gfiksel wrote:
Not sure what this test proves or disproves. Diffraction is a property of the lens, not the sensor. Printing or viewing, everything else being equal it's always advantageous to 'digitize' a diffraction-blurred image with a higher resolution sensor since it results in a higher overall lens+sensor resolution.
It proves that when you compare the 7D to the 5D2, stopping down to f/16 will put the 7D at a decided disadvantage in terms of sharpness, at least when viewed at 100%. Myself and perhaps a few others were not sure how much of a difference it might be. I think it's pretty clear that at f/16, it's pretty significant.
Yes, it's a property of the lens (the aperture to be more exact), but with the higher pixel density of the 7D, you can see the effects more readily as the "airy disks" expand beyond the dimensions of the individual pixels.
Well, all it shows that 7D is capable of resolving the Airy disks, while a lower resolution camera might not be able to do that.
It does not mean that the 7D is at a disadvantage. Just the way around, if you compare this image with an image taken with a lower resolution camera at equal output sizes, print, or screen, you will see that 7d reproduces more detail.
Again, the total resolution of a system is determined by a compounded resolutions of its subsystems, in this case 1/Rtot=1/Rlens+1/Rsensor. Increasing the sensor resolution always brings the total resolution up.
Tom_W wrote:
Quick diffraction test, 7D, ISO 400, 85/1.2 lens, tripod mounted, MLU, 2 second shutter timer. RAW converted in DPP, Standard picture style, Tungsten WB, Low NR, sharpening set at "3" (I forgot to turn it down, but the difference is a little more noticable here anyway). These are all 100% crops. I tested at f/5.6, f/8, f/11, and f/16:
chez wrote:
In this example you can even see the softening at f8. Does this mean the sweet spot for the 7D is below f8...more like 5.6?
No, it does not.
It's like saying that because old records have a narrow frequency bandwidth it is better to play them on a cheap stereo system with a bad amp and crappy speakers.
The truth is that no matter what the record quality is, it is always better to listen to it with a high quality receiver and broadband speakers.
chez wrote:
In this example you can even see the softening at f8. Does this mean the sweet spot for the 7D is below f8...more like 5.6?
I'm not quite able to answer that. I would consider that we're looking at an 18 mpx image at 100%. So, seeing a small difference here isn't likely to show up in a print unless it is quite large.
I took an f/4 image with the 7D also, but to me, it looks the same as the f/5.6 image so I didn't crop it and upload it. I could barely tell the difference between f/5.6 and f/8 to be honest.
I had read from someone that f/6.8 was the "diffraction limit" for this camera, but I don't buy the 'hard limit' theory. As you can see here, the effects gradually increase as you stop down - there is no sudden, objectionable line beyond which one should not cross.
I see no reason to avoid f/8, and precious little reason to avoid f/11 unless I'm cropping heavily or printing very large. And that's based on these relatively unscathed images - some judicious use of unsharp mask and contrast curve tweaks could work wonders as well.
gfiksel wrote:
No, it does not.
It's like saying that because old records have a narrow frequency bandwidth it is better to play them on a cheap stereo system with a bad amp and crappy speakers.
The truth is that no matter what the record quality is, it is always better to listen to it with a high quality receiver and broadband speakers.
I've got around 250 LP's in a plastic container in the basement, where they've been awaiting the starting of my project to record them onto CD. By the time I do that, I'll be so old that I won't be able to hear them anyway.
I agree with you here - I have to stress that I am showing 100% crops. We don't print that big. And even if we do, we don't sit 18 inches away from the photo and stare at one little spot - we take it in as a whole. The 100% is the electronic loupe and everybody has one!
If this 7D were being compared to the 40D or 50D, it's output would have been size-matched and the advantage of the additional pixels would be apparent. Comparing with the 5D2 is almost cheating - the results should be expected, especially in the realm of 100% viewing.
gfiksel wrote:
Well, all it shows that 7D is capable of resolving the Airy disks, while a lower resolution camera might not be able to do that.
It does not mean that the 7D is at a disadvantage. Just the way around, if you compare this image with an image taken with a lower resolution camera at equal output sizes, print, or screen, you will see that 7d reproduces more detail.
Again, the total resolution of a system is determined by a compounded resolutions of its subsystems, in this case 1/Rtot=1/Rlens+1/Rsensor. Increasing the sensor resolution always brings the total resolution up.
Agreed, to a point. Although, if I were printing 36 X 24 inch prints from the 5D2 and 7D, I would certainly prefer keeping the 7D at f/11 or larger. I wouldn't be as worried about the 5D2's aperture (though you can see that the f/16 shot definitely shows a bit of diffraction compared to larger apertures on that body).
