Pixel Density and Reach--Any Definitive Tests?

1Balance said:

Maybe I've got it wrong (as is often the case), however I think you miss the core of what Alan's test shows. Identical (cropped) composition from the same distance with the same lens, varying only the camera. He's showing the final subject detail you will realize in focal-length limited situation. This is, very effectively IMO, testing the 'reach' ability of the three cameras in a real world type of situation.

Click to expand...

Yeah - I agree with your "point of the test" summary.

But ...

That is not testing reach. Alans test simply compares image quality for a subject at the same distance.

A true test of reach would be to see how much further away the subject could be with the 7D whilst still obtaining similar IQ. In theory (practice would be a little different of course), you'd be able to crop the 7D down to around 8MP and still have similar IQ. That would, of course, allow the subject to be much further away - hence ... more reach .

I can't believe some of you have been so completely mislead by this test. LOL!

clk_walker said:

One other consideration that must be accounted for is camera and/or subject motion. If you are using the 5DII at minimum shutter duration for the given scenario then the 1DIV and 7D would have to have shorter shutter durations to compensate. This would lower the SNR some or if not compensated their images would be blurred somewhat.

Click to expand...

No, the blurring in the resulting images of the subject would be exactly the same with the three cameras at a given shutter speed.

clk_walker636320 said:

Another consideration concerning maximum resolution: This only occurs at the plane of best focus (PBF). On either side of the PBF, out to the limits of the DOF, the resolution will be degraded. Birders will typically try to focus on the bird's eye. If not then image resolution will be degraded. Increasing the F# to increase the DOF will have more of a negative impact on the 7D than the 5DII due to diffraction.

Click to expand...

No again. At a given f-ratio diffraction will be the same in the resulting images. All cameras use the same size effective sensor area to record the image.

Alan, thanks for all this. I want to offer up an alternative interpretation and see if I have it correct.

You take a shot with the 600mm on a 7D and compare to the crop of a shot taken with the 600mm on a 5DM2, such that the fields of view are the same. You find no important/tangible/nonPP differences in image quality. This suggests two things:

1) If you want to capture the FoV of 600 on FF, you can shoot instead with the 400 on the 7D and not lose in image quality. (Cut me a break, everyone, on the difference between 1.5x and 1.6x, thanks!) So, the 7D increases reach, without image quality loss, in that one can shoot the same FoV with a shorter lens. And there is the significant added advantage, not mentioned in this thread that I saw, of a considerable reduction in weight.

2) The same concept in reverse. If a field of view can be captured on a 7D, the same scene can be shot on a 5DM2, with the same lens, with no improvement in image quality, but with a significant amount of extra image content on all four sides that can be used to expand the scene, shift it around a bit (say, to get the subject in a "perfect" rule-of-thirds spot), and so forth.

So to me the FF/crop tradeoff, for a focal-length constrained situation, is a matter of weight vs. flexibility in framing.

Well, I'll summarize a bit differently, same concepts, and add a question (as before, please ignore the difference between 1.5x and 1.6x crop sensor, I find the exposition easier to write if I use real lens focal lengths):

1) You are in the field with a 5DM2 and a 600 mm and no converter. You have a subject that fails to fill the frame. What are your choices? A) digital crop, B) put a 7D on the lens instead. Test result: no difference

2) You are in the field with a 7D and a 400 mm and a subject that fills the frame. What do you get by instead using a 5DM2 and a 400 mm and then digital crop? Test result: no difference

3) You are in the field with a 7D and a 400 mm and a subject that fills the frame. What do you get by instead using a 5DM2 and a 600 m? Test result: not applicable. Speculation, anyone? Side note: tremendous weight gain. An example of one form of increasing lens reach.

JRMDC said:

Alan, thanks for all this. I want to offer up an alternative interpretation and see if I have it correct.

You take a shot with the 600mm on a 7D and compare to the crop of a shot taken with the 600mm on a 5DM2, such that the fields of view are the same. You find no important/tangible/nonPP differences in image quality. This suggests two things:

1) If you want to capture the FoV of 600 on FF, you can shoot instead with the 400 on the 7D and not lose in image quality. (Cut me a break, everyone, on the difference between 1.5x and 1.6x, thanks!) So, the 7D increases reach, without image quality loss, in that one can shoot the same FoV with a shorter lens. And there is the significant added advantage, not mentioned in this thread that I saw, of a considerable reduction in weight.

