Canon 5D3 vs 7D for birding thoughts

[dr croubie]

- I am sure a 1D4 is much better than a 7D, but that is not the question.  The question is whether it is worth keeping the 7D along with a 5D3.

Well, my outloud-musings were along the lines of:
1D4 >> 7D
1D4 ~= 5D3. (give or take a bit)
ergo, 5D3 > 7D.

Seeing as noone else seems to have compared the two directly, you could be the first? Forget the T/C to begin with, just the 70-200 on a tripod, same iso, aperture, shutterspeed, crop the 5D3 to 3568*2384 (that's still an 8x12" print at 300dpi), and see if 8.5MP really does beat the 7D's 18MP. That's a fair whack of difference in resolution, but if it's the choice between a noisy shot, a blurry shot, or a clean 8MP shot, the latter might win...

Personally, i wouldn't mind the 2fps loss and less pixels on target, in return for the extra IQ, FF sensor, higher-iso, 61pt AF. I can't justify the extra expense from selling my 7D for the 5D3, but seeing as you've already got both, the question is between keeping both bodies or selling the 7D for cash towards something else (like the front element of a 600/4, you'd have to fund the rest of the lens from elsewhere).

If you like having 2 bodies for backup, keep the 7D, but if it were me i'd be trading it for better glass...

[dlleno]

glad you are asking this!  The 7D definitely has a higher pixel density than the 5d3. Hope this thread attracts some expert technical light on this and the following:

In terms of pixel density, the 5D3 is similar to the 40D, so one could conceivably compare these two in terms of resolution, when cropping the FF to match that of the 40D . In fact, If Canon were to introduce a 1.6 crop mode into the 5D3 like the D800 I'm assuming such a camera would be capable of producing 1.6x cropped images at 40D sizes. Except for the musings about Canon producing such a camera, I would like to know if this is accurate or if am I out in the weeds here. 

Ignoring the ISO, noise, focus performance, shadow detail and other IQ differences between 5D3 and 40D, and concentrating on resolution, is it possible or meaningful to compare the following results:

1. Take the 5d3 image and crop it in PP to match the 1.6x crop sensor FOV. How would this IQ (resolution) compare with the same image photographed by the 40D using the same lens at the same focal length with no PP crop?

2. use today's 5d3 in the hypothetical 1.6x crop mode. How would this compare with the native 40D image.

I realize the original question is about 7D, but I thought it would be instructive to think of the problem in terms of equivalent pixel density -- where the cropped 5D image is probably more similar to a 40D image than it is a 7D image. 

[ajay]

First things first, if you are going to test both camera's image quality for birds, you have to test them with the exact same subject from the exact same distance. If you do not, it's not a true comparison. At least when I am photographing birds, 99% of the time, I have to photograph them from a fixed distance unless they are tame birds.

In other words, in order to test both cameras fairly, you have to simulate taking the same exact shot standing from where you were when you took the image with the other camera.

I have done this test fairly with both cameras. Of course, the 5DM3's image is going to have a much wider FOV than the 7D in this scenario. In order to compare them identically, I cropped the 5DM3's image to match the FOV of the 7D. (This is something you would normally do in post-processing work.)

Here's the result taken at ISO 800, Canon 70-200 f/2.8 IS II:


The 7D's image is on the left, 5DM3 is on the right.

Since this only shows part of the full image, you can't see that there is more noise with the 7D's image under certain color/shade variations. The bottom line is that the 7D's resolution is slightly better than the 5DM3's while the 5DM3's noise level is slightly lower. After applying NR and post-sharpening to both images, they are barely indistinguishable. This holds true from ISO 100 to ISO 3200.

Of course, this test is really only valid for subjects that you cannot move closer to. If you could move closer with the 5DM3 to frame it identically to the 7D, the 5DM3 would have better image quality. This test is only valid for photographing birds/wildlife in which you cannot move closer.

BTW, this test result is not too different from a similar test I did with the 7D, 5DM2 and 1D Mark IV. This is what you should expect from the 5DM3's sensor which is almost identical to the pixel resolution of the 5DM2.

Alan
www.iwishicouldfly.com

[candyman]

Clicky Linky for full article

OK, that's a setup, but having an onboard-flash controller like the 7D leaves your 550/580EX free to add more light on the subject...

Ouch. And the bird needs to make an appointment with the optometrist or optician.....


Seriously, what happens with the bird when you press the shutter?

