The answer for I was I1 & I3 = Zeiss. The inclusion of the 135/2 was supposed to make things a bit more interesting as its colors are actually more like a mix of Zeiss and Minolta rather than typical Canon.
denoir wrote:
The answer for I was I1 & I3 = Zeiss. The inclusion of the 135/2 was supposed to make things a bit more interesting as its colors are actually more like a mix of Zeiss and Minolta rather than typical Canon.
Ok thanks for putting the images back up. I think this is a wonderful thread to keep track of when folks claim there is no difference in IQ.
I got the "I" choices correct, and I agree that the 135/2 is more Zeiss like than other Canon lenses.
Mike K
Thank you Luka for completing such a detailed test. It was fun and the results interesting. It's clear that rendering differences can be discerned by people but perhaps it's possible that some of that could be due to the fact that:
Set A: Compared $400 Canon prime to a $1300 Zeiss prime
Set B: Compared a Zeiss prime to a Canon Zoom
Set C: Compared a Canon Zoom to a Zeiss Prime
Set D: Compared a $1300 Zeiss prime to a $600 Canon prime
Set E: Compared a $1300 Zeiss prime to a $400 Canon prime
Alt users could probably pick Zeiss vs Canon pretty well regardless but perhaps the Canon forum readers would have been less successful if the playing field had been a little more level with respect to lens quality? Who knows. Not knocking your test - I did find it valuable and fun. Thoughts?
If we were looking at 100% crops evaluating raw optical performance, I'd agree with you. But I think when looking at web sized shots it's more a question of different rendering styles rather than optical performance. So I don't think the results would have been significantly different had L primes been used. As it is, I only have one L prime left (135L) and no Zeiss matching that focal length, so I had to use the lenses I have.
denoir wrote:
The 100L is optically the same as the pre-L 100/2.8 Macro. I originally had the 100L but I replaced it with the older version after the L was in for repair three times with a broken IS motor. I did extensive side by side testing and there was really no difference as far as the optics goes.
It must be copy variation. I could definitely see a difference between my 100L and 100 non-L. Even the color balance was different. Also the L is a good 2/3 of 1/3 of a stop faster so if the non-L is called a 2.8 then the L is really more like a 2.5. It's not an extreme difference by any means but my L definitely had a bit more bite to the micro contrast.
I believe lens rentals tested a ton of copies from both and, IF I am recalling correctly, the L was sharper on average than the non-L although the very best few non-L copies were better than the worst L copies. The sharpest L was sharper than the sharpest non-L. The least sharp L was sharper than the least sharp non-L. From what I recall, by far most people would end up getting a sharper copy if they went with the L, although there was a modest chance of being unlucky and doing merely the same and very small chance to even do worse.
denoir wrote:
Yes, that does indeed look suspicious. However I had a control set (the pair consisted of the same Leica shot) and a clear bias would have shown there. A proper test would of course not let people see other peoples' responses. Difficult to do on a forum though. If I asked for the answers by PM the participation would probably be very low. Ideally one should set up a website where you can vote and where the samples are randomized for each user.
Canon is pretty neutral in its rendering. It also looked from the responses that people gave that they were primarily trying to guess the Zeiss lenses.
It did not so much show a difference in rendering style as in image quality. In short the Canon zoom sucked while the Zeiss prime was very good. 100% crop:
The results showed that both in this forum and the Canon forum it was the set that people got the best results on. That would indicate apriori that they expected the Zeiss to be better - at least in a technical sense. Furthermore after the results had been presented the replies indicated strongly that people preferred the Zeiss images. A few users generally known to be hostile to Zeiss expressed thing like "the difference is too small to be worth so much extra money" and "you could make them identical in photoshop", indicating that even they preferred the Zeiss.
The thing is that Zeiss and Canon design philosophy is entirely different. Zeiss has its style and enforces it. You can't do much about it in PP - it will still look like Zeiss rendering. Canon on the other hand produces a much more neutral image with the design goal to offend as few people as possible and to allow extensive postprocessing. It makes sense. They cater to a mass market....Show more →
are you sure the canon crop there was taken from an in focus picture? if that was anywhere near center frame and not near edges/corners then that really looks to be simply out of focus to me
Look at the POTN test, not that there are not differences. Test A with the 85 1.8 vs zeiss 85 1.4, the contrast/color difference (no clue about CA, sharpness, micro-contrat since the image has been re-sized so small) is pretty darn easy to spot and in the favor of the zeiss there.
for test B the exposure difference is way to extreme to be able to judge anything at all fairly, zeiss has more sky detail and crisper distant trees but less shadow detail, etc. but hey that is because the exposures are totally different!
for C it's sort of hard to tell, different color balance but a little hard to tell here otherwise (edit: hmm i see they were the same lens! so that sort of makes sense other than why is the color balance different?!)
D is tricky because of the huge vignetting differences and apparent exposure differences, they look super different and one of them has the bite on the bark but it's hard to tell whether it is lenses or vignetting or exposure for sure what is going on
E2 has better bite and I see it is the zeiss (i should say that the 100L also has better bite than the 100, it also gets tricky though since the L and non-L vary in effective aperture so diffraction or max sharpness don't align by aperture)
I also need to say that even the tiniest difference in focusing can make a difference, 10x liveivew at least six trials is the only sure way.
F1 looks better, zeiss again (however against 16-35 and not a 24 T&S II or 24 1.4 II etc.)
Just came across this thread via a reference in the thread on lens rendering--I know this original thread is years old. I found it very interesting to read through, at first.
