Landscape lens odyssey part VI - Optical design & Engineering

The 6th post in a series looking at how I arrived at a stable of lenses for my landscape photography as a Canon 1Ds3 owner and dedicated tripod user. 

I had been contemplating an upgrade from the Canon 70-200/4 to the the IS version, which everyone online raves about, even though the MTF charts don't suggest a radical improvement in areas where the non IS version is found wanting. OTOH the IS version may be a stellar performer in the 80-190mm region, if so, Canon might wish to reconsider presenting their optical data similar to the way that Leica does, as the 70-200mm f4 is a much better lens away from the ends of the zoom range.

20 lens elements in 15 groups

10 and 30 lp/mm % contrast curves  blue colour f8- solid lines sagittal,  dashed - meridional. Canon MTF charts for 70-200mm f4 IS

There were a few Leica R lenses that interested me, but a used Vario Elmar 80-200mm f4 was the one that I purchased first to test the water, mostly because I was looking for better performance from a tele zoom and could make a direct comparison to my copy of Canon's 70-200mm f4 in a woodland setting, where there would be 'high spatial frequency detail' over most of the image. There were also theoretical MTF measurements from the manufacturers and actual MTF measurements from tests at  www.photodo.com.

16 lens elements in 13 groups

10 and 30 lp/mm % contrast curves  blue colour f8- solid lines sagittal,  dashed - meridional. Canon MTF charts for 70-200mm f4

10, 20 and 40 lp/mm % contrast curves solid - sagittal,  dashed - meridional www.photodo.com MTF chart for Canon 70-200mm f4

10, 20 and 40 lp/mm % contrast curves solid - sagittal,  dashed - meridional www.photodo.com MTF chart for Leica 80-200mm f4

Leica tech data Vario Elmar 80-200mm f4 @ 80 mm

Leica tech data Vario Elmar 80-200mm f4 @ 100 mm

Leica tech data Vario Elmar 80-200mm f4 @ 135 mm

Leica tech data Vario Elmar 80-200mm f4 @ 180 mm

Leica tech data Vario Elmar 80-200mm f4 @ 200 mm

12 lens elements in 8 groups 

Testing of zooms, at stopped down apertures, at infinity, on scenery with randomly orientated 'high frequency spatial detail' in my opinion needs a wooded valley of some width and a variety of views across and along the valley axis with sky lined branches / horizon. On my doorstep, the topography of the Forest of Dean and Wye Valley fortunately has a few elevated locations for telephoto zoom lens testing, which is handy when the infinity focus at 200mm and f8 is ~140m. There were also elevated view points inside the woods where image comparisons at the wider end of the zooms could be made. Testing setup was tripod positioned on hard ground, MLU, cable release and manual focus in live view @10x magnification across the image to ensure everything was in focus. I'll add that when testing lenses for resolution, wind can be a pain on subject matter and good air clarity.

The quality of both lenses needed Pixel peeping of test images to establish the resolution capability,   which subjectively determined that -

The R displayed minimal chromatic aberrations.

It was a close call for resolution in the dead centre of the image, but not at all focal lengths.

Sometimes the L lens, sometimes the R, away from the centre the R outresolved the L for fine detail across the image into the edges and corners.

I never noticed any effect of field curvature effects stopped down and live view focusing.

The lenses were also tested over some manually set white balance settings and in my opinion the Leica images had more vivid colour. The analogy I'd use would be to compare an image file in Adobe RGB improperly converted to the sRGB colour space and the colour takes a hit through a smaller gamut envelope.

I didn't test specifically for flare resistance between the two lenses, but subjectively I'd say the Canon has better flare resistance and the Leica has a built in hood for good reason, it's worth employing the hood at all times in all lighting.  Typing of hoods and there's a good reason Canon doesn't show the hood attached in any marketing information, it makes for a great weathervane in any wind ...

Build quality of the Leica R zoom was impressive, in a different league to the Canon L lens, as was the manual focus compared to the Canons, to be fair to the Canon though it can auto focus and has electronic aperture control.

The Leica is slightly physically smaller, but is heavier than the Canon (1.05 kg incl leitax adapter and caps vs 0.81 Kg incl hood and caps).

Leica R 80-200/4, Canon L 70-200/4 and lens hood

Some caveats to the testing.  The Canon was a hands down winner over the  70-80mm focal length. The L lens was purchased from new and has seen heavy use, I've no idea on the Leica zoom history of use. I didn't notice vignetting as being an issue for either lens and never checked for distortion, the woodland scenery was mature oak and beech, with downy and silver birch trees, not subject matter that lends itself to distortion analysis. Over 6 years of use, the Canon lens had seen some exposure to wet weather and dusty/pollen conditions and despite its reported environmental sealing limitations, the internal glass is clean and I've had no issues. I could find no information on the Leica environmental sealing. There's also the consideration of a lens mass produced for a photography market requiring autofocus capability vs a far lower production volume lens for a niche of manual focus photographers.

What I found interesting was how optical lens diagrams didn't correlate with actual optical performance. The 12 element 8 group optical design of the Leica R zoom and the 15/12 and 20/15  optical designs of the Canon L f4 zooms.  I had been led to believe additional lens elements groups were used to correct for higher aberrations, this wasn't the case from my subjective testing. Anyone researching Leica R lenses will end up reading the observations and insights of Erwin Puts, and it soon becomes apparent that he has an enquiring mind, has some knowledge of Leica's history, products and access to Leica's designers and engineers, more importantly he writes from the perspective of a photographer, not a marketing drone. The review of the 80-200mm f4 mentions six elements with anomalous dispersion glass and/or high refractive index and a total of eleven different glass types used.  The lens rendition of colour was also a surprise and not something I had expected, I've no idea why the colour was better - coatings, exotic glass types, better colour correction, fewer lens elements reducing the number of air/surface interfaces or maybe a combination of factors, oh and engineering tolerances?

