Camera Labs

Well here is some food for thought:
Now that Nikon has announced their first FF-body - the much anticipated D3 - many of the already existing lenses and esp. the new 14-24mm and the equally new 24-70mm come into special focus, because they can deliver an image-circle that is large enough for the brand-new 12MPix Nikon-D3-"FX"-sensor.
Putting aside the "FX"-designation as pure marketing-hype and using the FF-denominator instead so that everybody knows what we're speaking of here, there comes up an interesting dilemma: How do you fully test a Nikon FF-lens (like my latest toy, the 50mm/f1.4) on a Nikon-body and do the lens justice?
Why should this be a problem? Well, if you put it on the D300, you can only test the APS-C part of the image-circle with a 12MPix sensor. But you really have no way of knowing haw sharp that little beast is in the FF-corners, don't you???
On the other hand, put it on a D3-body and you can measure to your hearts content all over the place, but you only have a sensor resolution of a 5MPix equivalent on a APS-C sensor
Well, what can happen then? Ok, e.g. that nasty color-fringing on the APS-C-sensored D300 could completely disappear on the D3-sensor ... and later reappear on a D7-sensor...

So you'll never get to know the real quality of that particular lens in toto (as the old Romans would have said). Even if you do a double test with both the D300 plus the D3 - you will never know how the corner-sharpness of that lens really is! Maybe you'll be quite disappointed with the performance of the lens you so expensively bought, when Nikon finally comes out with that 25MP-FF (sorry "FX) sensor.
And no-one out there has a solution to this challenge! That's pretty crazy. So for the time being you just have to trust Nikon, that their lenses which are sharp on a APS-C 12MPix-sensor, also will be sharp in the corners on the up-and-coming 25MPix-FF-sensor.

Good luck, Nikon(-users)!
Canonites clearly have it better in this respect

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

I don't understand this. The D3 has a full frame sensor so you can fully test the full frame lens at the full 12Mp resolution. The D3 only drops to 5Mp when you put a DX lens on it.

_________________
Nikon D80, Nikkor lenses: 35mm f1.8 G AF-S DX, 50mm f/1.8 AF D, 18-70mm f/3.5-4.5G ED-IF AF-S DX, 70-300mm f/4.5-5.6G IF-ED AF-S VR, SB800 flash
My photos on SmugMug

That may not be as true as you think. To quote from the Northlight Images 5Dmk2 /3D/ 7D rumours page

Canon has several programmes of upgrades it wants to pursue:


• Upgrading more of the wide primes (as per the 14mm L Mark II) Many of these are relatively old designs. These sell to FF professionals and will need upgrade as sensors climb into the 25MP+ range...

Maybe Nikonites aren't so badly off after all!

With your excellent lens tests continuing (I hope) in the future you'll have to bend Nikon's arm and see if they'll offer you a long term D3 "loaner".

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

@Phil: The 5MP resolution of the D3 on an APS-C-"crop" just shows how "bad" the comparable resolution of the Nikon D3-FX-senor is in comparison to the D300 12MP sensor! So linear resolution of the D300 is 54% better than of the D3-Sensor but only can "measure" the inner circle of the lens. On the outer circle you have to be content with the limited res of the FX-sensor...

@Bob: Why? Canon already has a 16MPix FF-sensor (+15% linear res against the D3) and will have a 21MPix FF-sensor soon (+32% lin res vs D3). So with this "test-tool" you can judge corner sharpness of a Canon FF-lens 32% better than you can judge the corner-sharpness of a Nikon-FF-lens.


I wish! But again: the D3 is not good enough to test lenses with! Klaus Schroiff confirmed that in a PM to me.

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

My mistake, I misread your post and thought you were bemoaning a potential lack of lenses good enough for a super high resolution sensor rather than the lack of such a full frame sensor to test the lenses.

Another snippet from the Northlight Images page that attracted my attention was

DIGIC IV uses less power, and is faster than the dual DIGIC 3 found in the 1D3 and 1Ds3. In particular it can support 16bit raw and the larger 40-50MP sensors under development. With the larger sensors, it also supports pixel-binning where pixels are aggregated to deliver low noise and higher ISO performance at lower resolutions.

It sounds like a tacit admission that while 50MP full frame sensors may well become available, if you want a really clean image you will still have to sacrifice resolution, presumably by a factor of two both vertically and horizontally. I assume that pixel binning inside the camera may offer a better result than performing the same operation on a RAW file. Maybe Nikon are doing a smart thing by not pushing the pixel count as fast as Canon are apparently planning to do?

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

Why should this be? You've got 2 (4) different signals from 2 (4) adjacent photocells. What can the in-camera electronics do, that a RAW-converter cannot do better?

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

I'm afraid I can't quote a reference but I am pretty sure I have read somewhere that the in-camera electronics have an edge when binning due to better handling of readout noise. Presumably there would have to be some additional support for DIGIC IV pixel binning from the sensor but any further speculation about this from me would be a perfect way to reduce the S/N ratio to unity as I cheerfully admit I am getting out of my depth.

