3 that do not cost $4,000+ can match the resolution of the 50D.
So, until Canon change their entire lens lineup, it is safe to say that the 50D outresolves most of Canon's lenses.
Somewhere between the 450D 3.7 MP/cm² pixel density and the 50D 4.5 MP/cm² pixel density, lies the absolute limit of resolution for most of Canon lenses and what's annoying is that the team of photography experts and engineers at Canon research facility somehow managed to overlook that.
Thank God we have a slew of experts on the internet to tell us what Canon staff missed.
astrolucida wrote:
You may not care about that in the detail level as being discussed here. But you are probably interested in the conclusions (when we get them). That is, whether you'll get more detail with 50D than with uprezzed images taken with 40D.
I'd say the difference between an upressed 40D and a 50D image is barely noticeable even for the eagle eyed viewer. After all the difference between them is only 864 pixels across and 576 pixels down. Try downsizing your image by that much pixel and upres it again. Or better yet, someone who owns a 50D and 40D can do the test and end this.
n0b0 wrote:
Somewhere between the 450D 3.7 MP/cm² pixel density and the 50D 4.5 MP/cm² pixel density, lies the absolute limit of resolution for most of Canon lenses
That's seems to be the case indeed.
It's a practical limit, though, not absolute.
... and what's annoying is that the team of photography experts and engineers at Canon research facility somehow managed to overlook that.
The engineers surely didn't.
But the marketing department is heavily betting that buyers will indeed overlook it - and will pay the premium over the 40D for the extra megapixels.
Thank God we have a slew of experts on the internet to tell us what Canon staff missed.
The Canon staff are surely not the only experts on the subject.
I was being sarcastic mate. George.ML wrote:
The engineers surely didn't.
But the marketing department is heavily betting that buyers will indeed overlook it - and will pay the premium over the 40D for the extra megapixels.
Ah yes, the marketing department told them to make a new body with more pixel density and not to worry if it outresolves most of their own lenses because they don't really want to sell those lenses.
George.ML wrote:
The Canon staff are surely not the only experts on the subject.
If the subject is what the camera bodies and lenses that they designed are capable of, I think they are.
n0b0 wrote:
Ah yes, the marketing department told them to make a new body with more pixel density and not to worry if it outresolves most of their own lenses
Canon would never even think about doing this - ever !!!
because they don't really want to sell those lenses.
Well, if you buy a camera that outresolves most lenses, you still need lenses, don't you?
bushwacker wrote:
You meant diffraction depends on 50d's sensor? so i were to use the same lens on 40D or 450D ... 40D might have it f/9 and 450D will have f11?
Precisely stated, diffraction depends on the optics - nothing else. However, as we sample the image with pixels, the important question is when we start seeing the effects of diffraction in the images. The smaller the pixel is, the sooner we get to the limit, which depends on the f ratio.
Yes, you are quite right that with different cameras the limit is at a different point. I've calculated the f ratio where the Airy disc equals four pixels. Airy disc is the smallest possible "dot" that theoretically perfect optics can produce. Four pixels, because of the RGBG matrix. For one pixel, the limit is half the value below.
40D has f8.7 as the limit, 450D f7.7, 30D f9.6, 1DIII f10.8 and 5D f12.2!
The new 1DsIII and 5D II both have f9.6, having equal pixel size to the 30D.
Remember, though, that no real optics is perfect. When you stop a lens down, it will become sharper. Until the diffraction kicks in - then no matter how good the lens is, the image will become blurry. In my experience one stop over the diffraction limit is only visible under perfect test conditions and only when flip-comparing two equal framings. Two stops and a discernible eye can see it in a single image at 100%. Three stops and then its already obvious at 100% (but not at 50%, so it depends on how big you print).
This all is good to know when you try to get out the maximum sharpness from your equipment and when your photographic goal allows a range of f stops.
astrolucida wrote:
40D has f8.7 as the limit, 450D f7.7, 30D f9.6, 1DIII f10.8 and 5D f12.2!
The new 1DsIII and 5D II both have f9.6, having equal pixel size to the 30D.
so let's say we load up a 40d and 450d with canon 17-40mm f4 L set apeture at f/10.0 --- you are saying I am already having diffraction probelms? this means we cannot use smaller apertures on these cameras.--- meanng it's useless?
n0b0 wrote:
Do you realise how inane that post of yours sounds?
Do you realise how rude this post of yours sounds?
George has a good point. Lenses will still sell. A camera resolving better than most lenses doesn't make your pictures any worse than a camera with fewer pixels, it just better shows the limits of the lens. In fact, high resolving cameras may well increase sales of Canon's more expensive L lenses as people try to get more out of their sensors.
Think about it this way: whenever you use your current camera, and select a lens and aperture, there is some mismatch between what your lens is resolving and what your sensor can record. It may even be that your lens throttles information in the corner, and your sensor throttles information in the centre. Your shutter speed, iso, and ability to focus accurately and hold the camera steady also throttle your image resolution, in many cases. It doesn't stop you getting great pictures.
The world doesn't end when diffraction takes its subtle hold. OTOH, a review of a camera's sensor performance ought to take care to show the maximum obtainable quality, as well as the average. That's where DPR seems to have failed, once again.