[stilscream]

I too was trying to decide 600f4 ii and use my 7d for awhile longer or 1dx (1st FF) and some L lenses-- perhaps the newer 24-70, or 14 f2.8+ 50 f1.2 or 24mm f1.4+ 85mm 1.2 (I have the 35f1.4 and love it-- do like landscape as well as birding. The thinking here is that the zoos are easier places to take good pics of animals even though it feels like cheating. So, 1dx improves my speed, IQ, low light, weather sealing, and AF over my 7d and offers a 2nd body to have multiple cameras with different primes. The downside is I really can't get the pics in the wild still. 600mm is a need for birding. Trying to curb my spending below $15k. (Nice problem to have, I know, but this is money coming from my late parents house inheritence.) I really swing wildly back and forth daily or even hourly. The non f8 focusing of the 1dx makes me think the 600 might be the better way to go, but the versitility of the FF makes me reconsider.

[briansquibb]

The downside is I really can't get the pics in the wild still. 600mm is a need for birding.

I use a tripod with gimbal for the 600

[neuroanatomist]

- I already have both the 5D3 and the 7D, so it is not a question about which camera to purchase.  I have already compared the two cameras - unscientifically - and found that even when the crop is considered the 5D3 produces much better images. 

Ok, so make it more scientific.  Set up a target of some sort, something with lots of detail, and a tripod at a distance you usually shoot at, then take a shot with the same lens on both cameras, crop the 5DIII to the FoV of the 7D, and compare IQ.  Then move the tripod to a further distance, such that you'd need tro crop the 7D, take the two shots, crop the 7D to the desired framing and the 5DIII to the same framing, and compare IQ (and also look at the resulting MP of the cropped 5DIII image to see if that will be enough for your uses.  I'd try the above both ISO100 and ISO3200 on both bodies.

[dlleno]

and just to keep these numbers in mind...  when a 5d3 image is cropped to 1.6x equivalent, the result is 8.7mp. 

22.3mp divided by 1.6^2 = 8.7

right?

Its astonishing to me that the comparison posted earlier, which was quite well done, showed only a marginal edge (in resolution) to the 7D's 18mp image, compared to the cropped 5D image at 8.7mp.  Then I had to remind myself that of something called a 1D mark II  which sported an 8.2mp sensor, and the FF, 11mp 1Ds which sold for $8K

[briansquibb]

and just to keep these numbers in mind...  when a 5d3 image is cropped to 1.6x equivalent, the result is 8.7mp. 

22.3mp divided by 1.6^2 = 8.7

right?

Its astonishing to me that the comparison posted earlier, which was quite well done, showed only a marginal edge (in resolution) to the 7D's 18mp image, compared to the cropped 5D image at 8.7mp.  Then I had to remind myself that of something called a 1D mark II  which sported an 8.2mp sensor, and the FF, 11mp 1Ds which sold for $8K

The real trick is to compose the pictures the same in ff, aps-h or aps-c by getting the right lens so there is no cropping to get the same picture.

That way you will get more mp on the subject.

Any yes I do know it will cost more, just in the same way a Ferrari cost more than a Mustang - you pay for what you get.

[CanineCandidsByL]

Seriously, what happens with the bird when you press the shutter?

I think we can fall back to that old adage "Poop happens".

That would have been a funnier picture with a roast chicken underneath the feeder.

[dlleno]

and just to keep these numbers in mind...  when a 5d3 image is cropped to 1.6x equivalent, the result is 8.7mp. 

22.3mp divided by 1.6^2 = 8.7

right?

Its astonishing to me that the comparison posted earlier, which was quite well done, showed only a marginal edge (in resolution) to the 7D's 18mp image, compared to the cropped 5D image at 8.7mp.  Then I had to remind myself that of something called a 1D mark II  which sported an 8.2mp sensor, and the FF, 11mp 1Ds which sold for $8K

The real trick is to compose the pictures the same in ff, aps-h or aps-c by getting the right lens so there is no cropping to get the same picture.

That way you will get more mp on the subject.

yes of course, +1 on that, to be sure!  But particularly in the wildlife scenario, where this comparison is important and meaningful (and part of the OP's original inquiry), one cannot simply decide to move closer, put a different lens on, or plead with the bird to take a different flight path.  What we are out to discover is the effectiveness of the 5D versus the 7D at photographing subjects at a distance and with the same lens.   

It takes a 46mp FF body to equal the pixel density of the 7D, such that one could crop the 46mp image at 1.6x to produce an 18mp image with the 7D FOV.   Clearly, pixel density is not the whole story, however ,because we have seen here that when the 5D3 image is cropped to 1.6x, the result is an astonishingly good 8.7mp image with the same FOV very close to the 7D itself in sharpness.   

maybe Canon doesn't want to let the secret out or maybe they want to preserve 7D sales I don't know.  But unless they can significantly improve the actual resolving power the 1.6x sensors,  it seems to me that the 'reach' advantage will soon be both a myth and a non sequitur, and that the FF boys will be able to produce the same IQ of the same subject at the same distance, with the same lens,  simply by cropping their high-mp FFs to match their 1.6x buddies standing next to them.   

the above makes a lot of amusing assumptions, but makes good conversation anyway

[neuroanatomist]

But particularly in the wildlife scenario, where this comparison is important and meaningful (and part of the OP's original inquiry), one cannot simply decide to move closer, put a different lens on, or plead with the bird to take a different flight path.  What we are out to discover is the effectiveness of the 5D versus the 7D at photographing subjects at a distance and with the same lens.   

