80D Heavy AA filter?

Valvebounce

CR Pro
Apr 3, 2013
4,555
450
57
Isle of Wight
Hi Folks.
Not withstanding Hillsilly's informative post, didn't we go through all this with the 7D when it was new, no predecessor to compare to and everyone convinced that it was soft especially when compared to cameras with the same sensor that followed?
Weren't people saying it was a heavy AA filter due to the idea that it would be a more video orientated camera?
Do manufacturers (Canon) intentionally put a heavier AA filter than is needed for a particular sensor, or is it more likely that the AA filter is as strong as is believed (calculated) to be needed for a new sensor until proven otherwise in real world scenarios and then it gets fine tuned?

Cheers, Graham.
 
Upvote 0
StudentOfLight said:
I was not not saying that the higher res body would be worse than the lower res body, but when shooting at f/8 on 24MP APS-C you'd be shooting at settings which erode the potential benefits (in fine detail) that the higher number of smaller pixels would have given. The 80D at f/8 would be worse than the 80D at f/4, assuming you had sufficient dof in both cases. Basically, f/8 - f/32 would be sub-optimal camera settings for the 80D if you are interested in retaining fine detail.

I totally agree with this. I'll further add that at f/8, an 80D with 24 MP vs. a 7D with 18 MP, the 24 MP sensor, all else equal, will still yield more detail, however slight.
 
Upvote 0
Jul 28, 2015
3,368
570
StudentOfLight said:
I was not not saying that the higher res body would be worse than the lower res body, but when shooting at f/8 on 24MP APS-C you'd be shooting at settings which erode the potential benefits (in fine detail) that the higher number of smaller pixels would have given. The 80D at f/8 would be worse than the 80D at f/4, assuming you had sufficient dof in both cases. Basically, f/8 - f/32 would be sub-optimal camera settings for the 80D if you are interested in retaining fine detail.

You seem to change tack halfway through your comment. I am not sure if the first part of your comment:
I was not not saying that the higher res body would be worse than the lower res body, but when shooting at f/8 on 24MP APS-C you'd be shooting at settings which erode the potential benefits (in fine detail) that the higher number of smaller pixels would have given.
was comparing 18MP to 24MP.
How do you define 'potential benefits' as opposed to 'actual' benefits?
How do you compare 'potential benefits' with what you actually see? It all sounds like theorising mumbo jumbo.
As has been said above diffraction is the same on both bodies so if you view both at 100%, why would diffraction 'erode the potential benefits' of a higher density sensor?
The 18MP sensor has diffraction which will reduce performance, the 24MP sensor will have diffraction that reduces the performance. The 24MP still gives higher resolution, diffraction or nay.
What problems do you foresee?

Then in the second part you seem to be comparing apertures on the same sensor:
The 80D at f/8 would be worse than the 80D at f/4, assuming you had sufficient dof in both cases. Basically, f/8 - f/32 would be sub-optimal camera settings for the 80D if you are interested in retaining fine detail.
You are saying that on the 80D, diffraction will be worse at f8 than at f4? Of course it will. Just like it will be worse at f5.6 than f4. Or worse at f4 than f2.8.
The question is, is it noticeable. People hear 'diffraction' and start freaking out - what they really mean is 'diffraction limiting' and even then, people far more experienced than I (including landscape pros) have said they have never had problems with diffraction below f16. Irrespective of camera body.
 
Upvote 0
Jul 28, 2015
3,368
570
ajfotofilmagem said:
StudentOfLight said:
I was not not saying that the higher res body would be worse than the lower res body, but when shooting at f/8 on 24MP APS-C you'd be shooting at settings which erode the potential benefits (in fine detail) that the higher number of smaller pixels would have given. The 80D at f/8 would be worse than the 80D at f/4, assuming you had sufficient dof in both cases. Basically, f/8 - f/32 would be sub-optimal camera settings for the 80D if you are interested in retaining fine detail.
Diminishing returns. Only that.

I stated that NO desire APS-C cameras with more than 24 megapixel because the benefits are restricted to diaphragm openings larger than F5.6.

A hypothetical APS-C camera with more than 30 megapixel, lose the benefit of additional sharpness in F4 openings, and this would make it limited to the use of wide landscapes. It would not be a worse camera than 70d, but would diminishing returns.

