If I stand 50' from a bird and take its photograph using my 100-400 f/4.5-5.6 L, using a full-frame 5D (or the upcoming whatever camera), then I take the same photograph with the same focal length and the same distance with a cropped sensor 40D (or say, the upcoming 50D) and then scale the images so that the bird is printed at the same literal dimensions with both cameras, which one will show more detail?

Will it be the same? Very similar?

Here is why I ask...

I have always liked the idea of the cropped sensor cameras for two reasons: One, the cropped sensor uses the central area or the "sweet-spot" of most lenses; Two, I get additional magnification from the cropped sensor while using the same lens.

But...what I am now thinking is that if the next generation 5D (whatever it is called) has the same or similar pixel pitch as the 50D, the magnification argument is irrelevant because I can simply crop the image in post-processing and still retain the same relative size and quality.

vignetting and barrel/pincussion distortion may be worse on the full frame camera, but that may be countered by other aspects of quality like dynamic range, or noise levels.

Comments...thoughts...

Well, one problem with your analysis is that these two cameras will likely not have the same or similar pixel pitch. The 50D sensor is: 22.3mm x 14.9mm = 332mm squared. The full-frame sensor is: 36mm x 24mm = 864 mm squared. We know that the 50D will have 15 mp. If the pixels were the same size, the 5D "evolution" would have about: 864/332 * 15 mp = 39 mp. Although the 5D is likely to have only about half that many pixels, they will be larger (nicer) pixels.

Yes, I see. So with that logic--under ideal conditions--the cropped sensor will have higher apparent magnification and more detail. It also won't be subject to as much potential vignetting, or distortion.

On the other hand, as the dynamic range of the scene increases and the lighting decreases (and the ISO increases) the full-frame will have a cleaner image and more dynamic image.

tradeoffs...tradeoffs...tradeoffs.

I am not one who automatically assumes the full-frame will be better for me, so I am trying to clearly assess the tradeoffs.

Of course, the proof will be in the actual image tests that eventually come out.

*Whups!* Didn't mean to post.

Well, the FF sensor should capture more detail if it turns out to have a higher pixel density than 15 mp. Cropped down to APS-C (50D) size, of course, the FF sensor would have fewer pixels and therefor capture less detail (ignoring other factors that contribute to resolution, such as noise).

And that kind of gets to the reason that I have generally preferred the cropped sensor. Since I am doing mostly panoramic photography, wide-angle lenses aren't an issue for me, so the extra resolution on long lenses is helpful and the lack of vignetting is important for doing panoramic stitches, particularly skies.

In the example, you will not get "the same photograph " with the cropped sensor. You will get the central portion of the one from the FF camera (same distance, same FL). Presumably you mean same "physical" dimension (when you say literal) - so an e.g. 10 mm high image of the bird on the sensors (it will be the same 10 mm in each case, given the same camera position and FL) you will need to specify the pixel pitch not the sensor size to know how many pixels are in the 10 mm. If you then print to get a 5 inch high bird, you will have a fixed geometric enlargement (of ~12.5x). While there are other mitigating factors, in the range you are considering the number of pixels plays a (or the) major role in determining the apparent detail in the print. The 5D would capture 1220 pixels in those 10 mm while the 50D would have captured 2100 pixels. The improved resolution would probably not be the full (2100/1220)*100% but it would be noticeable at many print sizes. The 5" print from the 50D would be at 420 ppi while that from the 5D would be 244 ppi.

I suppose another consideration would be the files themselves. A camera with more than 20MP will require more storage per shot than a 15MP camera. Moving from a 20D to a 40D made my whole workflow a lot more demanding of data speed, to the point where I needed a faster processor in the computer to make it perform as well for the same amount of processed images. Personally, I wouldn't want to shoot more MP than I would print. But if I needed to make a 20+ MP print, I'd have to go with the 20MP camera.

If the bird is either printed or viewed on monitor at the same literal dimensions, a 50D image will be comprised of approximately 3x the number of pixels (as the current 5D) and will therefore look sharper or more detailed to some extent, depending on the quality of the lens at the aperture used, and assuming a sufficiently fast shutter speed was used to get a sharp image.

The issue has already been covered in a recent thread on the 50D. As lenses are stopped down to apertures where diffraction becomes apparent, the benefits of increased pixel density diminish, but will always be apparent to some slight degree if the print is large enough, however insignificant that might be.

Edit: It is of course assumed that the bird will be cropped in one or both images, otherwise it would not be possible to view or print the bird at the same literal dimensions.

