The 50L isn't the perfect 50mm nor is it a mini 85L, but once calibrated it should be able to deliver more reliable results compared to what you are currently experiencing.
Can't understand those who write that this is about correct, it isn't. The lens is better than this.
It may lack contrast wide open and at MFD and it isn't the ideal "test lens" certainly not with test charts and b&w print, it is surprisingly good once you let go of these tests and start using it in the real world.
Again, you copy is mis focussing (front?), so sending it to Canon for servicing should solve your problems.
I have never owned the Canon 50mm f1.2. I do own the Sigma 50mm f1.4. The Sigma is a little soft at f1.4. This is my third copy of this lens and all of them were soft wide open. It does sharpen up nicely at 1.8 and at 2.0 everything is crisp.
What I am saying is that I have learned to live with the fact that I cannot shoot at f1.4 w/ this lens and get outstanding image quality. Anything from 1.8 upwards works great and the lens is stellar in all other areas.
If you like the lens alot and can live with shooting from f2.0 and higher then I would keep it. If not then find an alternative.
pkupcik wrote:
From what I understand the focus shift does not happen at f/1.2, but above it and it shows as a back focus. My problem is front focus at f/1.2.
What I see here is a camera that front focusses in combination with a 50 which backfocusses. The camera is borked by shoddy manufacture, whereas the lens is borked by design. First get the the camera properly calibrated by Canon, and then learn the three-hand grip necessary to use the 50L at f2 and f2.8. The three-hand grip involves holding down the depth-of-field preview button while focussing manually. If you have liveview you should be able to verify. In my experience the 50L works fine at f1.2, but focus-shifts at f2 - f2.8. A lens such as the 50L can expose distinct focus calibration errors in your camera's AF system and in its manual focus geometry.
Thanks all again for your suggestions. As a quick reminder all these shots were done on a tripod and with calibrated 5D. I've actually sent it to Canon last month for the mirror fix and asked them to recheck/adjust AF, which they did. The camera AF is spot on with 24L at f1.4 and 85 at f/1.8 and bunch of other f/2.8 lenses. Obviously 50mm f/1.2 is different, but still this is not normal to have every single tripod shot front focus, obviously something is wrong with the lens so hopefully the replacement copy is better. There are plenty of 50L users around here with better results so I'm not giving up on it yet.
I've lost count on how many new 50L's I've tested. They all have the identical characteristics because of the defective design flaw. I'm seeking a "less defective" copy that does not focus shift too much at f/2.8 to f/4 (aprox 5feet distance).
Even if I put my pixel peeping habits to rest the fact that stopping the lens exhibits focus shift seems unacceptable. An analogy is like driving a car with a bad alignment and you must fight the steering wheel to make the car go straight. I've been told using the outer focus point reduces or even eliminates focus shift. So basically using this lens at f/2.8 to f/4 you should be further than 5 feet to allow more dof to compensate for the focus shift in the lens. Thats broken to some (me) or good for others with different shooting styles.
I hate sigma's due to QC and calibration issues. I've tested many 50mm's too but I may just buy another one and pray in getting decent sigma copy for now. So far my experiences with the Sigma's have been horrible and unacceptable. I'd rather deal with the 50L's calculated focus shift than incaccurate sigma AF.
Having f/1.2 is nice to have but upclose this will simply have 1 eye in focus portraits unless the eyes are on the same plane.
Due to frustration I bought a 35L to be a "sorta wide" normal prime.
pkupcik wrote:
I think you guys may be right about the front-focus. It seems to be more pronounced with longer distances. It's really weird. At close distances on a test chart at 45 degrees I don't see any front/back focus, just blurry shot. I would think I should see the front focus on a test chart. However after I reviewed some of my long distance shots, it seems that the further I am from the subject, the more it front focuses. Maybe it's a combination of both front focus and lower sharpness in general. That would probably also explain why I'm not seeing the focus-shift problem - the front focus is compensating for it
I did try manual focus and at close distance couldn't make any improvement over AF. At longer distance I couldn't really manually focus due to the subject distance. I wish I had 5D II so that I could try liveview and micro adjustments.