When I get a chance I will put my Zeiss 21 on a 50D and see if that will deliver a full 15 megapixels of detail. I expect it will. It certainly makes a 450D sing in a way that some other lenses don't.
brainiac wrote:
Do you realise how rude this post of yours sounds?
Did you even read his previous post before deciding to chime in? I suspect not.
He said and I quote, George.ML wrote:
The thing is, out of the ~60 Canon lenses, maybe only 2-3 that do not cost $4,000+ can match the resolution of the 50D.
I find that to be an absurd claim without any evidence to back it up. It also sounded more like parroting what others have said. Seems like Canon bashing is the cool thing to do at the moment. Like someone said in another thread "most consumers are drones, just ask anyone in marketing". Too bad for the company it's a double edge sword.
Now I don't have a 50D but I do have a 450D which is the closest to 50D in terms of pixel density and maximum image resolution and I have no problem with image sharpness taken with a cheap pentax lens at 100% crop as you can see here or with the MP-E macro lens here even with diffraction setting in at effective f/stop of 20.
I really don't think the addition of 480 pixels across and 320 pixels down to the 450D maximum image resolution somehow manage to outresolves most of Canon lenses.
brainiac wrote:
Do you realise how rude this post of yours sounds? n0b0 wrote:
Did you even read his previous post before deciding to chime in? I suspect not.
Yes, I did, and I didn't see anything in it that provoked a rude response.
He said and I quote,
I find that to be an absurd claim without any evidence to back it up.
I'm still not seeing a reason to be rude.
One reason for this difference in opinion might be that different photographers have different attitudes to sharpening. It seems to me that many lenses don't fully resolve the sensor at these high densities, at many apertures. Sometimes a bit of sharpening makes up for it, depending on how much micro-contrast is lacking. Some photographers like to shoot at f2.8, or f16, and some care about corners much more than others. Putting a Zeiss 21 on a 450D is quite a revelation. George may well be right or wrong, but his proposition isn't absurd. There is some pixel density at which the lens becomes the limiting factor. Some photographers will feel that limit before others.
I'll be honest with you, he annoys me, but let's keep that between him and me aye? and I promise you any further correspondence with him will happen in PM, ie. not involving anyone else here.
There is some pixel density at which the lens becomes the limiting factor. Some photographers will feel that limit before others.
Sure, I can understand that, if the difference in pixel density is huge, not when the difference between 450D and 50D is so small. Even if it is true, it's probably limited to some of the cheapest lenses instead of the majority of lenses as claimed.
It's just a pity that people listen to those pixel peepers but I suppose they also help Canon sell old camera stocks.
bushwacker wrote:
so let's say we load up a 40d and 450d with canon 17-40mm f4 L set apeture at f/10.0 --- you are saying I am already having diffraction probelms? this means we cannot use smaller apertures on these cameras.--- meanng it's useless?
Of course you can use it. Just that you get a (slightly) less sharp image than with the same lens at f8!
So, if you don't need the depth of field, you should use f8 instead of f10. That's why I want to find out the limit of diffraction for each camera. When I need the maximum sharpness (but not the maximum DoF), I use that aperture (or larger, if the lens is sharp still).
astrolucida wrote:
Of course you can use it. Just that you get a (slightly) less sharp image than with the same lens at f8!
So, if you don't need the depth of field, you should use f8 instead of f10. That's why I want to find out the limit of diffraction for each camera. When I need the maximum sharpness (but not the maximum DoF), I use that aperture (or larger, if the lens is sharp still).
The f- stop values that you posted for each camera sensors... are they from Canon Technical documentation or we just did a manual calculations?
I'd like to have one of these documentations so i can maximize lens-sensor image sharpness.. and also for each lens' diffraction and maximum resolution values.
thanks very much... this cleared a lot of my confusions.
oh by the way... since we are talking about this... you posted that the 20d\30d has the same pixel size 1DsIII and 5D II --- from this we can safely say that image sharpness from 20D\30D is much greater [ compared to new - higher mega-pixel EOS bodies ] because it has more usable lens f-stop without hitting diffraction.?
bushwacker wrote:
The f- stop values that you posted for each camera sensors... are they from Canon Technical documentation or we just did a manual calculations?
I'd like to have one of these documentations so i can maximize lens-sensor image sharpness.. and also for each lens' diffraction and maximum resolution values.
Diffraction depends only on the f-stop. However, where the maximum resolution for a specific lens is, can only be found out by actual measurements. Typically primes are at their best 2-3 stops from wide open until f8, although some supertelephotos max already at f5.6 (but are not bad at f8, either).
As for the diffraction vs pixel size, I have made my own Excel sheet, which for each camera first calculates the pixel size (sensor size divided by the number of pixels - both directions give the same result) and then the f stop where the size of the Airy disc equals twice the pixel size.
Here's the formula: f-stop=(X*1000)/(550*2.43932)*2
X is the pixel size in um, 1000 scales it to nm, 550 is the wavelength of yellow-green light in nm. 2.43932 scales it to f stop values and 2 makes it two pixels wide.
