John Sheehy wrote in post #19223244
The question, I think, is "compared to what?". If you use 800/11 instead of 800/5.6, then you get more diffraction, more noise, and less background blur, AOTBE, just as if you took the same shot with an 800/5.6 lens, one at 5.6 and another at f/11 with the same shutter speed. If you use 800/11 instead of cropping from from 400/5.6, however, all the analog qualities are the same, the only differences being that the 800mm puts 4x as many pixels-on subject, which resolves the subject better. Yes, the diffraction blur, as measured in pixels, has twice the radius for f/11, but the subject is also twice as tall and wide, so the net effect of diffraction blur size relative to subject size is the same; you have the same analog subject-normalized diffraction. By sampling the subject with 4x as many pixels, though, the coarseness of the color filter array and the radius of the anti-alias filter halve in each dimension, relative to the size of the subject. What you get is pretty much the same as if you had used 400/5.6 on a a 20MP m43 camera, except that the RF800/11 seems to have nearly zero aberration, based on how impressed many people are with it. I don't know if the 400mm lenses typically used on m43 cameras are as low in aberration.
It's the same thing when you put TCs on the already "slow" 800/11, the diffraction gets larger, but so does the subject, for a net zero change in their relationship, ISO goes higher, but the subject is larger, making up for it, and the coarseness of the CFA and the radius of the AA filter both get even smaller and have less negative effect.
The question, I think, is "compared to what?". If you use 800/11 instead of 800/5.6, then you get more diffraction, more noise, and less background blur, AOTBE, just as if you took the same shot with an 800/5.6 lens, one at 5.6 and another at f/11 with the same shutter speed. If you use 800/11 instead of cropping from from 400/5.6, however, all the analog qualities are the same, the only differences being that the 800mm puts 4x as many pixels-on subject, which resolves the subject better. Yes, the diffraction blur, as measured in pixels, has twice the radius for f/11, but the subject is also twice as tall and wide, so the net effect of diffraction blur size relative to subject size is the same; you have the same analog subject-normalized diffraction. By sampling the subject with 4x as many pixels, though, the coarseness of the color filter array and the radius of the anti-alias filter halve in each dimension, relative to the size of the subject. What you get is pretty much the same as if you had used 400/5.6 on a a 20MP m43 camera, except that the RF800/11 seems to have nearly zero aberration, based on how impressed many people are with it. I don't know if the 400mm lenses typically used on m43 cameras are as low in aberration.
It's the same thing when you put TCs on the already "slow" 800/11, the diffraction gets larger, but so does the subject, for a net zero change in their relationship, ISO goes higher, but the subject is larger, making up for it, and the coarseness of the CFA and the radius of the AA filter both get even smaller and have less negative effect.
How does the ISO affect the diffraction?
John Sheehy wrote in post #19223244
There is no generic loss in subject quality with maintaining the same entrance pupil (71mm in this case) with longer and slower options, when compared to cropping and shooting from the same distance; what you have to watch out for is loss of AF ability, darker sensor dirt, and any added aberrations from the TC (which are probably quite small with Canon's latest TCs). High-contrast edges with lighting high in invisible light may blur more with TCs, too, as they are optimized for visible light.
If you also get farther from the subject with a TC, your entrance pupil stays the same size, but it looks smaller to the subject because you are farther, which means more subject-level diffraction and noise, and your background gets less blurred relative to subject size, because it is less out of focus, plus the entrance pupil also looks smaller to the background, even if not as much smaller as it is to the subject. Also, when you get farther from the subject, the minimum usable shutter speed to keep motion blur small relative to subject size also increases, increasing subject-level noise.
When you don't get farther away, but use the TC as an alternative to cropping, you can increase shutter speed if you want, but it isn't as necessary as it is with being farther away, because subject-normalized camera motion blur has nothing to do with focal length, but rather, with distance. Think of a laser pointer attached to a long lens and pointed at the center of the frame; how much it moves around on a subject has nothing to with focal length, and everything to do with distance. So, even if you don't up your shutter speed for an added TC, you won't get subject captures worse than without the TC, and on the fraction of frames where there is the least camera/lens motion, you will get subject capture that is more detailed.
There is no generic loss in subject quality with maintaining the same entrance pupil (71mm in this case) with longer and slower options, when compared to cropping and shooting from the same distance; what you have to watch out for is loss of AF ability, darker sensor dirt, and any added aberrations from the TC (which are probably quite small with Canon's latest TCs). High-contrast edges with lighting high in invisible light may blur more with TCs, too, as they are optimized for visible light.
If you also get farther from the subject with a TC, your entrance pupil stays the same size, but it looks smaller to the subject because you are farther, which means more subject-level diffraction and noise, and your background gets less blurred relative to subject size, because it is less out of focus, plus the entrance pupil also looks smaller to the background, even if not as much smaller as it is to the subject. Also, when you get farther from the subject, the minimum usable shutter speed to keep motion blur small relative to subject size also increases, increasing subject-level noise.
When you don't get farther away, but use the TC as an alternative to cropping, you can increase shutter speed if you want, but it isn't as necessary as it is with being farther away, because subject-normalized camera motion blur has nothing to do with focal length, but rather, with distance. Think of a laser pointer attached to a long lens and pointed at the center of the frame; how much it moves around on a subject has nothing to with focal length, and everything to do with distance. So, even if you don't up your shutter speed for an added TC, you won't get subject captures worse than without the TC, and on the fraction of frames where there is the least camera/lens motion, you will get subject capture that is more detailed.
Please explain the highlighted text, what invisible light do you mean? If the TC is optimized for visible light (as is the lens used for most of our cameras), "invisible light" shouldn't be a factor.