So are you saying that if you put the same lens on a 15MP APS-C and on a 30MP APS-C and print to 12x20, you will see more diffraction on the 30 MP body?
 
Upvote 0
Let us present a matrix bottom line:
A) The more pixels - the better the resolution even beyond diffraction limits.
Remember that most sensors have a Bayer array, so you need to expand the theoretical per pixel diffraction limits proposed by text by at least SQRT(2) = 1.41 each direction = double MP. The more MP you can throw at a target, the higher resolution you can achieve. We will probably need to hit 40MP (APS-C) before the AA filter becomes useless. Unless Sigma's Foveon sensor is involved.
I have recently seen that B&W film still out-resolves all current sensors at f8 - including the 5DsR - by a long way.
B) Higher DR - Provided all photons are utilized for high QE, the more pixels - the greater the dynamic range as long as the random noise of photon capture is less than the read noise per pixel to maintain SNR. We are getting close to ISO-less capture (eps with Canon sensors now - just one step behind Sony - an impressive leap), so this will hold true for well over 100MP for an APS-C sized sensor. Just look at the 5Ds compared to other Canon sensors with the old tech.
Bottom line - a large number of pixels is not going to diminish image quality. Sub-sampling the airy disc of confusion is still a long way off - even at f8 with 24+ MP APS-C sized sensors. Higher MP sensors will keep giving you better res (yes diminishing returns, but still better) with a higher dynamic range!
 
Upvote 0
Mikehit said:
ajfotofilmagem said:
StudentOfLight said:
I was not not saying that the higher res body would be worse than the lower res body, but when shooting at f/8 on 24MP APS-C you'd be shooting at settings which erode the potential benefits (in fine detail) that the higher number of smaller pixels would have given. The 80D at f/8 would be worse than the 80D at f/4, assuming you had sufficient dof in both cases. Basically, f/8 - f/32 would be sub-optimal camera settings for the 80D if you are interested in retaining fine detail.
Diminishing returns. Only that.

I stated that NO desire APS-C cameras with more than 24 megapixel because the benefits are restricted to diaphragm openings larger than F5.6.

A hypothetical APS-C camera with more than 30 megapixel, lose the benefit of additional sharpness in F4 openings, and this would make it limited to the use of wide landscapes. It would not be a worse camera than 70d, but would diminishing returns.

So are you saying that if you put the same lens on a 15MP APS-C and on a 30MP APS-C and print to 12x20, you will see more diffraction on the 30 MP body?

Yes, if you have a 15 MP sensor and put on a 50mm lens at f/8, and then you have the same lens at f/8 on a 30 MP sensor, the image from the 30 MP sensor will show more absolute diffraction, in that the sensor was able to resolve more diffraction than the 15 MP sensor. But that's where things go awry in this thread. The total amount of detail gained from the 30 MP sensor will still be greater than the image from the 15 MP sensor, yes, even despite more diffraction. You will never ever do worse, all else equal, with a higher pixel density sensor. If you keep the same lens and the same aperture, relatively speaking there is always something to gain with a higher resolution sensor with regards to detail. The misconception here is that you can have a 40 MP sensor at f/8 performing worse than a 10 MP sensor at f/8. That will never happen.
 
Upvote 0
bdunbar79 said:
Mikehit said:
ajfotofilmagem said:
StudentOfLight said:
I was not not saying that the higher res body would be worse than the lower res body, but when shooting at f/8 on 24MP APS-C you'd be shooting at settings which erode the potential benefits (in fine detail) that the higher number of smaller pixels would have given. The 80D at f/8 would be worse than the 80D at f/4, assuming you had sufficient dof in both cases. Basically, f/8 - f/32 would be sub-optimal camera settings for the 80D if you are interested in retaining fine detail.
Diminishing returns. Only that.

I stated that NO desire APS-C cameras with more than 24 megapixel because the benefits are restricted to diaphragm openings larger than F5.6.

A hypothetical APS-C camera with more than 30 megapixel, lose the benefit of additional sharpness in F4 openings, and this would make it limited to the use of wide landscapes. It would not be a worse camera than 70d, but would diminishing returns.

So are you saying that if you put the same lens on a 15MP APS-C and on a 30MP APS-C and print to 12x20, you will see more diffraction on the 30 MP body?

