6.21.2021

The Clash of Titanic Lens Design Philosophies. Modern versus the quest for perfection.

this is a Lastolite 4 x 6 foot diffusion panel. I use it a lot in the studio.
Right now, during the heat wave, I'm using it to block the strong afternoon
 sun from a direct assault on my living room. It seems like the right tool for the job today.

All lenses are baskets of compromise. It seems that every generation of lens designer is concerned with a different set of variables and outcomes than proceeding designers and understanding this is key to becoming immune to what are almost certainly transient styles in the look given by lenses rather than a sign of radical improvement in the overall quality of them as products. 

I noticed, for example, that with the current big, heavy, fast primes, that the primary objectives seem to be the delivery of lenses with sharpness that extends further across the frames and into the corners, at full aperture, when compared with previous offerings.

By making lenses which feature bigger and bigger maximum apertures, with flat fields, buyers are accepting the trade-offs of having lenses that are much bigger and much heavier than the smaller, slower lenses which were more normal in the past. It's ironic since camera buyers of today are demanding smaller and lighter  cameras than ever before; almost as if a rush to zero weight is somehow a benefit, but one profoundly offset by their lens purchases.

If I compare sizes between f1.4 lenses for mirrorless cameras and equivalent lenses made for SLRs from the 1980's and 1990's it's easy to see that sizes have increased by a lot. I own a handful of manually focusing 50mm f1.4 lenses, acquired over decades, and it's only in the past ten years that they've ballooned up huge, and also graduated to a much higher weight class. Of course, some of this is the result of adding robust auto focusing mechanisms to the newer lenses but it doesn't tell the whole story. 

Older lenses, according to literature I've read, were knowingly designed to be very sharp in the central region, adequately sharp in the middle regions of the frame and allowed, by design parameter, to become less sharp in the corners; at least when used at maximum apertures or close to maximum aperture. 

Knowledgeable photographers who actually needed "flat field" lenses knowingly bought macro lenses which were corrected for flat fields if they needed to shoot flat subjects such as paintings or the sides of houses. But by allowing for less sharp corners didn't mean a lens was less capable. If a well designed lens from the end days of the last century needs to be used to photograph a flat subject much of the time stopping down to f5.6 or f8.0 will render the entire frame with high sharpness. Even the corners. The positive trade off for older designs versus the latest designs is that the previous generation of lenses was constructed to be used in the field; they are small and light enough to be easily transported and more discreet in operation. Having to stop down for so incidences was part of an acknowledged bargain. 

The new generation of large, prime lenses with high speed apertures has ushered in lenses that use 67-82 mm filters and are, as in the case of the Panasonic S-Pro 50mm f1.4, about overall the volume of my Canon, Nikon and Contax Zeiss 50mm f1.4 lenses combined. And the newer lens is actually heavier than all three of them taken together. Yes, it is razor sharp at f1.4, all the way into the corners. Do I shoot it in that mode? Hardly ever. 

As I look at current lens measurements and compare them with older lenses of the same overall parameters (focal length+speed) I note that while the older lenses have sharpness fall off (mostly as a result of lens curvature) from center to corner they compensate by having less geometric distortion and much less vignetting. Since it was impossible to fix lens design compromises, after the fact, with film cameras (no software on film!) the older lenses were designed with the fewest image-oriented compromises. Today, a $5,000, 50mm f1.4 lens might have as much as three stops of vignetting at maximum aperture while a classic lens might only have 1.25%. Yes, on modern cameras, it's possible to correct for each compromise via in camera software but if you want to use older cameras or adapt the new lenses to different platforms you lose some of the advantages you are paying through the nose for. 

Many 50mm lenses of previous generations had simple barrel or pincushion distortion which is easily correctable in post processing while their modern successors have more complex distortion profiles (some called "mustache" distortion) which are much harder to correct in post processing software; outside the current cameras. 

As all of our lenses grow in size and weight, and escalate in price, the real question is whether or not they are visually as pleasant in the final rendering of an images as are the lenses we grew up with, pre-digital. I contend that as they technically became more and more "proficient" they've lost their individual character. Their optical fingerprints which made some lenses long time favorites. Keepers. Gems.

The real question is whether the trade-offs are working to your advantage. I am a portrait photographer. I'm happy to have a bit of sharpness fall offs in the corners of my images because my subjects are in the sweet spots of the imaging circle and the lowered sharpness of the corners is a pleasant way of further separating the sharper subjects from the less sharp backgrounds. 