When I say 'disadvantage', I am referring to the softness that is obvious in the 7D's f/16 shot compared to that of the 5D2 with it's much larger pixels (and technically less resolution per pixel, even though greater resolution per total image). I think that this concept isn't common knowledge, and considering at least 2 comparisons I've seen on the internet so far, it is serving to bias opinions against the 7D. I think that shedding a little light on the diffraction effect might remove that bias - people should be judging the camera on its legitimate merits and downfalls.
gfiksel wrote:
No, it does not.
It's like saying that because old records have a narrow frequency bandwidth it is better to play them on a cheap stereo system with a bad amp and crappy speakers.
The truth is that no matter what the record quality is, it is always better to listen to it with a high quality receiver and broadband speakers.
I don't quite get what you are saying but every camera has it's best aperature to shoot with. If the aperature does not matter ( IE: far away landscape vista ), then I want to use the aperature that gives the best image quality. I don't automatically put the aperature to f16 for landscapes...even with the 5D. If I don't need f16, I will use f8 as for the 5D, that is it's best aperature to shoot at. Looks like for the 7D, this aperature is around 5.6.
mfurman wrote:
The test may not prove anything but it shows that those 18 Mpixels, EOS 7D gives us are not as good as 21 Mpixels from 5D mkII.
I have to agree, at least in equally-framed subjects shot with the same lens and the same aperture.
But, at the risk of opening up some kind of can of worms, I think that there are ways of comparing these two cameras where the IQ of the 7D would most likely be superior. I think that we'll find that the 7D excels in situations where the focal length of available lenses becomes an issue. Birding. Sports. Wildlife. The occasional moon shot. Maybe astrophotography. I think that when you throw a 100-400 on the 7D as compared to the 5D2, you will find benefit in having the ability to land more pixels on your target. Not just a few more - a lot more. More than enough to compensate for the slight sharpness disadvantage that is apparent in these tests.
I haven't had enough time to compare the two bodies in these situations thoroughly, but I can assure you that my 800mm (400/2.8 + 2X TC) 7D moon pictures are somewhat superior to my 1120mm (400/2.8 + 2X + 1.4X TC's) 5D2 moon pictures. And I've captured quite a few respectable 600 mm shots with the 7D. Plus, it works very well with the 100-400 (which really surprised me).
My opinion is that the 7D is NOT a replacement for the 5D2. I think it serves as a great accompaniment to the 5D2 though.
Tom_W:
I think that there are ways of comparing these two cameras where the IQ of the 7D would most likely be superior. I think that we'll find that the 7D excels in situations where the focal length of available lenses becomes an issue.
There is no doubt that you are right. 7D is an excellent camera as far as focus and handling is concerned. It is definitely better than any 8 Mpixel camera and a 8 Mpixels crop from 5D mkII. What I have seen so far, does not convince me that its 18 Mpixel images are equivalent to 21 Mpixel images from 5D mkII, if the FOV is the same. Moreover I am convinced that if cropped to 10 Mpixels (to increase the reach by 1.35x "converter"), the image is going to be inferior to un-cropped image from 40D. In other words, we will not get as many more good pixels on a target, as the resolution of this sensor would indicate. By the way, I think that 100-400 zoom is not really that great and pixel density of 7D may exceed its capabilities.
chez wrote:
I don't quite get what you are saying but every camera has it's best aperature to shoot with. If the aperature does not matter ( IE: far away landscape vista ), then I want to use the aperature that gives the best image quality. I don't automatically put the aperature to f16 for landscapes...even with the 5D. If I don't need f16, I will use f8 as for the 5D, that is it's best aperature to shoot at. Looks like for the 7D, this aperature is around 5.6.
Please help me out here, if you need for a scene to be sharp from 3 feet to infinity what aperture would you use? I have read that you should use f/22 and focus at 3 feet and you will have from 3 feet onwards enough DOF to have the whole scene sharp... is there another way to achieve sharp landscapes? What aperture would you use and where would you focus?
mfurman wrote:
There is no doubt that you are right. 7D is an excellent camera as far as focus and handling is concerned. It is definitely better than any 8 Mpixel camera and a 8 Mpixels crop from 5D mkII. What I have seen so far, does not convince me that its 18 Mpixel images are equivalent to 21 Mpixel images from 5D mkII, if the FOV is the same. Moreover I am convinced that if cropped to 10 Mpixels (to increase the reach by 1.35x "converter"), the image is going to be inferior to un-cropped image from 40D. In other words, we will not get as many more good pixels on a target, as the resolution of this sensor would indicate. By the way, I think that 100-400 zoom is not really that great and pixel density of 7D may exceed its capabilities.
