Why No Two Lenses Are Ever the Same

When I say no two lenses are the same, I do not mean that various models have differences between them (though they do, of course). Rather, I mean that every individual lens — even those of the same model — differs from the next, and it is well worth understanding the implications of that to make sure your lenses are of the highest quality possible.

Coming to you from Michael The Maven, this great video talks about a rarely discussed issue involving lenses: how individual lenses within the same product line can vary in quality. The issue comes down to manufacturing tolerances. Simply put, no manufacturing process is perfectly precise, and as such, individual copies of lenses may vary slightly in their quality. Manufacturers put limits on how much variation is allowed in any one copy, however. This is why you will sometimes hear photographers talk about having a "sharp copy" of a lens. Nonetheless, whenever you purchase a new lens, it is important to not just assume that your copy is perfect; be sure to take time to carefully examine shots you take with it to make sure it is performing to the standards you expect of it before the return window closes. Check out the video above to learn more.

Log in or register to post comments

20 Comments

Ivan Lantsov's picture

bad robot make bad lens?

Rod Kestel's picture

I dunno, this seems a bit too clever. A reputable manufacturer will set tolerances to maintain quality. Sure, there'll be minor differences, but if you bought a lens from them that 'failed'... they're not reputable.

If anybody ever rejected a lens, how bad was it? And how much do lenses age (but not abused)?

Sound advise in Michaels article as "sample variation" is a very real, little known and frustrating thing.

I figured I could save as much as a grand by picking up the older Nikon PC 85 tilt shift lens when the "E" version came out. I rejected 2 and ended up buying the new version after all.

Teo Lab's picture

Sample variation is sufficiently poor with some lenses that to me it completely changes the value of a lens (ie the amount of money that I would be ready to spend on it), and even worse the value hierarchy that I'd establish between different lenses (good copy of lens A > average copy of lens B > bad copy of lens A).

My first copy of Olympus' 12-40mm f2.8 was softer than the kit zoom for the whole right third of the frame at all focal lengths, while the second one was much sharper everywhere. I failed to find among seven samples a copy of the 25mm 1.8 which focal plane wasn't titled and three or four of them were atrociously bad (if I focused at 10m on the right hand side, the left hand side would be focused around 2-3m).

It can also affect out of focus areas : https://www.dpreview.com/forums/thread/4415177#forum-post-62954181
In regards to the link above, if indeed that is a question of manufacturing issues (both hypothesis raised in the thread would suggest that it is), I'd be ready to spend €1000 on a hypothetical copy of that Nikon completely devoid of that problem, €250-300 on a copy with a mild version of that problem (most sample galleries seem to have a mild version of it) - although in that case I'd rather shoot with the Sigma 45mm if it were available on that mount -, and €0 on a copy with a severe version of that problem (I just don't want it in that case).

Lensrentals has extensively published articles on these issues, I encourage you to read their blog.

To test lenses, however, I do not recommend at all the methodology proposed in the video, it doesn't control variables well enough. There are various methodologies presented on the web (siemens chart, upside down test, etc.) that are better but operator error and other variables are difficult to eliminate to a sufficiently good enough degree. The rule of thumb is : always question your methodology first before questioning the lens.

Kirk Darling's picture

"The rule of thumb is : always question your methodology first before questioning the lens."

For sure. I don't know how many times in recent years I've read someone complaining about lens quality and then he says, "I went by the shutter speed/focal length rule, so I know it's not blurry from hand holding it."

Teo Lab's picture

I agree. For some of my lenses I've tried to avoid reaching a conclusion as I never found my tests' results particularly logical. I've tried to compare two copies of a Canon 100mm f2.8 L IS USM but, for example, the upside down test yielded incoherent results and I experienced some shot to shot inconsistencies that I couldn't solve.

But I also think that the reverse is true. I believe that some people carry in their bags pretty rubbish copies and, being unaware of sample variation, blame anything but the manufacturing when they see something that's not quite right in their pictures, or pass judgement on (or even review) a lens in these conditions.