You must also realize that along with the increases in pricing, etc. you are getting lenses that you are less and less likely to feel comfortable hauling around with you and practicing styles of photography that rely on your mobility and your ability to respond to scenes without too much pre-planning. And to do so inconspicuously. I question whether any but the most driven camera technicians are comfortable hauling around gargantuan lenses while mostly using only a fraction of their potential performance. 

I don't think that lens designers pursued the designs they've recently opted for in a vacuum. I don't think some head designer came in and told his team, "from now on we're emphasizing lens speed over package size." And then foisted the results on unsuspecting consumers. Rather, I think the constraints for the first three or four generations of widely accepted, interchangeable lens camera systems drove consumers to demand more speed. Why? Because, in case you forgot, the first ten years of popular digital camera featured sensors that were very, very noisy at any but their base ISOs. To compensate for lots of icky color noise at ISO 400 or 800 camera makers pressed their teams to make faster and faster lenses. Since camera users wanted cleaner files they went along for the ride. 

It's all pretty ironic since photographers seemed to be able to handle photographing with slower film emulsions and less over-designed lenses in the last years of the last century. Most of the faster lenses in those days were designed to overcome the viewing and focusing limitations of earlier optical viewfinder cameras. As light dropped it became harder and harder to focus on the focusing screens of the time so users bought the faster lenses in order to put more light onto the screens and to focus at those brighter, maximum apertures and then stop down, or let the cameras automatically stop down, to a smaller shooting aperture that made for better overall picture quality. The limited depth of field as a result of focusing wide open helped accentuate what was in focus and what was not.

An interesting question would be whether or not the faster current lenses, with wide open apertures, drove the pictorial style of using very shallow depth of field or if that style was always there, in potential, but only realized as a result of lens "progress." 

Now that the race for less noisy digital camera sensors seems to have been won by new sensor tech from Sony and others it will be interesting to see if there is a reverse trend back to more compact and size appropriate lens development. The Sigma i Series of f2.0 prime, Contemporary lenses sure points to this direction --- and is being quickly copied by Sony...and others. 

Another point of interest for me is whether the ability to put lens profiles into camera software is driving lens design or not. I think of the choices Leica seems to be making with their SL lens line. They seem to be designing for flat field and high resolution. The files I see from their systems are both contrasty and highly detailed but I wonder how much of the mix is driven by the in camera or lens-delivered processing software which optimizes the final, visual performance of the lenses. 

It's hard to design for both high contrast and high resolution simultaneously but while resolution is mostly determined by the physical properties of the lens design and construction while contrast, and even micro contrast, can be augmented with really good software or firmware design. Contrast can be ramped according to taste and branded, engineered looks. The higher the basic resolution the more micro-contrast can be tweaked in camera. But in the absence of an in-camera or post "look" or "lens profile" do the lenses still deliver the goods? I haven't tried it with a contemporary Leica lens but it would be interesting to use one on a camera, in a raw format, with no profile applied. I would conjecture that the lenses would still deliver high resolution and corner sharpness but the color and contrast would be different and perhaps less distinct from very similar, competitors' products. 

I would also presume that while the sharpness and detail would be maintained that in the absence of the embedded profiles the images straight out of camera would show higher degrees of both distortion and vignetting. In wider lenses to an even greater degree. 

Which brings me back to why I tumbled down this rabbit hole in the first place. Last week I'd taken the Sigma 35mm Art lens for L-mount out on a photo walk and worked at what I consider to be its most favorable aperture (if you are looking for the least vignetting, the highest sharpness, the least computational correction, etc.) which most published tests indicate is around f4.0. The photographs were uniformly great. Sharp, contrasty and with very nice, saturated colors. 

I did the same basic routine yesterday with the much more traditional, older, much smaller and lighter, and one stop slower, Zeiss ZF 35mm f2.0. While each lens delivered a different "look" the technical performance results were amazingly similar. They presented files that are equally sharp, and equally detailed. Even at 200% there are scant differences in the amounts of detail rendered. One defining difference was the more airy and open feel of the Zeiss lens. Not a significant difference but enough to make me realize that even in this age of being able to measure everything there are different "characters" associated with different lenses.

I'm sure if I had tried each lens at f2.0 the Zeiss would not resolve corners as well. And, as I got closer and closer to a flat surface subject I'm sure the gradation from sharp to soft; center-to-corner would grow demonstrably worse. The Sigma would outperform in that test. But does that make it the uncontested winner if your use case is to carry a 35mm lens with you everywhere and to shoot mostly at f4.0 and f5.6? Would the nearly twice as heavy, and 50% bigger, Sigma still hold your affections to the same degree?

And that, my friends, is lens design compromise in a nutshell. 