The recent thread is what got me thinking about this topic. I thought, what if I used a full frame camera and overlapped my panoramic stitches over the central portion of the sensor (the overlapping area of the APS-C sensor if it were laid on top of the FF sensor) and then made a macro that cropped the Full-Frame output to the cropped sensor size . By doing this, I wondered, would I eliminate the vignetting and distortion from the edges of the lens and get a similar resolution. It seems unlikely that the 5D replacements(s) will be 39MP to match the pixel pitch of the 50D, but it might be more comparable to the 40D or the 30D (the one I currently own).

So, I am evaluating this odd approach of cropping FF images to get some of the benefits of the cropped sensor while retaining the benefits of the FF sensor performance. In this case, it may be more helpful to compare the new 5D to a 40D or 30D.

Cropping is cropping, whether it's done by the camera's sensor or later in Photoshop in relation to the wider angle of view one gets from the larger format. The advantage of the cropped format has always been one of higher pixel densitiy and the possibility of a more detailed image using the same lens from the same position.

However, normally one doesn't use the same lens if one is at all concerned with image detail and general sharpness. One tries to fit the focal length of the lens to the composition and FoV. If one does this, then generally there wouldn't be much difference in image detail between a 15mp 50D image and a 12.7mp 5D image, depending again on the quality of lens at the apertures chosen.

For example, I would guess that a 5D image taken with the Canon 85/F1.2 portrait lens at F2.8, would be better to some degree than a 50D image taken with the Canon 50/F1.4 at F1.8. In both cases the FL and apertures are matched for equal DoF and FoV, but the 85/1.2 is sharper at F2.8 than the 50/1.4 is at F1.8.

Understood, but what about using my 100-400. I really can't afford anything substantially longer. My wallet and my shoulders can't handle the longer lenses, so for those lengths, getting that extra 1.4X can be nice.

You were correct at the outset. Assuming the bird fits into the crop sensor area you will get more pixels and, therefore, more (potential) detail with the higher pixel pitch of the 50D in the situation you describe (same lens, viewpoint etc).

There may be other tradeoffs around diffraction limitations as diffraction effects will start to become visible earlier on the 50D than the 5D (say f5.6 to 8 instead of f11 to 16), but you will never be worse off. Other factors that may also come into play would include noise performance of the camera system, lens aberations limiting resolution and possibly (but I think to a lesser dfegree than is suggested by the amount of webtime expended on the topic) the relatie impact of the AA filters on each camera.

Mike

Even with the 8mp 20D, I get noticeably sharper results using my 100-400 at 400mm and F8 or F11 than the 5D gives me with the same lens at the same apertures. After cropping the 5D image to the same FoV as the 20D image, one is comparing just under 5mp with 8mp. Comparing 5mp with the 15mp of the 50D would be no contest.

That is the question I was trying to answer. It seems likely that this will hold true with the next gen 5D or 7D or whatever-D.

So the fact it is a crop sensor is not the reason you are getting the better detail (1.4x sensor size does nto mean a 100m lens becomes a 140mm lens in terms of msgnification at the sensor). It is down to the sensitivity of the sensor which in turn some comes from the pixel pitch/density.

okay...if you say so.

I tire quickly of semantic distinctions like this. The apparent magnification of the lens is an intrinsic property of the APS-C sensors when compared to FF sensors. You are correct that the image projection that the lens produces remains the same size and that there are limitations imposed based upon the lens quality. But, prints made from an APS-C sensor have higher apparent magnification and more actual detail than full-frame sensors with the same lens and of the same generation.

So, if magnification of your subject is a major concern (getting more apparent length from your long lenses) and you can accept slightly reduced dynamic range and slightly increased noise at higher ISOs, the cropped sensor makes great sense at a much lower cost. Depending on your needs, either APS-C or FF could be a better tool.

So, to summarize my understanding of the tradeoffs between FF and APS-C...

Benefits to FF
More wide angle lens options.
Better dynamic range.
Better high ISO performance.
Diffraction effects are less pronounced.

Liabilities to FF
Vignetting will be more pronounced when it is present.
Lens distortion is more noticeable, particularly near the edges of the frame.
Telephoto lenses don't benefit from cropped-sensor 'magnification.'
Substantially more expensive (including body and longer lenses)

Benefits to APS-C
Increases the effective focal length of lenses
Decreased vignetting.
Substantially less expensive (including body and lenses)
Lens distortion is less noticeable, typically because it occurs near the edges of the frame that are 'cropped' from view.

Liabilities to APS-C
Increased noise at higher ISOs.
Decreased dynamic range.
Diffraction effects are noticeable a few stops sooner.
True wide angle lenses are hard to find.

With the bird at 50' using the 100-400 L neither camera will show any detail to speak of. Get closer, much closer.

It's an ostrich.