I'm going to exchange this for another copy and hope for better result. I never had to return any lens, but I guess it had to happen some day. The last lens I got, 24mm f/1.4 II, is magnificent wide open. I knew 50L won't be able to match it wide open, hopefully the next copy will be different....Show more →
This lens has been discussed to death so I hope you read all the threads about this lens. Many of us have purchased it, tried our hardest to like, and failed.. Its "quirks" are not something I could live with, especially when those quirks meant that the photo was unusable.
for me, I was OK at 1.2 - 1.8, but at 2.0 to 4.0 it was mush. I knew that for my intended use, street shooting, I would not have time to manually focus the lens that often, but still wanted a shallow DOF. For $1200 (my price at the time from B&H), I thought I should get a working lens. Canon apparently didn't think so....
I can live with imperfect lenses but I wouldn't get another if I were you. I'd get something that works, and the 50L ain't it...
hyperion wrote:
Your results are exactly what pros experience when using it - and why most pros don't own it.
Really?
As what I knew lots of top wedding photographers use 50/1.2L a lots along with 24/1.4L II....if you shoot at wider than f/2, you will not really notice the "AF issues" in the real world...
The focus shift issue is not a design flaw per se. It is, to the best of my knowledge, intentional.
One needs to understand how focus shift arises to see why it is not always corrected.
Focus shift arises from undercorrected spherical aberration in very fast lenses. The focusing distance needs to be relatively short and the diaphragm stopped down only slightly. You cannot observe focus shift when the lens is wide open. When the lens is focused far away, or stopped down farther, the DOF is large enough to obscure the presence of focus shift.
When a lens is undercorrected for spherical aberration, what this means is that the light rays coming from a point on an object in focus, and entering the lens, do not precisely converge to a point on the image plane, as they should. Instead, the point is rendered as a circle of confusion (albeit a small one). When manually focusing with the aperture wide open, you are selecting the focusing distance for which that circle of confusion is smallest.
However, if you take the exposure when the lens is stopped down slightly, the position of the circle of least confusion along the optical axis is no longer the same, because the marginal rays at maximum aperture are blocked by the aperture stop. Because the spherical aberration causes these marginal rays to converge at a different depth than the near-axial rays, the elimination of the marginal rays in an undercorrected situation results in the circle of least confusion moving very slightly backward, and now your focus is slightly off.
Why, then, would you permit this phenomenon to occur? Why not correct for spherical aberration as precisely as possible? The answer has to do with the appearance of the background blur. Lenses with undercorrected spherical aberration tend to exhibit very pleasing--i.e., smooth--blur characteristics of objects behind the plane of focus. This is because the blur circle for these objects are not uniform in intensity, but drop off toward the edges, increasing the soft appearance. However, for objects in front of the plane of focus, the result is the opposite--you will see slight "ringing" of the blur disk, where the intensity increases toward the edge. But since it is less common to have objects of interest in front of your point of focus, undercorrection is the more desirable choice if soft backgrounds are desired.
And this explains in part why the 50/1.2L is designed the way it is. It is so that one achieves very pleasing bokeh, at the expense of some focus shift at near distances. For many photographers, this is a worthwhile compromise, especially for those who use the 50/1.2L for full-body portraiture.
The 50/1.0L presumably did not have focus shift because, to be frank, the lens has such shallow DOF and is so soft overall that the design didn't need smoother bokeh. And the 85/1.2L II, with its very large, heavy aspherical element in the focusing group, is also largely free of focus shift.
Toothwalker explains the focus shift/spherical aberration phenomenon far better than I can, and has diagrams to assist. Please see here:
Your observation is right on. But first you have to calibrate the lens for the "lens shift" aka front/back focus. Once that's done (each lens-camera combination is unique) you have a baseline of the sharpness of the lens. I use the 50L relatively frequent, but never fully open, at least at f2 but mostly at f5,6-11. In that range the lens is outstanding and the contrast/color is on par with the 85L, which i also shoot a lot.
Now, did i expect more from this, a 100%, but i can live with it and as cameras get cleaner at higher ISO's it's really no issue at all (especially as i usually need a little depth of field). If you feel you'll shoot mostly between 1.2 and 2.8 and don't need similar renditions than the 35/85L i would seriously consider testing the Canon 1.4 50mm or a third party lens (Zeiss anyone?). My personal favorite to shoot wide open is the 35L...
But seriously, why can't Canon figure out a prime/pro line up of L lenses (24/35/50/85/105, like the new 100 macro) that all start at f1.4, that has an outstanding build and optical quality with super rapid AF, a consistant filter thread diameter, IS and the ability to focus manually to infinity once off the camera (a real 85L problem) for around $1200-1600.