The most glaring examples of over-design versus pleasant handling and look trade-offs come via a comparison of two products of close to the same focal length from one Company. Sigma has a 40mm f1.4 Art lens that's enormous and brutally heavy. I venture to say that a person working, handheld, with one of these would toss in the towel after half an hour of carrying it around. Overwrought is the word that comes to mind...

In contrast Sigma also makes a 45mm f2.8 lens that's small, light weight and a joy to carry around. The use cases are phenomenally different but is one lens "better" than the other? And how does the potential of the camera and the lens profiles factor in? I voted, rationally, with my credit card, I have two of the 45mm f2.8 lenses and zero of the giant 40mm f1.4 Art lenses. While there might be ten or fifteen times in a year that I could even make use of a 40mm, shot wide open, there are daily cases when I can and do happily use a 45mm f2.8 lens stopped down to f5.6. I've never given any consideration to spending a thousand dollars more, or getting a gym membership so I can train to power lift a lens. 

The compromises in lens design are partly (mostly?) a result of consumers clamoring for what they think they want in order to solve a technical problem in the moment. The longer term issue comes with the amount of time it takes to design and deliver a new style of product to market. In the case of lenses many of the high speed lenses hit the market only after the sensor noise issues were largely solved. 

I personally like lenses with high center sharpness and I'm always pretty happy to trade extravagant resolution and flat field for more image contrast and less native distortion. And I'm even happier to have lenses that need to be physically focused because they don't contribute to battery drain or laziness. But an even higher priority is a lens that's easy and fun to carry.

Let's tackle image stabilization next time. Thanks for reading,  Kirk
Blanton Museum. Sculpture in the upstairs gallery. 
S1R in monochrome. 

Ellsworth Kelly Chapel at UT Austin.







 

15 comments:

Anonymous said...

How does your SR1 do with an R72 Infrared filter on it?

EdPledger said...

Break out your PenFT 40mm f1.4. Small, delightful…you have said so yourself. I would try it on something like a fuji xe2 or xe3. Just saying. Or if you have an LM to big ole L adapter….prob sharp in middle and weird in edges on FF.

Rich said...

similar contrast between my sig 35f2 and (adapted) tamron 35f1.4. 350g vs 900g. They are both sharp wide open! The Tam does pretty good AF w/ MC21. Bottom line for me: The speed of the Tam is its only advantage. But speed is great on the street at night! Weight and being conspicuous? = not so much

Kirk Tuck, Photographer/Writer said...

"How does your SR1 do with an R72 Infrared filter on it?"

How would I know that? Never, ever done infrared photography with any camera. I'd probably use a YF-12A filter on it instead. Or just default to some preset IR filter in post. But....so not interested in that.

ODL Designs said...

I don't think we can underestimate the effect of modern online lens testing and comprisons, increasing megapixels and digital photography in the drive for "perfect" lenses.

What I find interesting is how so many are both chasing perfect lenses, then processing for a lofi feel, adding grain, light leaks tweaking colour etc. The most perfect lenses to post 1920x1920 images on instagram... Or just in case they every get around to printing and putting on a show :)

While I have both the f1.2 and f1.8 lenses from Olympus, the f1.8 set get the most work. I have been cycling and working from the road a lot, and I carry a camera and a few lenses with me, the 12-45 f4, 25mm f1.8, 30mm f3.5macro and 40-150 f4-5.6. Together they probably weigh less than many of the newest performance beating lenses and give me a lot of fun shooting while i am out.

I think part of the push for perfection is the need to move the bar for sales. How do you convince photographers they need the next greatest thing if it really isn't greater?

Andrea Bellelli said...

Part of the reason for cumbersome lenses with very large apertura is that It Is trendy to have extreme "bokeh". In the film days (but for me even today) web accepted that it was food to have a portrait in which the subject was in focus and the background was out of focus but recognizable. Today it seems fashionable that the background is "creamy" and you obtain this only with very large apertures.

Michael Matthews said...

It could be that the desire for shallow depth of field comes from cinema. We’ve been immersed in the imagery almost since birth. The subtle focus shift between two people talking…almost in the same plane of focus, but not quite…a foreground object in closeup, all else dissolving into mist. It’s what we expect to see on screen.

As to flat field with edge-to-edge sharpness, I really wish I hadn’t sold my Olympus 60mm f/2.8 macro lens. I ditched it because it became clear I was never going to get into macro photography. Later, looking back over my Lightroom catalog at landscape shots I was amazed to find that all of the best were made using the 60mm. My original use, photographing artwork for reproduction, no longer applied so I punted off to Ebay what was probably the best lens I’ve ever owned. Gah!