Those who know about these matters state that manufacturing and assembly tolerances can have the greatest impact on image quality, from precision grinding of a lens surface, to alignment of lens elements inside the lens barrel, there will be an engineering tolerance build up. Which could create a Catch 22 scenario, where increasingly complex lens designs aimed at improving optical performance actually degrade image quality through engineering tolerance build up. Those lenses that pass quality assurance - inspection and testing - of a defined performance point should demonstrate some variability in performance courtesy of tolerance stacking. The excellent www.lensrentals.com blog has a few posts demonstrating the variability of lens performance.

http://www.lensrentals.com/blog/2013/09/there-is-no-perfect-lens

http://www.lensrentals.com/blog/2015/03/a-quick-zoom-variation-demonstration
http://www.lensrentals.com/blog/2015/06/measuring-lens-variance

And it seems optical performance variation also exists within 50mm f1.4 primes, the Zeiss Outus 55/1.4 variability might be due to a complex aspherical element ...
http://www.lensrentals.com/blog/2015/07/variation-measurement-for-50mm-slr-lenses

Landscape lens odyssey Part V - Divining MTF charts and cinematographers lens evaluation

The 5th post in a series looking at how I arrived at a stable of lenses for my landscape photography as a Canon 1Ds3 owner and dedicated tripod user. 

Canon produces a publication titled 'Lens Work' with a chapter on Optical Terminology and MTF Characteristics, which unusually for texts on optical matters is remarkably lucid in describing what a layman needs to know and one of the few publications that refers to a lens Colour Contribution Index. An ideal MTF chart illustrates the measurements of contrast of an actual lens optical performance under a test environment and can potentially indicate areas of the lens strengths and weaknesses. MTF charts provide no information on colour rendition, flare, distortion and build quality.

A little knowledge is dangerous, as I could appraise Canon's calculated mtf charts with images made by the lenses I owned and observe how the optical performance indicated by the 30 line pairs per millimetre (lp/mm) of the sagittal and meridional curves appeared to correlate with the resolution of fine detail in an image, with the difference between the sagittal and meridional curves indicating astigmatism and a nod to field curvature, astigmatism, chromatic aberration and an assumption my technique was adequate.

It was an exercise that provided an insight into correlating actual performance with theoretical, I'd be lying if I said, I didn't try and visualise what the image might look like if the 30 lp/mm curves were closer together, higher up the %contrast scale and flatter across the image. My eyes also suggested that my copy of the 17-40 was better away from the ends and not that bad focused in the corners and stopped down at 20mm, performance at the tele end was also visibly better when stopped down to f11, which for landscape photography is not a bad thing at all.

Every article I have read on divining mtf charts has a caveat that charts shouldn't be compared with other manufacturers nor across different focal length lenses. It appears that some manufacturers charts are not tests of actual lenses, but are calculated curves; some manufacturers calculated measurements, may be more theoretical than others and may or may not, include design tolerances, assembly tolerances, geometric losses etc. Some knowledgeable commentators hint that a lens could be designed to produce a respectable mtf characteristic that wouldn't necessarily be reflected in actual performance.  So ideally, it would be good if someone independent had the wit to test and measure real lenses and correlate manufacturers MTF charts with reality. It just so happens that a photography website once did just that http://www.photodo.com/browse-lenses/sort-mtf-score-desc and graded lenses to some criteria and also published the actual MTF charts of lenses tested by Hasselblad and the 10, 20 and 40 lpm spatial frequencies at full aperture and f8, at infinity focus. The downside was that the lens tests dated back to 1990's/early 2000's, but they did highlight, to my eyes, some potentially high performing lenses that could be adapted to the EF mount. Just an observation from analysing the photodo MTF's, but if the 40 lp/mm sagittal and meridional curves are above 60% contrast across the image width at f8, there's a high probability the lens is highly regarded in the photography community.

Leica's MTF chart for the Macro Elmarit R 60mm f2.8

Back in late 2011, despite Canon's extensive lens line up, the future in my eyes suggested more image stabilised lenses, faster autofocus motors and a suspicion Canon was focusing on video cameras and cine lenses. So, I decided to look elsewhere for lenses to meet my desired angle of view and optical performance. For a few years EF mount landscape photographers had migrated to Carl Zeiss ZE lenses, legacy 35mm and medium format film lenses that could be mounted on the EF mount by using adaptors. In my opinion there's an entrenched tribalism to photography brands and gear, which unfortunately manifests itself on photography gear forums, as highly subjective opinion based on pseudo science/delusion/personal agendas/prejudice and anecdotal evidence. There are some genuinely useful opinions, but it is a dispiriting search for pearls of wisdom

Leica's MTF chart for the APO Macro Elmarit R 100mm f2.8

Cinematographers were also starting to seriously test and evaluate legacy manual focus 35mm lenses for their needs, their incentive was driven by the price of cinematography lenses and the fact that 35mm DSLR's were incorporating video and increasing capability. The evaluation of legacy and current lenses by the cinematographer community was insightful and informative for its transparency and reliance on characterising a lens performance on actual images under varying subject and lighting scenarios. 

Leica's MTF chart for the Vario Elmarit R 28-90mm f2.8-4.5 @ 70mm

Based upon the manufacturers published MTF correlation with photodo's actual MTF testing and the insights gained from the cinematographer community evaluation, the surprising upshot was that Leica's R lenses would be the ones to further investigate.

Leica's MTF chart for the Vario Elmar 80-200mm f4 @ 135mm