Click to expand...

If I am following this correctly, you are suggesting doing the same amount of cropping with both cameras using different lenses to equal out the FOV. If this is the case, the image quality will not be the same.

As an example, if I crop an image taken with the 5DM2 at 50%, then do the same 50% crop with the 7D image, the 7D's image will be much noisier and image quality will suffer. This has to do with the pixel density of the 7D vs 5DM2, plus the MP's of the 5DM2 is 3MP greater.

(If you want to try this test for yourself, go to imaging-rescource.com and download full-size images from both cameras at the same ISO and crop each identically. You will see the difference in image quality.)

In the test that I performed, the 5DM2's image was cropped much, much more than the 7D's image but the outcome was very similar in quality.

To put it another way, you can crop heavier with the 5DM2 than the 7D and still get similar image quality.

As I stated in my article, the great equalizer in doing this test had to do with how heavily the images were cropped. The 7D's image was barely cropped, the Mark IV's image was moderately cropped and the 5DM2's image was substantially cropped. After doing this, all three images were fairly close in image quality.

JRMDC said:

2) The same concept in reverse. If a field of view can be captured on a 7D, the same scene can be shot on a 5DM2, with the same lens, with no improvement in image quality, but with a significant amount of extra image content on all four sides that can be used to expand the scene, shift it around a bit (say, to get the subject in a "perfect" rule-of-thirds spot), and so forth.

Click to expand...

This is true and as I stated at the end of the article, I'd much rather have a full-frame sensor that has good resolution with a fast fps and a good autofocus system. There's less chance of clipping a wing of a bird out-of-frame when photographing a BIF.

The 1DX sounds like it will fit the bill, although (1) it's going to be expensive and (2) I really want to see how well it will resolve detail being that it's only 18MP and a full-frame camera.

If the 5DM2 had a better autofocus system and was 6-8fps, I would forgo a slight decrease in resolution (compared to the 7D) for a full-frame sensor. Maybe the 5DM3 or 5DX or whatever it will be called, might be the ticket.

JRMDC said:

So to me the FF/crop tradeoff, for a focal-length constrained situation, is a matter of weight vs. flexibility in framing.

Click to expand...

JRMDC said:

Well, I'll summarize a bit differently, same concepts, and add a question (as before, please ignore the difference between 1.5x and 1.6x crop sensor, I find the exposition easier to write if I use real lens focal lengths):

1) You are in the field with a 5DM2 and a 600 mm and no converter. You have a subject that fails to fill the frame. What are your choices? A) digital crop, B) put a 7D on the lens instead. Test result: no difference

Click to expand...

I guess it depends on how much the subject fails to fill the frame of the 5DM2. It might actually be too large with the 7D. But let's say that the subject fits in the FOV with both cameras...I would pick the 7D, not so much because of a slight advantage image quality-wise, but for the better autofocus and fps.

JRMDC said:

2) You are in the field with a 7D and a 400 mm and a subject that fills the frame. What do you get by instead using a 5DM2 and a 400 mm and then digital crop? Test result: no difference

Click to expand...

The image quality would be very close with a slight advantage going to the 7D, but there would be an advantage of having more wiggle-room with the 5DM2's sensor...less chance of clipping the subject out-of-frame and having the advantage to crop the subject in a more pleasing way (rule of thirds, subject flying into the frame, etc.).

JRMDC said:

3) You are in the field with a 7D and a 400 mm and a subject that fills the frame. What do you get by instead using a 5DM2 and a 600 m? Test result: not applicable. Speculation, anyone? Side note: tremendous weight gain. An example of one form of increasing lens reach.

Click to expand...

The 5DM2 will produce a much better image in case #3. I described this commenting in your earlier post.

JRMDC said:

You are in the field with a 5DM2 and a 600 mm and no converter. You have a subject that fails to fill the frame. What are your choices? A) digital crop, B) put a 7D on the lens instead. Test result: no difference

Click to expand...

Sorry, but I don't think this is what either liquidstone's or aayjay's test results indicate.