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

Another thought on this topic: If some lenses go to 3000 LW/PH resolution (which is about 1.4x todays 10MPix-sensor's linear resolution) and you want do digitize their image faithfully, the sensor should have double that resolution (Shannon's sampling theorem) linear. That means, you need a (1.4x2)square x 10Mpix = 80MPix sensor for a lossless A/D-conversion of such a lens.
And I'm discarding that today's sensors only have b/w-resolution as quoted but are 1/2 behind in linear color resolution due to the bayer-filter: that would mean to have a 320MPix (!) sensor if you want to capture the color details lossless.
Well, in this light a 1/2.5" 10MPix sensor is quite enticing as the pixel-pitch of such a sensor would make possible a 360MPix FF-sensor. Such a sensor would be totally transparent to todays best lenses.

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

Alternatively, one could mount such a 1/2.5" 10MPix sensor behind a substantially modified lens based IS system capable of moving the image from all parts of the rectangle representing a full frame of 35mm film so as to fall on the sensor. One press of the shutter release would then trigger multiple exposures which could be stitched together using the camera's on-board processor.

Naturally, it would take a little time to take the sequence of shots to be assembled together so one would also need a third generation in-body stabilisation system which could compensate totally for any camera shake for the second or two it would take to capture the complete image.

Any portrait shots would have to be taken with the subject braced by a rigid frame invisible to the camera but that technology was mastered back in the days of daguerreotype.

Unfortunately, nature photography would have to be restricted to windless days and stuffed animals.

Seriously (but not for too long), if I have followed your argument correctly I think you have demonstrated that, short of a revolution in sensor design, current technology means we are unlikely to see DSLR sensors capable of outperforming today's best lenses. The gains in linear resolution would result in so much picture noise at normal ISO settings that the professionals at whom such a sensor might be aimed would not be interested.

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

Your conclusion is correct: I just tried to make a point against people saying that todays DSLR sensors already are limited by todays lenses.
This was inspired by Ken Rockwell's ode to the "Full-frame advantage", where he did some serious lens-bashing.
Clearly with a relatively lower res FF-sensor (like in the Canon 5D or the Nikon D3), you will not see so much lens-problems as on an similar res APS-C sensor. But that really doesn't do justice to the lens-quality.
This is not to say that FF has no advantages. But certainly it's not a good way to measure lens-performance...

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

I thought that was a really good article by Ken there - good reading for anyone who wonders what the point of full-frame is...

Yeah, but comparing cheap lenses on a 12MPix FF-body to expensive lenses on a 12MPix APS-C-Body with 50% more linear resolution is not an apples by apples comparison!
Ken also exagerates in his statements because looking at his (very informative) test-shots also shows that some of the DX-lenses could hold up quite well...
Oh, and by the way: To get the same teleshot you need a lens with 1.5x the focal-length, which at the same aperture surely costs 3x the price of the perfectly adaquate lens for the D300. E.g. like using the (VR 70-)300mm/f5.6 for 462€ (we are still waiting for your full review here, Gordon!) on the D80-300 vs the need to use at least the (VR 80-)400mm/F5.6 for 1371€, not to speak of any other Nikon-gear reaching 400mm and further

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

Choosing products to compare is always a challenge and open to criticism - I know the cropped sensor cameras were placing greater demands on the lenses, but that's the point. I think it IS fair to compare them against full-frame bodies if you want to illustrate the benefits of the latter.

Many people assume that two cameras with the same resolution will perform similarly, especially at their lowest ISO and under good light, so articles like these illustrate that's not the case.

Anyway, if you compare one apple against another very similar apple, there'll be no difference to talk about! So more apples vs oranges please! And maybe a kiwi fruit too!

As for the 70-300mm VR review, patience, patience!

Hi Gordon. I didn't get my point across clearly enough!
I'm not complaining about comparing FF against APS-C! I'm only complaining about the use of a 12MP-FF body for lens-testing and comparing the results with a 12MP-APS-C-body for the same or another lens. This is totally inadequate as suddenly you're happy with color-fringing and sharpness on a sensor that has 1.5 times less linear resolution today.
Then you go out and buy this lens, mount it to the brand spanking new 25MP FX-body to come in 2008/9 and are totally disappointed, because suddenly all the ugly color fringing and corner-sharpness problems reappear, that the 12MP-FX-sensor simply could not "see"!
E.g. I'm pretty sure that the Sigma 12-24mm makes for quite a nice appearance on the D3, a lens that Ken even refused to test at all in his ultra-wide review - what a shame! Now Ken is telling us that all the second grade glass looks good on FF/FX But this is only true until the 25MP-FX-sensor comes out...
This is what I'd call an unwise advice
As lenses should be bought with future-proof in mind...

_________________
Thomas (beware: Nikon-fanboy and moderator!) My Lens Reviews, My Pictures, My Photography Blog
D800+assorted lenses

Trouble is, unless you have an optical bench for testing lenses without a body, you're kinda reliant on the manufacturer figures for guessing how future-proofed you might be!