Focusing ZEISS DSLR Lenses For Peak Performance PART ONE: The Challenges
Lloyd Chambers takes an in-depth look at the challenges in obtaining peak sharpness, revolving primarily around focus, using ZEISS DSLR lenses.
by Lloyd Chambers, diglloyd.com
Decades ago, I recall shooting Air Force resolution targets in a basement to check for lens sharpness. Little did I realize just how error-prone such efforts were, with issues like film flatness and thickness, shutter vibration, calibration errors of the camera, etc. Today’s perfectly flat sensors and tiny photosites are even more unforgiving. And even with perfect focus, optical behaviors can result in out-of-focus results.
ZEISS DSLR lenses for Canon and Nikon are manual focus. Since the #1 issue in getting peak lens performance is accurate placement of the zone of sharp focus, manual focusing technique is critical, along with understanding lens behavior that can affect focus.
Part 1 covers the challenges in obtaining peak sharpness, revolving primarily around focus.
Part 2 provides specific tips and techniques for focusing challenges.
A few other topics are touched upon for context and relevancy, such as diffraction. This introductory video discusses some of what is to be covered.
Video: Introduction to focus and related issues
Since the ultimate goal of accurate focus is getting peak lens performance, a brief discussion of general lens performance is warranted. Setting aside the “bad sample” possibility, lens performance issues are frequently focus-related, with “focus” being taken in its full sense to include optical behaviors that can “pile on” such as focus shift and/or field curvature. Frustrations with lens performance tend to run along these lines:
- “The lens front/backfocuses. It must be the lens because I focused it perfectly.”
- “The corners [or mid zones] are soft.”
- “My images are not quite sharp even at f/5.6, so it must be the lens.”
- “I stopped down to f/16 and even then the lens is soft.”
- “I focused at infinity but it’s not sharp, so the lens is the problem.”
- “My quick test shows that Brand X is sharper than Brand Y” or “My quick test shows that Lens A is sharper than Lens B.”
This next video discusses lens performance in general, in order to provide context for the overall article. Focus is not the only issue, but focus is the key variable over which the photographer has control. The rest of this article speaks primarily to focus with ZEISS manual-focus lenses.
Video: Common frustrations with lens performance
Manual Focusing Challenges
It’s a high-resolution autofocus-oriented world today. Consider these stumbling blocks:
- Even high-end DSLR optical viewfinders are challenging for manual focusing accuracy. For starters, the focusing screens are ill-suited for manual focus and generally are no better with an f/1.4 or f/2 lens than an f/2.8 lens. In APS-C DSLRs, the “porthole” viewfinders make manual focusing extremely difficult.
- The smaller the sensor pixels, the more dramatic the sharpness loss caused by a focus error. A 50-megapixel sensor is less forgiving than a 24 megapixel sensor. The tiny photosites of a 24-megapixel APS-C sensor are less forgiving even than a 50MP full-frame sensor.
- Stopping down as much as 3 or 4 stops may be needed to compensate for relatively small focus errors, and even then the zone of sharp focus may be biased too far forward or back in a way that alters the visual impact.
- Fast lenses (f/1.4 and f/2) and lenses of medium and longer focal length are unforgiving of tiny focus errors at wider apertures. The better the lens performance, the more visible the error.
- Making crisp images with ZEISS manual-focus lenses requires spot-on focusing, but achieving precise focus is more nuanced than it might seem. The challenges discussed here in Part 1 set the stage for the Tips that follow in Part 2.
Challenge #1: an optical viewfinder (OVF) is error prone
In terms of critically sharp focus, the optical viewfinder of a DSLR is a hit and miss approach:
Alignment: there are two light paths and two sensors in a DSLR: the light path straight to the imaging sensor, and the twist and turns of the light path to the focusing sensor along the path to the optical viewfinder. If the two light paths are not exactly the same distance (more than 0.02 mm is a significant error), then focus will be skewed even if the user has perfect vision and focuses the image perfectly (as perceived). The camera manufacturer may be able to adjust the camera to fix such alignment issues.
Focusing screen: even with a perfectly adjusted viewfinder, the surface structure of modern focusing screens makes it difficult for the eye to distinguish between in-focus and out-of-focus, particularly off-center. In effect, the focusing screen doesn’t allow focus discrimination much better than f/2.8, so f/1.4 and f/2 lenses tend to be hit and miss for focus. Also, f/1.4 and f/2 lenses are generally not much brighter than f/2.8 lenses with these modern screens. The traditional split image and microprism focusing aids have disappeared from Canon and Nikon DSLRs. There may also be markings that overlay the screen and obstruct the view for manual focus, though some camera models have options to hide these markings.