Even after applying NR at 1600 ISO, the 7D image contains substantially more detail. Look at the top half of the feather. Then look at the bottom half of the feather.

http://forums.dpreview.com/forums/read.asp?forum=1019&message=40486279

aayjay said:

JRMDC said:

3) You are in the field with a 7D and a 400 mm and a subject that fills the frame. What do you get by instead using a 5DM2 and a 600 m? Test result: not applicable. Speculation, anyone? Side note: tremendous weight gain. An example of one form of increasing lens reach.

Click to expand...

The 5DM2 will produce a much better image in case #3. I described this commenting in your earlier post.

Click to expand...

Although this issue goes beyond the scope of this thread, and although I don't want to start a crop vs ff flame war here, IMHO, whether and to what extent the 5DII will produce a significantly better image (assuming the two lenses are both high quality, high resolving lenses) will largely depend on the ISO at which the images are shot.

As to weight, not only will the 5DII/600mm combination will be much heavier, the 5DII/600mm photographer's wallet will be much lighter.

bhollis said:

As to weight, not only will the 5DII/600mm combination will be much heavier, the 5DII/600mm photographer's wallet will be much lighter.

Click to expand...

So true! But to make the comparison "fair", as we are in the digital world, we really should compare the amounts of digital money and a credit car's weight does not vary (except psychologically).

aayjay said:

If I am following this correctly, you are suggesting doing the same amount of cropping with both cameras using different lenses to equal out the FOV. If this is the case, the image quality will not be the same.

Click to expand...

You did not follow it correctly, but you followed everything else and I understand it all, now. I presume.

As for my #3 in the other post, I now realize I was comparing entire-sensor images from the 5DM2 and 7D so of course the former has both more pixels and larger pixel sites so IQ will be better. Just wasn't thinking.

For my own purposes the combination of expense and weight dominates, considerably. So FF would not be a real choice. Nice to know the crop alternative is so good, and to better understand how it gets there and where it can still fall short.

Thanks for all the help, everyone.

JRMDC said:

Well, I'll summarize a bit differently, same concepts, and add a question (as before, please ignore the difference between 1.5x and 1.6x crop sensor, I find the exposition easier to write if I use real lens focal lengths):

1) You are in the field with a 5DM2 and a 600 mm and no converter. You have a subject that fails to fill the frame. What are your choices? A) digital crop, B) put a 7D on the lens instead. Test result: no difference

2) You are in the field with a 7D and a 400 mm and a subject that fills the frame. What do you get by instead using a 5DM2 and a 400 mm and then digital crop? Test result: no difference

3) You are in the field with a 7D and a 400 mm and a subject that fills the frame. What do you get by instead using a 5DM2 and a 600 m? Test result: not applicable. Speculation, anyone? Side note: tremendous weight gain. An example of one form of increasing lens reach.

Click to expand...

In none of these above examples is the subject far enough away to provide any trouble at all for the 7D.

What about ...

4) You are in the field with a 7D and 600mm lens + TC and a subject fill about one-third of the frame. What do you get by instead using the 5DII? Test result: - well, Alan hasn't tested this. But Romy has. ;-)

The 7D has more reach.

I completely agree with schmegg. I can't believe you guys don't get it. This is what matters to me when needing reach, which is almost always. In fact he worded it very well earlier. Put 8 mp on your subject with each test and see how much further away from the subject you are with the 7D.

Also has anybody mentioned how it's easier to keep an AF point on your subject (like on the eye of a bird) when the bird fills more of the frame. Or tracking a moving animal when it is larger in the frame. Forget heavy cropping of the 5D2 after the fact if you cant focus exactly where you want.

--
'The truth is rarely pure and never simple' Oscar Wilde

mattr said:

clk_walker said:

One other consideration that must be accounted for is camera and/or subject motion. If you are using the 5DII at minimum shutter duration for the given scenario then the 1DIV and 7D would have to have shorter shutter durations to compensate. This would lower the SNR some or if not compensated their images would be blurred somewhat.

Click to expand...

No, the blurring in the resulting images of the subject would be exactly the same with the three cameras at a given shutter speed.

mattr said:

Another consideration concerning maximum resolution: This only occurs at the plane of best focus (PBF). On either side of the PBF, out to the limits of the DOF, the resolution will be degraded. Birders will typically try to focus on the bird's eye. If not then image resolution will be degraded. Increasing the F# to increase the DOF will have more of a negative impact on the 7D than the 5DII due to diffraction.

Click to expand...

No again. At a given f-ratio diffraction will be the same in the resulting images. All cameras use the same size effective sensor area to record the image.