But then I was shocked to see which lenses were chosen to represent Zeiss and which were used to represent Canon in the five comparisons.
Seems to me the lenses chosen obviate the results because of the confounding of the very different quality level of lenses chosen for each maker-- including two comparisons of a Zeiss prime to a Canon zoom, two comparisons of Canon consumer lenses to Zeiss professional-level lenses, and one comparison of an outdated and replaced Canon macro (for an image at infinity distance) to a top Zeiss lens. In all cases, there were better Canon lenses available than those that were chosen.
Of the five comparisons:
two compare a Canon zoom to a Zeiss prime!!!
one compares a much disliked and inexpensive Canon 50 1.4 to a very fine and expensive Zeiss 50/2--the comparison should have been to the Canon 50mm/1.2
one compares a very fine and expensive Zeiss 85/1.4 to a consumer grade Canon 85/1.8--the comparison should have been to the Canon 85/1.2 II
one compares an inexpensive, out of date, and replaced Canon 100/2.8 macro to a very fine and expensive Zeiss 100/2 MP--the comparison should have been to the Canon 100 2.8L IS, which MTF tests show to be much better than the Canon lens that was used.
There is a complete confounding of the lens maker with the quality of the lenses chosen from each line. Not actually informative about the differences between Zeiss and Canon lenses in general.
I saw this video comparing a zeiss and a fuji camera with various lenses which I think shows some good examples of different rendering.
It would be nice if some lens engineers who had worked in the industry a few years could share what kinds of decisions they have to make when designing lenses and what impact that has on the look of photos produced.
I've noticed that the more money I spend on the lens, the better I'm able to discern depth. I think it's called microcontrast and if a lens is high quality, the subtle shadows that help our eye/brain interpret depth are not washed out. The result is that we believe something looks more 3D.
The opposite of good contrast or microcontrast to me is computer animation. Computers can easily draw shapes, but the subtle shadows just look off so to my eye at least, everything looks fake. Rendering though is also color information and certain lenses seem to make colors look more saturated. But the video I think shows really nice examples.
chiron wrote:
But then I was shocked to see which lenses were chosen to represent Zeiss and which were used to represent Canon in the five comparisons.
Okay, so you clearly have not understood what this thread was about ... this thread was about seeing why some lenses are better than others. It was not about an objective comparison of Zeiss and Canon.
I'm no lens designer but I've read through the various Zeiss and Leica documents of lens designers publically available. Even a relatively simple 6 element all aspherical lens has already about 20 independent parameters, which means no supercomputer can actually brute force the possible choices. And ultimately not two lens designers agree on whats the best image quality, and to actually evaluate a lens you have to first actually build it.
Thats why the concept of one company knowing the secret sauce to make great lenses is a misconception. All these companies will give you is an approximation of the best result they could get. Thats why even lesser lens companies can have a Heureka and why better lens companies can have an underperformer, and thats why truely outstanding lenses are so rare.
There is one advantage however Zeiss, Leica and Voigtländer have over other optical companies, and thats crystal glas. Nikon, Canon etc wont use it simply because its so damn heavy, and thats a problem in more than just one way, specifically its a problem with autofocus, since you want the autofocus unit to be as lightweight as possible.
Interesting post, Sauseschritt. Especially about crystal glass. If you have more on this or references for it, I'd like to know more.
I'm of a slightly different mindset regarding 'an approximation of the best result they could get' because this appears to imply that all are striving to attain the goal of 'best' (however defined) and sometimes arrive at the peak but sometimes fall short. So a poor - good - very good - best quality scale.
My current take (we all change opinions according to new info, persuasive arguments and new or revised thinking) is that the people responsible (design staff, optical managers, range managers, marketing, etc.) combine to deliver lenses with a certain look, though they may also aim for range characteristics (color matching, design constraints like size, actuations, ergo design etc.). All creative work is subject to small things that are overlooked or deprecated yet which inject magic to the process. The 'unknown unknowns' if you will, the things not foreseen.
Even in the area of deliberate choice of knowns, companies choose what to aim for in their lenses. Certainly Zeiss do this, others who knows? Lens makers are clearly producing for lens rendering, drawing styles, use the terms you are comfortable with.
They are very deliberately creating how your image will look, and lenses are the major contributor to this process - which is why users can move their familiar and loved lenses across platforms easily and effectively. So called color science of camera makers is less of an issue than many believe, therefore, acording to this reasoning.
So I don't buy the 'quality scale', at all. If this was true, all would aspire to the clear winners or the lenses at the level they can afford, and a simple lens review would tell you all you need to know. It's not so, reductionism doesn't work here, in our art. You buy the best match, lenses that give you the best look for your tastes and preferences, for your light and your shooting environment. The makers are shaping the look our images give to the world. Quite a responsibility, and quite a challenge.
And a few things follow on from here, if we accept that lens design is converging in the modern era. Less diversity means less exposure to other ways of doing things, a decrease in the scope of what is possible. We may end up with the auto example, in which almost all cars look and even behave much the same. All that eccentricity in French and Italian cars, the distinct appearances - all gone, immersed into a vat of homogeneous design that offends no one and delights no one - the 'world car'.
I don't want to see the 'world lens' appear anytime soon. It would probably look like Sigma - lack of aberrations, but lack of character too. After all, the most successful makers determine the tastes of the greatest number of people, and this in time becomes the standard, the reference point for all players in the market. Canon and Nikon have both produced over 100 million units of their AF lenses, for their current mounts.