This would make some lenses lighter and cheaper (50 and 85L), as we're not paying for the technical exercise to create the fastest lens in the segment.
The technology is available on various lenses, but combine it, streamline the appearance (love the new 100 Macro) and make this the true professional L line.
Canon has to consider that they're about to become potentially the sole survivor of the digital pro photo market and high end shooters (as well as DP's used to Panavision lens sets) expect a lens line up comparable to the Hasselblad H or V series, or the Mamiya RZ lenses, with the same abilities and design and not a mix of options.
Beautiful! Thanks for posting that aberrant lens behavior.
I think this lens should be christened "The Mushball" or some other derogatory moniker.
it's it hard to believe that Canon would let something like that escape from the factory, but I can only conclude they do it as a joke. So many people complain but Canon just keeps selling it.
I tested on a 5d and 1Ds2 but deleted all the photos. I wondered, "Why keep test shots from one of the worst lenses I've ever owned"? I trashed everything...
hyperion wrote:
I rented it for my MK4s and could not stand it. IMO its a complete waste of money.
Your results are exactly what pros experience when using it - and why most pros don't own it.
wickerprints wrote:
The 50/1.0L presumably did not have focus shift because, to be frank, the lens has such shallow DOF and is so soft overall that the design didn't need smoother bokeh.
Where on earth have you got that info from ? I can almost certainly tell it's not from experience with multiple 50 f/1.0 copies.
PetKal wrote: wickerprints wroteThe 50/1.0L presumably did not have focus shift because, to be frank, the lens has such shallow DOF and is so soft overall that the design didn't need smoother bokeh.Where on earth have you got that info from ? I can almost certainly tell it's not from experience with multiple 50 f/1.0 copies.
It's just my hypothesis, my best guess as to Canon's design goals for the 50/1.0L. I'm not saying the lens is an underperformer--quite the contrary. For some reason, their engineers did not see a need to undercorrect the spherical aberration. In fact--I just want to add this--all else being equal (i.e., price not a factor), I would prefer to have the 50/1.0L over the 50/1.2L. I think the former was a better design overall. Contrast and resolution are not the end-all and be-all of photography.
wickerprints wrote:
The focus shift issue is not a design flaw per se. It is, to the best of my knowledge, intentional.
One needs to understand how focus shift arises to see why it is not always corrected.
Focus shift arises from undercorrected spherical aberration in very fast lenses. The focusing distance needs to be relatively short and the diaphragm stopped down only slightly. You cannot observe focus shift when the lens is wide open. When the lens is focused far away, or stopped down farther, the DOF is large enough to obscure the presence of focus shift.
When a lens is undercorrected for spherical aberration, what this means is that the light rays coming from a point on an object in focus, and entering the lens, do not precisely converge to a point on the image plane, as they should. Instead, the point is rendered as a circle of confusion (albeit a small one). When manually focusing with the aperture wide open, you are selecting the focusing distance for which that circle of confusion is smallest.
However, if you take the exposure when the lens is stopped down slightly, the position of the circle of least confusion along the optical axis is no longer the same, because the marginal rays at maximum aperture are blocked by the aperture stop. Because the spherical aberration causes these marginal rays to converge at a different depth than the near-axial rays, the elimination of the marginal rays in an undercorrected situation results in the circle of least confusion moving very slightly backward, and now your focus is slightly off.
Why, then, would you permit this phenomenon to occur? Why not correct for spherical aberration as precisely as possible? The answer has to do with the appearance of the background blur. Lenses with undercorrected spherical aberration tend to exhibit very pleasing--i.e., smooth--blur characteristics of objects behind the plane of focus. This is because the blur circle for these objects are not uniform in intensity, but drop off toward the edges, increasing the soft appearance. However, for objects in front of the plane of focus, the result is the opposite--you will see slight "ringing" of the blur disk, where the intensity increases toward the edge. But since it is less common to have objects of interest in front of your point of focus, undercorrection is the more desirable choice if soft backgrounds are desired.
And this explains in part why the 50/1.2L is designed the way it is. It is so that one achieves very pleasing bokeh, at the expense of some focus shift at near distances. For many photographers, this is a worthwhile compromise, especially for those who use the 50/1.2L for full-body portraiture.