J Williams said...

I very much agree with your thoughts on modern fast, huge and expensive lenses. I've owned exactly one large, fast, heavy and expensive prime in my life, the Canon RF 50 1.2. Other than that all have been rather mundane aperture instruments of reasonable size and cost. Somehow I managed to take decent and sometimes rather good photos with them knowing they had soft corners wide open, but I learned their weaknesses and also where they worked well.

In a moment of weakness after reading one too many reviews of the RF 50 1.2 I accidentally wondered over the the Canon site and checked the refurb section and they had a really good sale on the refurbished lens and I just couldn't resist. I have found it to be a fantastic lens that I have to admit allows me some freedom aperture wise that my more pedestrian lenses lack. I'd probably love it just as much if the corners were soft at 1.2 and it weighed half as much.

Is it possible to produce such a lens with all of its optical goodness and only sacrifice corner sharpness wide open while relieving some of the size/weight penalty? I'm not sure but even so I think marketing would preclude producing such a lens as the ultimate sale price would take a hit because of all the complaints about the corer softness wide open. "Such great potential squandered because of the soft corners wide open" would probably be a typical review.



Rene said...

Thank you, Kirk, for an excellent tutorial on lenses and lens design. I love when you write this kind of post as I learn so much. Looking forward to the image stabilization post.

Anonymous said...

Travel, especially intercontinental weight restrictions in monkey class, make the modern mastodont lenses (of any kind) rather limited value. If you have to travel on Middle East Airlines, weight limit is 7 kg (some airlines in all classes), and for domestic flights inside Brazil and China, government restrictions is 5kg per person (not always checked, be
ut when it is, you'll hear a lot of howls and begging and... ;-) So no modern high end cameras seen there anytime soon - even where travel is allowed for the lucky few (1.4 billion Chinese citizens have ample opportunity to travel inside china).

I haven't bought a new camera since 2017 (and that was a "new" LX100 for travel use), and no lenses since 2016. Both cameras and lenses started growing around that time, and the current crop of mirrorless "giants", made me decide to buy a lowly EOS 90D and reuse my old Canon glass again, now that it counts as "tiny" compared to the new crop of gear. The Canon gear (then an EOS 40D) was sent out to pasture by the arrival of the Panasonic GH1 (including microphone), that served me well until it was replaced by GH2, then by... but soon Panasonic decided on trying to get a foothold in the "luxury sector" (secondarily the available infidels in the pro sector ;-)

The EOS 90D will last my years, but will not be used for travel, but my selection of APS-C lenses really deserved to come into use again. They are delivering, and my, oh my, how wonderfull it is, to work with USM/HSM lenses and Phase detection focus instead of Panasonic DPD Contrast focus. One FUMP, and that's it, instead of wacky-bumpety-thunk-wirrrr-bump, in some cases ;-)

Modern lenses may be a tick better, but... they seem both larger, heavier and... paying real money for even more strained muscles at the end of the day is not my wish anymore.

But if Panasonic would produce a successor to the GX80 (the GX9 was not really an alternative), with decent sensor, phase detection focus, AAAAND both microphone input and headphone output, I could be persuaded to buy a u43 camera again to use my nice arsenal of Olympus and Panasonic lenses.

Otherwise.... I'll have to make do, with what I have, and a combination of iPhone 12 Pro and Panasonic LX100 (for available light photo/video), when traveling. Maybe the "gigantosaurus trend" will cease, when Japanese company accountants are barred from deciding on product design ;-) This would be a first, but... smile... dream... wish... sigh!

atmtx said...

Thanks, Kirk. A great writeup on the compromises of product development.

Anonymous said...

Nice photos. I really like your square format eye

Mark

Unknown said...

I notice the photos you added to this post all have a film grain effect. I like it. Would you care to comment on how you added it and why?

Kirk Tuck, Photographer/Writer said...

Hi Unknown,

I used the Monochrome D. color profile in the camera, turned the noise reduction slider in the camera way down (minus 3 clicks) and used the grain feature in camera as well. Small grain size. Medium effect. Added some contrast in post and bumped up the shadow slider a bit in Lightroom.

Thanks. I like the effect as well.

jp41 said...

Kirk,

The look of this set of black and white architectural images you've posted and Icon 3d-printed buildings have been bouncing around in my mind. The texture and shape of 3d-printed walls would provide an array of interesting shadow and depth. Icon is located in Austin.

https://www.buildshownetwork.com/blogs/3d_concrete_printed_homes
https://www.iconbuild.com/updates/new-story-and-icon-unveil-the-first-permitted-3d-printed-home