Vision: the visual acuity of the photographer’s eye is another issue. I (Lloyd) have astigmatism in my right eye that makes it not good for focusing, so my left eye must be used for focusing. Fatigued eyes, dim light, eyeglasses, smudged viewfinder, or the diopter adjustment being off just a tad all degrade the ability to focus accurately and consistently over a range of conditions.
The bottom line is that it is awfully hard to focus a lens manually through the OVF to the precision required for peak lens performance, let alone do so consistently with a variety of subjects in varied lighting.
As shown below, all three attempts to focus by eye under ideal target and lighting conditions through the Nikon D810 optical viewfinder resulted in unacceptable focus.
Challenge #2: DSLR focus assist (“green dot”) is prone to error
Many DSLRs have a focus assist mechanism: a dot lights up when the camera determines that the image is in focus. There may also be directional indicators (Nikon). The focus assist feature is based on the autofocus sensor, which can have accuracy and precision issues: a common focusing frustration with autofocus is frontfocus (too close) or backfocus (too far). Frontfocus or backfocus thus apply to manual focus lenses when using focus assist. A manual-focus lens focused with camera focus assist is subject to similar frontfocus or backfocus issues as an autofocus lens. Whether auto or manual focus, focus shift can be present, further confusing matters (covered further on).
Focus assist has other challenges such as a limited number of sensor positions in fixed positions (some of which are less accurate than others). Sometimes there can even be a mismatch between the apparent and actual focus sensor position (very confusing). And once focused, recomposing shifts the plane of focus.
Below, a situation in which focusing is as easy at it gets: an ultra high performance lens on a high contrast brightly-lit flat target. The Nikon D810 focus assist (“green dot”) feature was used. Four variants are shown at f/1.4: (1) Optimal focus using Live View, (2) focusing near to far, (3) focusing far to near, (4) guesstimate focusing within the in-focus range indicated by focus assist.
None of the focus assist results seen below are optimal, though the guesstimate in-between result is close (blind squirrels do find acorns sometimes). Under real world field conditions results are likely to be much more variable (lower contrast, more ambiguous target, dim lighting, etc). A camera’s focus assist feature delivers variable results and is not viable for consistently critically sharp images.
It can be even trickier: the focus assist feature depends on the autofocus system, which on a DSLR typically “sees” only light rays corresponding to stopped-down rays (something like f/5.6 for Canon DSLRs). This is a big deal for lenses with spherical aberration and focus shift.
For a lens with focus shift, the AF system thus indicates focus way off the mark for wider apertures. Below, the Canon EOS 1Ds Mark III delivers a badly blurred image for the ZEISS Planar T* 1,4/85 (which has significant focus shift caused by spherical aberration). The camera autofocus system “sees” rays that are optimal for ~f/5.6, but the exposure is at f/1.4. The results with Focus Assist are mush, but with the lens set to the infinity focus, fine details are captured.
Challenge #3: the infinity focus mark may be wrong
ZEISS sets the infinity focus mark precisely, but cameras can vary slightly in flange-to-sensor distance (even 0.02mm is significant for a high-res sensor). Relying on the infinity focus mark or hard-stop for infinity focus is unwise, so it’s important to confirm whether a particular lens + camera body combination is accurate at the infinity mark—compare to a reference frame obtained using magnified Live View focus.
The camera variability point was driven home to me using a ZEISS wide angle lens a few years ago. Two Canon 5D Mark II camera bodies were used, with the ZEISS lens racked to its mechanical infinity stop. At f/2, one image was sharp but the other was slightly blurred. Why? The flange to sensor distance was slightly different for each camera. Thus the camera body is itself a variable that makes the infinity focus mark unreliable.
With some lenses with special glass types, infinity focus changes slightly with temperature so that there exists no fixed infinity focus, no hard-stop, as seen with the ZEISS Otus 1.4/85 focusing ring shown focused “past infinity”.
Video: Infinity Focus Can Vary
Challenge #4: focus distance varies across the frame (FIELD CURVATURE)
With most lenses, optimal focus distance varies across the frame. This is known as curvature of field (“field curvature”), and it is present to a varying degree in nearly all lenses. Field curvature should not be confused with optical distortion, which warps shapes and curves straight lines.
In general, f/1.4 lenses and wide angle lenses have the most field curvature; it is one of many optical tradeoffs. A lens with strictly controlled field curvature is termed a “flat field” lens, the ZEISS Makro-Planar T* 2/100 being an outstanding example. Field curvature is the rule not the exception, and it’s why lens comparisons can be so misleading: two lenses with differing field curvature intersect a subject to advantage or disadvantage, depending on how the curved zone of focus intersects the particular subject matter.
In a lens with significant field curvature, there exists no perfect focus for the frame in its entirety, only a best overall focus in some average sense. But when a key element needs to be in critically sharp focus, the choice is simple: focus on that element and whatever else happens, happens.