EL PIC's picture

If it’s bad send it back ! I always do a test or two with any new cameras and or lens.
Could be one or more problems BUT .. it is rare and it’s usually GearHead OverReaction ..
You will see more problems with high res cameras .. you don’t need such cameras pixel peepers.
Don’t knock your head against the wall with Micro Adjust. Micro Adjust was invented to correct for Extreme Thermal Effects not a bad lens or GearHead OverReaction !!

Rumor/myth has it that Stanley Kubrick would order 10 or so copies of a lens, test them all, and go with the one he liked best. So clearly, he thought there were variations. And these were high-end cine lenses.

That's true, Joe Dunton worked closely (owner of one the biggest rental houses in the UK) with Stanley Kubrick & mentions that in this very interesting video on Kubricks lenses:https://youtu.be/fb7Meqaz7Aw

Ryan Davis's picture

My first "L" lens was a 24-70. It was good- very good, but after picking up my next L lens, a 70-200, I was floored by the quality of the images produced. I thought maybe it was that the models are different, but just to check I borrowed somebody else's 24-70, and it was slightly better. Don't get me wrong, mine was still good, it was just on the low end of the L level of quality. And my 70-200 is on the high end. This issue is definitely real.

I've seen distinct variation in new off the shelf Canon, Nikon,Sigma and Leica over the years and it baffles me any buyer would preorder an unproven expensive optic knowing "sample variation" exists.

Ansel Spear's picture

What utter Bollocks!

Wes Perry's picture

Quite accurate actually. My first Sony 35 1.4 was strongly decentered. My second is only moderately decentered. I know there are better copies out there, but apparently both lenses were within the manufacturer’s tolerances.

Ansel Spear's picture

Well that's down to appaling QC. Every lens is supposed to be produced to fall within a set of fine tolerances. Any variation within these are deemed to be barely noticeable - and acceptable to even the most demanding of photographers.

If it's a hit and miss process that i) the lenses frequently fall outside these tolerances and ii) it's subsequently not picked up in QC, then something is seriously amiss with both the manufacturing process and staff training.

Wes Perry's picture

I sent my lens back to Sony and they tested it and said it was within manufacturing tolerances. They then sent me another that they tested and found to be “superior” in performance, and it was also decentered. Apparently those tolerances are much wider than people realize.

Derek John's picture

Mmm, while I'm sure there is sample variation, in a competitive field such as photography it is negligible. What is also probable is that in countries where there is the facility to exchange such products easily the variation is probably higher, mainly due to the fact that the exchange is possible.
Another factor in the variation is in the loose nut behind the lens. Sometimes I have sample variation on the same lens mainly due to my not adhering to the rules.

I currently own 19 lenses, have been through many Nikon/Fujifilm/Canon and Olympus lenses in my time and never had cause to return any of them in nearly 30 years other than that they may not have suited me or having fell out of favour....some of them are 40 plus years old and are still fine.

Operator error aside-let’s say a retailer has 4 of the exact same legacy lens in his stock inventory-testing them (policy permitting) will give a person (assuming aspiring test pilot possesses ability to discern any differences) clue's as to manufacturer sample variation between the exact same optics (or not).

Recently the frustrated and picky private seller of a Zeiss Loxia 21mm I tested pressed me to quantify me declining purchasing his (from all outside inspection) pristine lens that he picked up brand spanking new. I needed the Loxia for astro and my criteria was/is that it not be decentered and sharp across the entire frame wide open at F2.8.

Same said owner doesn’t shoot astro and revealed he had never critically shot this lens wide open-seller apologized after sending him samples illustrating lens decentering and the man's frustration is now squarely aimed at manufacturers spitting out high end product with ambiguous manufacturing tolerances that end up in the laps of unsuspecting buyers.

Derek John's picture

Interesting. David how do you go about testing the lens? Do you take an image of a specific subject or a chart? Flat surface, brick wall? Did you manage to get a better copy of the lens?

Sourced a sound used example.

Accurate article. I've had two L lenses that were 'soft' out of the box and swapped. Then again, my latest 5d mkiv needed microadjustment with my 70-200 which was perfect on other cameras.