+1, that's exactly the issue.  In my case, I can get the 1D X, or put that money towards a supertele prime.  Since only 10-15% of my shooting is birds/wildlife, the 1D X makes sense for now (meaning 12-15 months to save for the 500/4 II).

It takes a 46mp FF body to equal the pixel density of the 7D, such that one could crop the 46mp image at 1.6x to produce an 18mp image with the 7D FOV.   Clearly, pixel density is not the whole story, however ,because we have seen here that when the 5D3 image is cropped to 1.6x, the result is an astonishingly good 8.7mp image with the same FOV very close to the 7D itself in sharpness.   

Indeed.  But...what about a crop of the 7D's FoV to 50% of its original area?  Would the resulting FF crop still be close in IQ, and is 4 MP sufficient for the end use?

[jm345]

So far, based on Alan's test, there is not much difference in IQ between the the 7D and 5DIII cropped to the 7D's frame size. But to neuroanatomist's point, it isn't always that simple - especially with small birds. There are times my 7D and 600mm with extenders won't get a frame filling shot and I will need to crop the 7D image. So the "extended" question is what happens to the 5DIII's IQ when it is cropped to to the same size as a cropped 7D image?

BTW, good discussion here.

[dlleno]

As I look at the crop/FF technology, the simplest common thread that pops out to me is that;

1.  the FF sensors always have a lower pixel density which allows them to control noise and advantage IQ
2.  the crop sensors with higher pixel densities have poor noise performance and limited actual resolving power
3.  To some extent, (2) above may be due to less expensive technology used in the crop sensors.

I'm using the term 'resolving power' because, as many have pointed out, increasing pixel density does not (necessarily)  a higher resolution system make.  Case in point; As many and as profound as the 7D's strengths are, compared to the 40D, resolving power isn't the biggest "a hah" discovery.  Its better, but not by leaps and bounds as the pixel count would imply. Which means -- the more pixels you cram into the same space the more physics dictates that you will have consequences (such as noise) that limit actual resolving power. 

So -- and I'm  making some broad hypothetical assumptions here -- if we assume that IQ limitations are dominated by the artifacts of pixel density,  then its clear why there is a benefit to longer lenses and bigger sensors, and that a crop sensor will never equal a FF sensor of the same pixel count.  Also, and following the same assumption, we would have to conclude that a FF sensor at 46mp (cropped to 1.6x) would perform identically to an 18mp crop sensor of the same technology-- because the actual resolving power would be identical. 

I would suggest that as long as the technology itself, in the FF, is superior -- that is, the artifacts of pixel density are better controlled, then we could see the crop camera offering no advantage (other than cost) over the FF, in the situations we are discussing.  Ajay's comparison is quite striking here, in that the 5D3 should behave more like the 10mp 40D, when cropped to 1.6x, instead of the 18mp 7D.  Moreover, if the 5D3 would have offered mid-thirties mp, I bet we would see equivalency to the 7D or even superior!

The  point that was raised about "over cropping" is interesting, i.e. how much can you throw away and still have a decent picture.  I'm not a physicist but let me suggest that the answer is still about resolving power -- the sensor that has the best resolving power will show the best image.  what does that mean -- to me it means that the comparison earlier, posted by Ajay, will continue to hold even as you crop things down further and further.  the 7D will continue to show a slight edge, but as a practical matter the difference between the two will become more apparent as you crop further and further until the resulting image is essentially a pixel-peep. 

Ajay you could demonstrate that by simply cropping the images further in Post.

[blufox]

I think another factor which needs to be kept in mind is that when you actually crop a FF image, especially when shooting wildlife or Birds etc, you have to keep exposure int mind too.

When you use a lens of 500mm Focal length on a 7D you are using it at 800mm. So, what happens is that your metering and luminosity is spot on at 800 mm because that is what the camera sees i.e the real distance and so do you from your viewfinder.

Now do keep in mind that light follows the inverse square law for luminosity, so the correct exposure for the same subject on a FF with 500mm lens is screwed up *if* you decide to crop for 1.6x factor.

So definitely there is a trade off between resolution and noise. Add to this that you have to decrease your shutter speeds further on a higher density APS-C sensor to get sharp images.

I think 1DMkIV with a 1.4TC is still a much better combo than anything out there. But that is just my opinion .

Please feel free to correct me, if I am wrong somewhere.

Thanks,