Yes, if you have a 15 MP sensor and put on a 50mm lens at f/8, and then you have the same lens at f/8 on a 30 MP sensor, the image from the 30 MP sensor will show more absolute diffraction, in that the sensor was able to resolve more diffraction than the 15 MP sensor. But that's where things go awry in this thread. The total amount of detail gained from the 30 MP sensor will still be greater than the image from the 15 MP sensor, yes, even despite more diffraction. You will never ever do worse, all else equal, with a higher pixel density sensor. If you keep the same lens and the same aperture, relatively speaking there is always something to gain with a higher resolution sensor with regards to detail. The misconception here is that you can have a 40 MP sensor at f/8 performing worse than a 10 MP sensor at f/8. That will never happen.
Absolutely! ++++
 
Upvote 0
Jul 28, 2015
3,368
570
bdunbar79 said:
Yes, if you have a 15 MP sensor and put on a 50mm lens at f/8, and then you have the same lens at f/8 on a 30 MP sensor, the image from the 30 MP sensor will show more absolute diffraction, in that the sensor was able to resolve more diffraction than the 15 MP sensor. But that's where things go awry in this thread. The total amount of detail gained from the 30 MP sensor will still be greater than the image from the 15 MP sensor, yes, even despite more diffraction. You will never ever do worse, all else equal, with a higher pixel density sensor. If you keep the same lens and the same aperture, relatively speaking there is always something to gain with a higher resolution sensor with regards to detail. The misconception here is that you can have a 40 MP sensor at f/8 performing worse than a 10 MP sensor at f/8. That will never happen.

So how do you tell the difference between the lower pixels (18MP body) and higher diffraction (24MP body)? I am not aware of a characteristic that defines the two.
I am intrigued because diffraction is an optical effect of the lens arrangement. Using an analogy from my long-distant physics lessons, diffraction on a pinpoint source of light causes rings and those rings are a fixed size for any given circumstances. The number of pixels does not change that. Take an image of those rings and the 18MP will still see the same spread of rings, those rings will just be a bit more blurry than with the 24MP. The edge of a leaf in a landscape will be just as blurry in both sensors when it comes to diffraction. In fact on those 'diffraction rings' you could argue that the greater blurriness of the 18MP means the spread of the diffraction ring is more smeared with the 18MP adding to diffraction problems not lessening them.

But overall I think you can only talk about 'resolving diffraction' if you able to look at an image and say 'See that, there. That is a diffraction pattern as opposed to fuzziness due to limitation of number of pixels'. So how do you know the difference?
 
Upvote 0
Jan 29, 2011
10,675
6,121
cazza132 said:
Let us present a matrix bottom line:
A) The more pixels - the better the resolution even beyond diffraction limits.
Remember that most sensors have a Bayer array, so you need to expand the theoretical per pixel diffraction limits proposed by text by at least SQRT(2) = 1.41 each direction = double MP. The more MP you can throw at a target, the higher resolution you can achieve. We will probably need to hit 40MP (APS-C) before the AA filter becomes useless. Unless Sigma's Foveon sensor is involved.
I have recently seen that B&W film still out-resolves all current sensors at f8 - including the 5DsR - by a long way.
B) Higher DR - Provided all photons are utilized for high QE, the more pixels - the greater the dynamic range as long as the random noise of photon capture is less than the read noise per pixel to maintain SNR. We are getting close to ISO-less capture (eps with Canon sensors now - just one step behind Sony - an impressive leap), so this will hold true for well over 100MP for an APS-C sized sensor. Just look at the 5Ds compared to other Canon sensors with old tech.
Bottom line - a large number of pixels is not going to diminish image quality. Sub-sampling the airy disc of confusion is still a long way off - even at f8 with 24+ MP APS-C sized sensors. Higher MP sensors will keep giving you better res (yes diminishing returns, but still better) with a higher dynamic range!

Why? The Bayer array means colour is dithered, not resolution. The AA filter will reduce outright resolution, but in a reasonably predictable manner so is generally comparatively accurately counteracted. Hence the reason final output resolution difference between worked 5DS and 5DSR files is not that noticeable most of the time, OOC yes, but not after optimal processing.

As for B&W film, I have seen no evidence to support that resolution claim. I have seen overly sampled B&W scans that illustrate grain structure, but not more detail/resolution. If you can point me to examples I'd be interested to see them.