The 50/1.0L presumably did not have focus shift because, to be frank, the lens has such shallow DOF and is so soft overall that the design didn't need smoother bokeh. And the 85/1.2L II, with its very large, heavy aspherical element in the focusing group, is also largely free of focus shift.
Toothwalker explains the focus shift/spherical aberration phenomenon far better than I can, and has diagrams to assist. Please see here:
Interesting and good to know. Though I've read previously that the "problem" was part of the design. I guess I wonder why Canon does nothing to correct the impression that the lens can be used as a regular lens. Why not come out and say "Good at f/1.2 or f/8-f22 only. It would certainly prevent those of us that had negative experiences from buying/bashing the lens if our use was outside the limited uses Canon intended, and mine clearly was.
jamesf99 wrote:
I think this lens should be christened "The Mushball" or some other derogatory moniker.
it's it hard to believe that Canon would let something like that escape from the factory, but I can only conclude they do it as a joke. So many people complain but Canon just keeps selling it.
I tested on a 5d and 1Ds2 but deleted all the photos. I wondered, "Why keep test shots from one of the worst lenses I've ever owned"?
While the focus shift is a nuisance, the lens can produce a certain look when used skilfully. Here it is on a 450D, where it seems to produce a reasonable sense of reality: http://cyberphotographer.com/450d/paul_lowrez.jpg
In my experience it is slightly better than the Rokkor 58 f1.2 and everyone goes mental over that in the alt forum. If you learn to work around the focus shift, it's one of the very best 50's out there.
trumpet_guy wrote:
The bottom line is this: If you don't like the performance it is giving you, then send
it back. Don't agonize over it too long. Despite it's AF flaws, one thing about the Sigma:
it is sharp. So, you have the Sigma as an alternative, if you don't want to play
Canon's "copy lottery." Of course, Sigma has a lottery of its own.
Regardless of the reason, if you can't get sharp shots, you won't use the 50L.
yes, I've heard Sigma has a lottery too, but compared to Canon, you can afford to buy 3 at a time, pay taxes, shipping, and still have cash left over at the end of the day.
If I buy another 50mm it will be the Sigma, but as much as I like other Canon lenses, I just don't think I'll ever own a 50 f/1.2L again.
One of Canon's design objectives for 50mm f/1.0 was: "To compensate for distance-sensitive aberration changes and provide quality images at all distances, by deploying a floating element construction." (Canon source).
The problem with 50 f/1.0 has been an extraordinary degree of copy variation which can not be explained by lens aging and overuse, since generally those lenses have been well "safeguarded" by their owners. Some copies are sufficiently or even surprisingly sharp at f/1.0, quite a few are very soft wide open. However, most copies seem to be very sharp beyond f/2 or so.
My hunch is that the lens fell off the EOS wagon because Canon for some reason could not ensure IQ consistency for the kind of money the lens was selling for. If they raised the price of an already expensive lens, they probably wouldn't have had much of a market for it. Again, just my best guess.
brainiac wrote: jamesf99 wroteI think this lens should be christened "The Mushball" or some other derogatory moniker.
it's it hard to believe that Canon would let something like that escape from the factory, but I can only conclude they do it as a joke. So many people complain but Canon just keeps selling it.
I tested on a 5d and 1Ds2 but deleted all the photos. I wondered, "Why keep test shots from one of the worst lenses I've ever owned"?While the focus shift is a nuisance, the lens can produce a certain look when used skilfully. Here it is on a 450D, where it seems to produce a reasonable sense of reality: http://cyberphotographer.com/450d/paul_lowrez.jpg
In my experience it is slightly better than the Rokkor 58 f1.2 and everyone goes mental over that in the alt forum. If you learn to work around the focus shift, it's one of the very best 50's out there.
I agree that looks good enough to keep.
I usually don't test lenses; they either work and do what I need them to do or as in one case in the last 15 years, I returned a lens (16-35) but immediately got another.
Now I wish I had kept all the test shots so people could see what I saw. I just couldn't use this one for the intended purposes. I'm just not a lens fanatic like many here where I pick apart every aspect of it looking for faults. I find ways to like them so I can use them. I guess my biggest disappointment is that I REALLY wanted to like the lens. I kept trying, but just couldn't justify keeping it. I'm glad it went back.
Keep yours if it does what you need. I certainly would, regardless of the bad press, a working lens is better than any review or complaint you'll read here.