Click to expand...

Okay allow me to provide some background. The amount of blur due to camera and/or subject motion is proportional to the movement of the projected image on a photodetector during the time, T, the shutter is open. Define W in units of radians per sec as the angular of the image projected on the iamge plane. The projected image will move an amount given by W X T. The blur will be the same if the image movement is the same relative to the photodetector resolution or IFOV (instantaneous field of view). The IFOV is simply (photodetector size) FL. Therefore the relative amount of blur is given by: (W X T) ((photodetector size) FL). The only variable is photodetector size if the other three are held fixed. Certainly FL and W are fixed for this scenario. For a smaller pixel size the relative amount of blur will increase if T is held fixed. I hope this is a satisfactory explanation.

As to diffraction you are again misinformed. As with motion blur diffraction blur is relative to the IFOV of the photodetector. The diffraction blur will be proportional to the F# which is the same for all cameras. But again the IFOV the 7D is smaller than the IFOVs of the other two bodies. Hence diffraction will have some blurring effect on the image of the 7D compare dto say teh 5DII. It is not the sensor area that is important, which as you say will be the same, it is the area of a given photodetector.
--
A bird in the viewfinder is worth...

schmegg said:

What about ...

4) You are in the field with a 7D and 600mm lens + TC and a subject fill about one-third of the frame. What do you get by instead using the 5DII? Test result: - well, Alan hasn't tested this. But Romy has. ;-)

The 7D has more reach.

Click to expand...

Yes, I have tested this. Your wish is my command (and last test too):

Here's an outdoor test, taken with the Canon 600mm f/4 IS lens + 1.4x TC. As per your request, this test cropped the 7D's image to 1/3rd of the original dimensions. I then cropped the Mark IV's image and 5DM2's image to match the FOV of the 7D's cropped image. They were all upscaled. ISO800.

Here's the target area:



Here are the results at 100%:



Here's the results after apply NR and sharpening:



That's about it for me...I'm toast. You guys are lucky that the weather is really crappy here in SE MN, otherwise I'd be out taking pics of birds and not spending my time here!

Alan
http://www.iwishicouldfly.com

P.S....The test images from left to right are: 1DM4, 7D, 5DM2

We are talking about images from the same lens, recorded by the same sensor area, and viewed at the same magnification. Sampling at higher spatial frequencies will never result in more blurry images.

aayjay said:

That's about it for me...I'm toast. You guys are lucky that the weather is really crappy here in SE MN, otherwise I'd be out taking pics of birds and not spending my time here!

Click to expand...

LOL! Thanks for the entertainment Alan ;-)

Nice job - good tests - much appreciated by all I suspect.

Still - the 7D has the legs when it comes to far off subjects (apart from the other attributes). It's still got more detail and the 5DII has turned to mush - though your ISO800 non-selective NR'ed images level the playing field a little more than I think is necessary for a really informative comparison ;-)

Spent some time in Minneapolis just a couple of years back - shame we hadn't had this discussion before that trip - I'd have bought you a beer or two

Hope the weather improves fast so you don't have to suffer the likes of me for too much longer :-D

clk_walker said:

Okay allow me to provide some background. The amount of blur due to camera and/or subject motion is proportional to the movement of the projected image on a photodetector during the time, T, the shutter is open. Define W in units of radians per sec as the angular of the image projected on the iamge plane. The projected image will move an amount given by W X T. The blur will be the same if the image movement is the same relative to the photodetector resolution or IFOV (instantaneous field of view). The IFOV is simply (photodetector size) FL. Therefore the relative amount of blur is given by: (W X T) ((photodetector size) FL). The only variable is photodetector size if the other three are held fixed. Certainly FL and W are fixed for this scenario. For a smaller pixel size the relative amount of blur will increase if T is held fixed. I hope this is a satisfactory explanation.

As to diffraction you are again misinformed. As with motion blur diffraction blur is relative to the IFOV of the photodetector. The diffraction blur will be proportional to the F# which is the same for all cameras. But again the IFOV the 7D is smaller than the IFOVs of the other two bodies. Hence diffraction will have some blurring effect on the image of the 7D compare dto say teh 5DII. It is not the sensor area that is important, which as you say will be the same, it is the area of a given photodetector.

Click to expand...