Here is a link to the last B&W film high resolution thread I was involved in.

http://www.canonrumors.com/forum/index.php?topic=27421.msg542751#msg542751
 
Upvote 0
Thank you all for sharing your knowledge on affects of diffraction on higher density sensors.

Granted...

But why in the examples I showed real life shots, from different sources ,is the 80D visibly softer? I am not imagining that softness nor theorizing it ..... It's there in plain sight for all to see... What is going on with them.... ?
 
Upvote 0
K-amps said:
But why in the examples I showed real life shots, from different sources ,is the 80D visibly softer? I am not imagining that softness nor theorizing it ..... It's there in plain sight for all to see... What is going on with them.... ?
I read several interesting theories, but the facts are:

24 pegapixel APS-C gives better sharpening results with F5.6 apertures or more open.
24 megapixel APS-C has its canceled sharpness advantage (largely) with F8 or more closed apertures.
24 megapixel APS-C without an AA filter is capable of higher sharpness, although with a higher incidence of moiré and other artifacts.

Given these facts, I see very little advantage in having an APS-C sensor with over 30 megapixel, while using a bayer array.
 
Upvote 0
ajfotofilmagem said:
K-amps said:
But why in the examples I showed real life shots, from different sources ,is the 80D visibly softer? I am not imagining that softness nor theorizing it ..... It's there in plain sight for all to see... What is going on with them.... ?
I read several interesting theories, but the facts are:

24 pegapixel APS-C gives better sharpening results with F5.6 apertures or more open.
24 megapixel APS-C has its canceled sharpness advantage (largely) with F8 or more closed apertures.
24 megapixel APS-C without an AA filter is capable of higher sharpness, although with a higher incidence of moiré and other artifacts.

Given these facts, I see very little advantage in having an APS-C sensor with over 30 megapixel, while using a bayer array.
You are highlighting my first point regarding the Bayer filter array:
'24 megapixel APS-C without an AA filter is capable of higher sharpness, although with a higher incidence of moiré and other artifacts.'
This means that we are far from out resolving the airy disc with a Bayer 24MP sensor!
 
Upvote 0
Jan 29, 2011
10,675
6,121
K-amps said:
Thank you all for sharing your knowledge on affects of diffraction on higher density sensors.

Granted...

But why in the examples I showed real life shots, from different sources ,is the 80D visibly softer? I am not imagining that softness nor theorizing it ..... It's there in plain sight for all to see... What is going on with them.... ?

Because you are comparing them at 100%, which means the 80D image, and any and all aberrations, are being magnified more.

It is the standard issue with comparisons, what are you actually comparing? Like for like (same sized view) or 100% view? Down sample the 80D to the 70D size and the difference in sharpness disappears, or print a set picture size, or look at them on screen at the same size.

Why do people not get that if you look at two different things, one magnified more, the one that is magnified less will look 'sharper', 'cleaner', 'have less noise', etc etc. Comparisons need to be normalised to be of any comparative value.
 
Upvote 0
Mikehit said:
bdunbar79 said:
Yes, if you have a 15 MP sensor and put on a 50mm lens at f/8, and then you have the same lens at f/8 on a 30 MP sensor, the image from the 30 MP sensor will show more absolute diffraction, in that the sensor was able to resolve more diffraction than the 15 MP sensor. But that's where things go awry in this thread. The total amount of detail gained from the 30 MP sensor will still be greater than the image from the 15 MP sensor, yes, even despite more diffraction. You will never ever do worse, all else equal, with a higher pixel density sensor. If you keep the same lens and the same aperture, relatively speaking there is always something to gain with a higher resolution sensor with regards to detail. The misconception here is that you can have a 40 MP sensor at f/8 performing worse than a 10 MP sensor at f/8. That will never happen.

So how do you tell the difference between the lower pixels (18MP body) and higher diffraction (24MP body)? I am not aware of a characteristic that defines the two.
I am intrigued because diffraction is an optical effect of the lens arrangement. Using an analogy from my long-distant physics lessons, diffraction on a pinpoint source of light causes rings and those rings are a fixed size for any given circumstances. The number of pixels does not change that. Take an image of those rings and the 18MP will still see the same spread of rings, those rings will just be a bit more blurry than with the 24MP. The edge of a leaf in a landscape will be just as blurry in both sensors when it comes to diffraction. In fact on those 'diffraction rings' you could argue that the greater blurriness of the 18MP means the spread of the diffraction ring is more smeared with the 18MP adding to diffraction problems not lessening them.

