100% crop useful for cameras of different pixel density?

I know this has probably been discussed... maybe even to death, but a forum search results in a deluge of threads with have nothing to do with the narrow topic of how useful (or not... I claim not) is using 100% crop to compare (for example) high ISO noise between cameras of different pixel densities but similar sensor size (e.g. APS-C).

It seems to me the useful comparison is the same image "physical" size rather than 100% crop, which automatically puts the higher pixel density sensor at a disadvantage (because it is "blown up" more) and is a mostly useless comparison in the real world of photographers. It has use to sensor designers, but not photographers.

Can anyone provide a link to a "definitive" thread on this topic?

If not, maybe we can start a discussion (again?) on it.

Sorry... posted in wrong forum... Can the mods please move it to the EOS camera forum? Thanks.

Maybe the best solution to this problem is simply not to measurebate too much.

The only meaningful comparison is to look at them at the same physical size, 8 x 10, for example.

RTPVid wrote in post #11223978
I know this has probably been discussed... maybe even to death, but a forum search results in a deluge of threads with have nothing to do with the narrow topic of how useful (or not... I claim not) is using 100% crop to compare (for example) high ISO noise between cameras of different pixel densities but similar sensor size (e.g. APS-C).

It seems to me the useful comparison is the same image "physical" size rather than 100% crop, which automatically puts the higher pixel density sensor at a disadvantage (because it is "blown up" more) and is a mostly useless comparison in the real world of photographers. It has use to sensor designers, but not photographers.

Can anyone provide a link to a "definitive" thread on this topic?

If not, maybe we can start a discussion (again?) on it.

It depends on how different the pixel densities are.

Comparing a 16 megapixel APS-C sensor to an 18 megapixel APS-C sensor at the pixel level is useful because they're both roughly the same pixel density -- enough that resizing one to match the other isn't going to make much of a difference. But comparing a 10 megapixel APS-C sensor to an 18 megapixel APS-C sensor (or even a 15 megapixel APS-C sensor) can't really be done properly on the pixel level since the pixel densities are sufficiently different that resizing to match (when done properly of course) will have a marked effect on the result.

I don't think 100% crops are good for looking at camera sensors at all.
I use it to help see how my focus, the bodies AF, and sometimes how lenses are behaving .. but only on the rarest of occasions.

For ISO noise, the trouble is that detail = more noise,. less noise = less detail. So for a 100% crop the image with the least detail will look best,. but when viewing the entire image at normal print sizes,. one will often quickly find that the image with the least noise had the least detail and therefore is not the best.
YMMV

Unless you have a point what's the point

Heh! It's all in what you want to see! For some types of photography and much "normal" use most noise doesn't matter much because it gets lost in small prints and image compression. For other types it matters a lot (things like birding and wildlife are examples because they often require close crops). So, examining the noise characteristics of various cameras can be important to some photogs and relatively meaningless to others. Now, your point about comparing cameras of a different pixel density is a point to consider, for sure, when viewing images at the same time, since the higher res image will be less enlarged and less noise may be visible, but then bear in mind that one of the big reasons why people buy the higher res cameras is because they want more cropping room, which then gets us back to the question of relative noise characteristics.

So, you get a feel for what's meaningful to you and your photography! My criteria may differ! And Joe Shmoe may have his own criteria!

In other words, don't fall into the syndrome of judging what other people are interested in by your own personal criteria...

Here's a thread that has been active over the past couple months with some interesting comparisons between the 7D and the 5D2:

https://photography-on-the.net/forum/showthre​ad.php?t=930196

I can see pixel density entering into the equation..

Consider shooting a full moon with a 400mm lens... First on a 5DII and then on a 7D.

The 5DII moon will be about 700 pixels wide and the 7D moon will be about 1000 Pixels wide. So should the 7D image be down sampled to the same 700 pixels as the 5DII image to truly compare the images?

Would downsampling reduce the effects of noise slightly?

Jim_T wrote in post #11226375
I can see pixel density entering into the equation..

Consider shooting a full moon with a 400mm lens... First on a 5DII and then on a 7D.

The 5DII moon will be about 700 pixels wide and the 7D moon will be about 1000 Pixels wide. So should the 7D image be down sampled to the same 700 pixels as the 5DII image to truly compare the images?

Would downsampling reduce the effects of noise slightly?

Well, for the moon you typically don't want to shoot at a high ISO!

But, sure, to compare directly with the 5D image at the same image size you'd have more pixels in the 7D image and sure you have the 7D image a bit more compressed, but when I'm shooting wildlife I'm not comparing it to a 5D2 image -- I often have to crop at or close to a 100% crop!

Looking at 100% crops is only of use if you understand and can account for what you are used to looking at.

Once upon a time I did my sharpening and noise reduction at 1:1 views with my 8MP 30D.

When I got a 12MP 5D I would often toggle between a 1:1 view and a 1:2 view. Same with the 10MP 1D Mark III, though I mostly watched 1:1 views with this camera for noise and sharpening.

Now with the 16MP 1D Mark IV I usually do my NR and sharpening at 1:2. I find that NR done at 1:1 generally leads me to excessive NR application and plastic looking skin in even 16x20 prints. 1:2 image viewing on a 16MP image is better for even large prints.

RTPVid wrote in post #11223978
Can anyone provide a link to a "definitive" thread on this topic?

I would argue that this thread I made covers exactly what you're talking about in a definitive manner:

Small pixel sensors do not have worse performance

Specifically, see the section on "Spatial Frequency" -- the number one most common type of flawed image analysis.

Wow! Someone who got it, right off the bat. Congratulations!

JeffreyG wrote in post #11226772
Looking at 100% crops is only of use if you understand and can account for what you are used to looking at.

Once upon a time I did my sharpening and noise reduction at 1:1 views with my 8MP 30D.

When I got a 12MP 5D I would often toggle between a 1:1 view and a 1:2 view. Same with the 10MP 1D Mark III, though I mostly watched 1:1 views with this camera for noise and sharpening.

Now with the 16MP 1D Mark IV I usually do my NR and sharpening at 1:2. I find that NR done at 1:1 generally leads me to excessive NR application and plastic looking skin in even 16x20 prints. 1:2 image viewing on a 16MP image is better for even large prints.

+1
If you can understand the relationship between your 100% view and your print at the desired print size, then it is a useful thing to look at.

If not, then you should make the prints you want and see how they come out with different processing methods *while you keep track of your enlargement factor*

If you only make the same print size all the time, then you don't have so much to keep track of.

However, if you vary your cropping extent but keep a fairly standard print size, your degree of enlargement will be all over the place. Therefore, knowing that it looks like X in the 100% view is a reference point that can be useful.

If you bought a higher MP camera/sensor so you could make bigger prints by using all the pixels, then 100% view will be a significant evaluation view.