I agree - I posted a high-ISO test on POTN between my 5D, 5D2, and 7D. The 5D2 won. The 5D and 7D were close, though the 5D had a little more contrast and detail while the 7D had a little smoother noise. I am certain that I could have tweaked back and forth between those two. Meaning, the 18 mpx on the 7D were more comparable to the 12.8 mpx on the 5D (and possibly just shy of the mark). Anyway, I didn't compare at lower ISO settings since I was concerned primarily with noise.
What was apparent in my comparison is how well the 5D2 retains detail compared to both the 5D and 7D. I basically stopped using my 5D when I got the 5D2 last March. I sold the 5D last week, and will miss it. But we have to move on.
I've seen a few comparisons between the 7D and 40D (and 50D as well) that show that it is superior to both. There is no need to crop a 7D to 10 mpx (if that's what you're saying) as it shares the field of view of both of these cameras. It is superior in noise and detail, though at 100%, it does show diffraction more quickly. However, if downrezzed to match the pixel dimensions of its two APS-C brethren, it still outresolves both.
Many people don't think highly of the 100-400. It's not my sharpest lens, and is no match for the 300/2.8. However, it works very well on the 7D, much better than what you might expect. I was surprised, although I think I have a very good copy of this lens.
mfurman wrote:
Thank you very much for these images.
Apart for the detail difference, I find it surprising that there is so much more noise in 7D image. I see that both were taken at ISO 200.
Remember these are HDRs from a 3 shot bracket, and that the HDR process introduces a lot of noise into the final image. On the original 0EV shot there isn't almost any evident noise at 100% crop.
Bryan Carnathan likes the camera a lot ... and confirms the softness.
Let's see what the defenders of the faith are going to say this time .
He's a very experienced reviewer.
Thanks for the link - that's a very good test (It's been out a few days, I've seen it). Not much to defend - it blows the 50D away at equal focal lengths, primarily by virtue of it's megapixel advantage and lower noise.
Tom_W:
I agree - I posted a high-ISO test on POTN between my 5D, 5D2, and 7D.
Tom, could you please post a link to this comparison? Thank you.
The reason that I mentioned cropping from 18 to 10 Mpixels is that it may allow for "more reach" with a lens. In other words, the resolution of 7D would be used not to achieve a better print but to "zoom" for a tele-converter effect. In that case the quality of the pixels would determine the coefficient of this converter effect.
hvilorio wrote:
Please help me out here, if you need for a scene to be sharp from 3 feet to infinity what aperture would you use? I have read that you should use f/22 and focus at 3 feet and you will have from 3 feet onwards enough DOF to have the whole scene sharp... is there another way to achieve sharp landscapes? What aperture would you use and where would you focus?
Thanks in advance.
Depends on the focal length and the body. Don't forget, if you are framing an equivalent scene with an APS-C camera as opposed to a full frame camera, you will be using a focal length that's approximately 1.6X shorter (if you shoot from the same vantage point). For example, if you're using an 85 mm lens on a 5D2, you would achieve the same depth of field with a 50mm lens on a 7D while stopping down 1 1/3 to 1 2/3 stops less. In other words, the DOF of an 85 mm lens on full frame at f/16 will be about the same as the DOF of a 50 mm lens on a 1.6X crop camera at f/9. So, even though diffraction is visible earlier as you stop down, you don't need to stop down as far - again, if you're shooting from the same vantage point and using a lens that provides the same field of view (1.6X shorter).
As for the aperture you'd use, again, it depends on the format of the body. I'd certainly focus somewhere between 1/3 and 2/3 out depending on what part of the image you think is more important.
Don't forget, the comparisons I made were 100% crops. Small differences in sharpness at that level will most likely not be visible in ordinary 8 X 10 or 11 X 14 prints.
mfurman wrote:
Tom, could you please post a link to this comparison? Thank you.
The reason that I mentioned cropping from 18 to 10 Mpixels is that it may allow for "more reach" with a lens. In other words, the resolution of 7D would be used not to achieve a better print but to "zoom" for a tele-converter effect. In that case the quality of the pixels would determine the coefficient of this converter effect.
So if I understand, what you're saying is that you're cropping the 5D2 to 10 mpx to match the framing of the 1.6X crop of the 7D and its predecessors? I'm not sure about cropping the 7D though. What am I cropping to compare with? I'm missing something here. Or are you saying that I should crop the 7D to obtain additional "zoom" effect on top of that gained by its pixel density.
It's getting late, and I've been up since 0400 this morning. Brain's clogging up!
Here's the test thread. I re-did the test later in the thread to accomodate some of the concerns of other posters...
.