But overall I think you can only talk about 'resolving diffraction' if you able to look at an image and say 'See that, there. That is a diffraction pattern as opposed to fuzziness due to limitation of number of pixels'. So how do you know the difference?

I understand your question. Of course don't lose sharpness to diffraction by increasing pixel density, because I agree with you in that the SIZE of the diffraction effect is strictly determined by the lens used. The airy disks are the same with any camera sensor so long as the same lens and aperture are used. If I have 50 MP sensor and a 25 MP sensor, all else equal, the 50 MP sensor will be able to reveal the effect of diffraction sooner. I'm saying that and NOT that the effect is stronger in the higher pixel density sensor. The effect is the same it's just the higher pixel density sensor is able to reveal its effect sooner (at wider apertures).
 
Upvote 0
bdunbar79 said:
Mikehit said:
bdunbar79 said:
Yes, if you have a 15 MP sensor and put on a 50mm lens at f/8, and then you have the same lens at f/8 on a 30 MP sensor, the image from the 30 MP sensor will show more absolute diffraction, in that the sensor was able to resolve more diffraction than the 15 MP sensor. But that's where things go awry in this thread. The total amount of detail gained from the 30 MP sensor will still be greater than the image from the 15 MP sensor, yes, even despite more diffraction. You will never ever do worse, all else equal, with a higher pixel density sensor. If you keep the same lens and the same aperture, relatively speaking there is always something to gain with a higher resolution sensor with regards to detail. The misconception here is that you can have a 40 MP sensor at f/8 performing worse than a 10 MP sensor at f/8. That will never happen.

So how do you tell the difference between the lower pixels (18MP body) and higher diffraction (24MP body)? I am not aware of a characteristic that defines the two.
I am intrigued because diffraction is an optical effect of the lens arrangement. Using an analogy from my long-distant physics lessons, diffraction on a pinpoint source of light causes rings and those rings are a fixed size for any given circumstances. The number of pixels does not change that. Take an image of those rings and the 18MP will still see the same spread of rings, those rings will just be a bit more blurry than with the 24MP. The edge of a leaf in a landscape will be just as blurry in both sensors when it comes to diffraction. In fact on those 'diffraction rings' you could argue that the greater blurriness of the 18MP means the spread of the diffraction ring is more smeared with the 18MP adding to diffraction problems not lessening them.

But overall I think you can only talk about 'resolving diffraction' if you able to look at an image and say 'See that, there. That is a diffraction pattern as opposed to fuzziness due to limitation of number of pixels'. So how do you know the difference?

I understand your question. Of course don't lose sharpness to diffraction by increasing pixel density, because I agree with you in that the SIZE of the diffraction effect is strictly determined by the lens used. The airy disks are the same with any camera sensor so long as the same lens and aperture are used. If I have 50 MP sensor and a 25 MP sensor, all else equal, the 50 MP sensor will be able to reveal the effect of diffraction sooner. I'm saying that and NOT that the effect is stronger in the higher pixel density sensor. The effect is the same it's just the higher pixel density sensor is able to reveal its effect sooner (at wider apertures).
That is exactly right! There is no disadvantage in using higher MP sensors. F-it go 40MP. Weaker or no AA filters would be required :)
I can say one thing - when you get the focus right, the Zeiss 135mm f2.0 at f2.5 gives me major problems with star alignment with PT Gui with a FS 6D. It's almost like the AA filter is trying to help me, but stars are green with one sub, then blue or red the next. Even the 5Dsr would have the same problem - the Zeiss is so dangerously sharp. A FS converted 5Dsr could be my next cam - it seams to hold detail well at ISO3200 with not much long exposure dark noise (a bit like the 7DII). A bit off topic there - sorry.
 
Upvote 0
K-amps said:
Thank you all for sharing your knowledge on affects of diffraction on higher density sensors.

Granted...

But why in the examples I showed real life shots, from different sources ,is the 80D visibly softer? I am not imagining that softness nor theorizing it ..... It's there in plain sight for all to see... What is going on with them.... ?
Crap lens - check out the longitudinal chromatic aberration! Pink halos around the blacks :/ It means that red is not focusing with blue. This flaw even shows up with the 5DIII test results!
 
Upvote 0