LOL! :-D

Nope - you've spent a lot of words just to show you don't really understand the difference between pixels and images.

As mattr says ...

mattr said:

Sampling at higher spatial frequencies will never result in more blurry images.

Click to expand...

And the proof is in the crops posted in this thread.

clk_walker said:

mattr said:

clk_walker said:

One other consideration that must be accounted for is camera and/or subject motion. If you are using the 5DII at minimum shutter duration for the given scenario then the 1DIV and 7D would have to have shorter shutter durations to compensate. This would lower the SNR some or if not compensated their images would be blurred somewhat.

Click to expand...

No, the blurring in the resulting images of the subject would be exactly the same with the three cameras at a given shutter speed.

clk_walker636320 said:

Another consideration concerning maximum resolution: This only occurs at the plane of best focus (PBF). On either side of the PBF, out to the limits of the DOF, the resolution will be degraded. Birders will typically try to focus on the bird's eye. If not then image resolution will be degraded. Increasing the F# to increase the DOF will have more of a negative impact on the 7D than the 5DII due to diffraction.

Click to expand...

No again. At a given f-ratio diffraction will be the same in the resulting images. All cameras use the same size effective sensor area to record the image.

Click to expand...

Okay allow me to provide some background. The amount of blur due to camera and/or subject motion is proportional to the movement of the projected image on a photodetector during the time, T, the shutter is open. Define W in units of radians per sec as the angular of the image projected on the iamge plane. The projected image will move an amount given by W X T. The blur will be the same if the image movement is the same relative to the photodetector resolution or IFOV (instantaneous field of view). The IFOV is simply (photodetector size) FL. Therefore the relative amount of blur is given by: (W X T) ((photodetector size) FL). The only variable is photodetector size if the other three are held fixed. Certainly FL and W are fixed for this scenario. For a smaller pixel size the relative amount of blur will increase if T is held fixed. I hope this is a satisfactory explanation.

As to diffraction you are again misinformed. As with motion blur diffraction blur is relative to the IFOV of the photodetector. The diffraction blur will be proportional to the F# which is the same for all cameras. But again the IFOV the 7D is smaller than the IFOVs of the other two bodies. Hence diffraction will have some blurring effect on the image of the 7D compare dto say teh 5DII. It is not the sensor area that is important, which as you say will be the same, it is the area of a given photodetector.
--
A bird in the viewfinder is worth...

Click to expand...

Yes you would need a bit higher shutter speed in order to take advantage of the extra sensor resolution but OTOH no you wouldn't do worse with the higher density sensor if you left the shutter speed the same.

And same for diffraction. Diffraction is a rare, rare issue to encounter when one is focal length limited anyway. For all this type of shooting even f/2.8 is usually good or even ideal and even when not usually not more than f/5.6 is needed. And diffraction effects roll in slowly anyway so the true limit where the higher density sensor would stop gaining anything extra is rather closed down indeed for the densities that have appeared on DSLRs so far (although with the higher density sensors you do stop losing the ideal crispness they can obtain somewhat early on, but again, they don't do worse, just as not much as better than the lower density sensors).

Sorry for your misunderstanding. Here is a real world example. Tale teh 400mm f/5.6 and put on the 1.4x TC and you now have an f/8 lens. Shoot with a 20D and 7D. Yes I know you cannot AF with this combo unless you tape pins. The best "reach" results will be those obtained from Alan's test. Both motion blur and diffraction will affect the 7D image more than the 20D because of the smaller pixels of the 7D. The "reach" advantage of the 7D starts to lessen. Totally disappear? No. But now any advantage tends to be not so apparent. Look I am in favor of the reach advantage of the 7D since I just got mine Thursday and will be shooting a U of Florida BB game this afternoon with it, but I also understand the physics of the situation. If unclear about the diffraction visit DXOMark. Look at the MTF at center of a good lens like the 100mm macro on both the 7D and 20D/30D. Compare say f/4.0 with f/8. The differences start to disappear.
--
A bird in the viewfinder is worth...

clk_walker said:

Both motion blur and diffraction will affect the 7D image more than the 20D because of the smaller pixels of the 7D.

Click to expand...

Motion blur and diffraction at the image level will be the same. Smaller pixels comprise a smaller percentage of the total image area. Net effect - zero difference.

clk_walker said:

Sorry for your misunderstanding.

Click to expand...

LOL! Indeed!