When field curvature is present, then when focus is made at center it is slightly out of focus for other areas of the frame (and vice versa). For example, ideal focus for the center might be at 2m, but focus might best at 1.95m in the mid-zones, 2.05m at the edges and 2.2m in the corners—it varies across the frame. This is why shooting a brick wall or similar planar subject in order to test a lens has limited value in understanding lens performance: sharpness may be extremely high across the frame but with some deviation outside the nominal plane of focus—so within the plane of the flat test surface, detail goes soft and thus misleads as to imaging potential for most subject (things other than that brick wall!).
Typically field curvature arcs forward, but sometimes it is balanced as a “wave”, arcing forward in the mid zones and inflecting rearward toward the edges/corners. Stopping down is the only real mitigation for field curvature—more depth of field. For a distant landscape scene, it is not uncommon for f/8 to be required for peak sharpness across the frame on a high-res DSLR.
Shown below, the ZEISS Distagon T* 2/28 ZF.2 has a field curvature that arcs roughly in the shape of the white line. Accounting for this behavior in the choice of focus can help optimize total image sharpness.
The rearward arc of field curvature can be seen cutting through the sand.
Challenge #5: focus can change aperture (FOCUS SHIFT)
Most of the ZEISS DSLR lens line has little or no focus shift, but the classic lens designs like the ZEISS Planar T* 1,4/50 and ZEISS Planar T* 1,4/85 have spherical aberration and hence significant focus shift (the high performance Milvus 1.4/50 and Milvus 1.4/85 designs eliminate focus shift as a concern).
The cause of focus shift is spherical aberration. With some wide angle lenses focus shift may be introduced as a “balancing” design tradeoff (using one aberration to balance another), particularly in the outer areas of the frame; the ZEISS Otus 1.4/28 applies this principle in the outer zones.
Focus shift can go either direction, and can sometimes be in opposite directions depending on the area of the frame—a tradeoff in the lens optical design. In the center of the frame focus shift is usually rearward (more distant focus). In the outer areas focus usually shifts closer, particularly with wide angle lenses.
Shown below are examples with the ZEISS Planar T* 1,4/50 at f/1.4 and f/5.6, obtained meticulously using a slider to alter camera position. The difference in optimal focus for f/1.4 vs f/5.6 amounts to a whopping 60mm (as marked on the ruler, shooting distance was about 130cm). The top row shows optimal focus for f/1.4, the bottom row shows optimal focus for f/5.6. View at full size to see the differences; they are substantial. Peak results for an f/5.6 exposure require significantly closer focus (at f/1.4) to account for the rearward focus shift stopped down.
The 4-way grid below shows the results of focusing focusing wide open at f/1.4, versus focusing optimally for f/5.6. Observe the sharpness on the eyes with both, but also the zone of focus on the ruler (open the image separately to view at full size).
Focus shift can be seen at work in this video using the ZEISS Planar T* 1,4/85 ZF.2.
Video: Focus shift with the ZEISS Planar T* 1,4/85
Focus shift can be graphed with an MTF chart as shown below for the center of the frame. Observe that at f/4, peak MTF of 80% is seen, but it is well off the zero point (the point of focus, focused at f/1.4). At the point of focus, MTF is 60% instead of 80%, the difference between a world-class lens and a mediocre lens at f/4. Moreover the entire zone of focus is shifted rearward, which throws off the balance of the image (e.g., bitingly sharp ears but with slightly blurred eyes in a portrait). Combined with a slight user focusing error, the image may be unacceptable.
Challenge #6: stopping down too far blurs details (DIFFRACTION)
One way to mask focus errors is to stop down, say to f/8 or f/11. However, a focus error also biases the entire zone of sharp focus too far forward or too far back, which has implications for visual impact (the juxtaposition of sharp vs unsharp diverts attention).
Stopping down too far brings with it the phenomenon of diffraction, which begins to degrade contrast and image detail starting around f/5.6 – f/8 for modern high-res DSLRs. When DSLRs increase in resolution increases to 60 megapixels or more, those figures drop to f/4 – f/5.6.
Practically speaking for camera resolution in 2016, use of f/11 degrades image sharpness mildly, f/16 degrades it strongly, f/22 is wantonly destructive, and f/32 is an unmitigated disaster. Shoot a series by aperture to see the effects, or see the Diffraction section in diglloyd Making Sharp Images as well as Mitigating Micro Contrast Losses from Diffraction Blur.
Observe in the pair of images below how the f/22 image has very low contrast with fine detail seriously damaged, caused by diffraction. When stopping down, some lenses also degrade in contrast over and above diffraction effects, due to internal reflections. Hence stopping well down to overcome focus errors has serious implications for image quality.
This video shows the effects of diffraction with stopping down, but due to limited resolution of the camera LCD, it is not an ideal demo. See the above still picture for the best sense of losses.