B&H: L's and macros only for EOS full frame DSLRs

The B&H newsletter today had an article on anti-shake technology, IS

in the EOS system. In the middle of the article it says that,

although any EF lens works well with EOS full frame film cameras,

Canon "strongly recommends" only L and macro lenses for their 1D and

5D full frame DSLRs (for large prints). This because, it says, the

top-of-the-line DSLRs are sharper than film.

 

First, I wonder where this strong recommendation is documented.

 

Second, this doesn't really apply to me immediately, as I'm happy

with my 350D. But I was sure hoping I'd still be happy with my EF 70-

300 f/4-5.6 IS USM in a few years when Canon makes a full frame DSLR

I can afford.

 

Third, if this is true, surely it means 12-16 megapixels is all one

could "ever" need for any but the most specialized applications.

Remember that the more expensive the lens, the more money B&H make selling it.

 

Better lenses are better? Duh, I could have thought of that ome myself!

 

You NEED "L" or macro lenses on a full frame DSLR? Total nonsense. You'd be wasting your money putting a cheap $100 consumer zoom on one, but that pretty much applies to film as well if you're shooting fine grain film.

 

The 70-300 performs quite well on a 5D. I know because I've used that combination. Not as well as a 300/2.8L IS of course, but then that's pretty obvious.

I have read about this myself in different photo magazines. What the reviewer was stating is that the chip in the 1Ds Mk II was picking up imperfections in the lower end lenses and to really get the best images the better lenses were needed. That holds true for anything if you ask me... But the effects become more dramatic the larger you output your images and a 16.7mp file lends itself to larger prints and cropping.

Michael - Is Canon going to step up to the plate with some 'film killer' statement? No. As long as Canon still sells film bodies, they are not going to change.

 

Is Canon going to tell people to buy $$ lenses to go with their $$ cameras? Of course. And, fo course, the people who buy 5D and 1D cameras are either filthy rich or are much more concerned with photography and the quality of their work than the total consumer who just wants a memory maker.

Norman has it right: the higher resolution the sensor, the more imperfections in the glass become apparent.

 

APS digital bodies and film also benefit from better glass. I would argue that many imperfections are more apparent on APS digital than on film as well, and that this is only magnified when you go full frame. This doesn't mean you can't use mid-grade glass on digital, nor that you can't produce good images with it. Just beware that differences in MTF, corner performance, and color rendition become more apparent.

 

As to why imperfections would be more apparent on APS digital: it's not about the extinction lpmm resolution, it's about the MTF. Digital sensors have much higher MTF than film, which means differences in lens MTF will be more apparent.

 

Similar situation with color rendition: film choice and processing typically obliterated any variations in lens color bias except for the most demanding professionals comparing with a single film and tightly controlled processing. But your sensor doesn't introduce the same magnitude of bias and shift color wise. So you might notice a difference between two lenses you would have never picked up on with film.

 

I'm surprised the recommendation is "L or macro" as there are numerous non-L primes which should do fine even on a 1Ds mkII, primes which match or beat L zooms.

 

FWIW, 12-16 MP FF is all most will ever need. Certainly there are photographers that want/need more. But at 16 MP you're well past saturating common print sizes with data.

Also it is much easier to pixel peep your DSLR images and view the smallest imperfections.

Most film shooters glanced at their 4 x 6 prints and that was that. Very few tossed slides or

negs on a light table and checked them out with a high powered loupe or low power

microscope. I did, and saw all sorts of imperfections around the edges of my chromes and

negs.

Okay, this is the equation. It's very simple:

 

Where R is resolution in lp/mm, (1/Rtotal)^2 = (1/Rlens)^2 + (1/Rfilm)^2

 

So if your lens is capable of delivering 100 lp/mm to the film and your film/sensor is capable of resolving 100 lp/mm - you will only actually get 70 lp/mm! If you want the full 100 lp/mm of your sensor, you have to have a lens that can deliver 400 lp/mm.

 

But just like loading a sharper film behind any lens DOES produce a sharper image than a less sharp film behind the same lens, so putting a higher resolving sensor behind any lens produces a sharper image than a lower resolving sensor behind that same lens.

 

Both lens and sensor are complementary limitations to each other, and neither is currently "maxed out" on any Canon camera with any Canon lens. THis is also why Nikon is able to squeeze so much more capability out of higher and higher resolving sensors.

 

HoweverI think the problem here is that we aren't talking about higher resolving sensors, but lower resolving sensors...albeit larger. We're comparing a Yashicamat D with a Leica. Yet, the Yashicamat will still produce an image that is more pleasing than the Leica tonally, if not in acutance.

Kirk, could you provide some proof, theoretical derivation or experimental demonstration of your resolution equation?

Probably not. It's an empirical obersvation which has been found to be approximately true for film. There's realy no sound theoretical basis for it, but it does appear in some Kodak literature. Actually you can often get a better fit by using an exponent other than 2. The equation is of the generally correct form but it's by no means exact.

 

However you can't really apply it to digital because the way digital resolution works isn't really like film resolution works. Film MTF gradually tails off towards higher and higher spatial frequencies.

 

Digital MTF on the other hand hits a brick wall at the Nyquist limit (which is somewhere between 60 and 80 lp/mm for most DSLRs). It does drop a bit before it hits the wall due to the anti-aliasing filter present in most cameras and due to the nature of digital sampling.

 

Besides this, resolution is NOT what you want to measure. What you want is the MTF of the lens multiplied by the MTF of the sensor in the region below the Nyquist sampling limit (60lp/mm to 80lp/mm depending on the sensor). That's what predicts sharpness, not "resolution" numbers.

 

See http://www.bobatkins.com/photography/technical/mtf/mtf1.html for a somewhat detailed (but still simplified) explanation.

Horses for courses...

 

If you want the best quality big prints from DSLRs, you need top quality sensors, top quality lenses, and top quality tripods. And a good eye to make that top technical quality photograph interesting. That is nothing new.