garyvot wrote:
Michael, it appears from these crops like your lens may be backfocusing a bit. The rear of the eyebrow (toward the side of the face) looks sharper while the indicated focus point is softer.... (this seems especially evident in the f/2.8 crop).
I saw that, but if you look closer its actually front focusing. Look closely at her eyebrows. They get more defined as they go towards her ears. She was facing to my right a bit. Whats odd is that I tested it for that before I did the shots and it was spot on.
madmax200 wrote:
Totally unacceptable for a 1500.00 L lens.
Your test results do look problematic. If you are convinced it's the lens (whether sharpness or focus), I'd consider returning it for another copy if possible. If that's not possible, send it to Canon Factory Service with a carefully worded letter stating your concerns.
I have found in the "AF era" that there is far more sample variability among lenses than with traditional MF lenses... and it's much easier to see these differnces now that we can all pixel peep. I've had good and bad copies of several Canon lenses and have now resigned myself to the fact the lens purchasing is an... adventure. The key thing is to arrange a return privelege before purchase, and test immediately to your satisfaction.
Sometimes it's not possible (or impractical) to return a lens at retail... Last year I lost my original 50 1.4 in India and obtained a new one through insurance... Wide open it exhibited significant CA across the frame (you could see yellow halos around every light colored object). I sent it to Canon with a letter and some sample images and it came back with the CA completely fixed, and (if anything) sharper than my original copy. In fact, I like my 50 1.4 so much that I don't feel all that compelled to go test a 1.2L.
That said, I would not expect any ultra-high speed lens to be "tack" sharp wide open. My 50mm 1.4 is certainly sharper at f/2.8 than at f/1.4, but I still consider its performance consistent with professional results.
FWIW, I had a similar experience with the 100-400. My original copy of this lens (purchased way back with my EOS-3s) as soft at 400mm wide open, but a more recent copy is tack sharp at all focal lengths... It's a damn shame to have to worry about this, but judging by the often conflicting reports of individual lens performance on the Web, it appears to be a fact of life.
Remford said: I find tremendous benefit from each of the 50's I own (the 50 f/1.0, 50 f/1.2, and 50 f/1.4)
Although you did not want to elaborate about how you access image quality (yes, sharpness as well!) of your 50 f/1.2L, I would like to ask you to explain why you really need these three 50 mm primes.
I completely understand that you just need all the subtle differences of all the lenses you own but I admit that I get completely confused having more than 4 lenses all together (never did). How can you concentrate on photography having so many tools?
By the way, in my mind sharpness, colour and contrast (in this order) are the starting points for judging the IQ of a lens. What am I missing (I already know that you would say that I am missing something)
mfurman wrote:
By the way, in my mind sharpness, colour and contrast (in this order) are the starting points for judging the IQ of a lens. What am I missing (I already know that you would say that I am missing something)
Well, there's also the slightly intangible aspect of bokeh - the character of the out of focus areas in an image. A lot of this has to do with aperture design, with the curved-bladed apertures with lots of blades looking a lot different than fewer straight blades. This quality is why some people really like the Lensbaby, even though it is not a "sharp" lens, nor a contrasty lens. The fact that all the apertures are circular gives it a quality (especially stopped down) that you can't really get with any other lens.
In this sense, comparing lenses wide open is totally valid. Wide-open, bokeh quality should theoretically be at its "best" since there's effectively no aperture in use, and so there should be the highest level of "circularness". So if you're comparing, say... the 50/1.2L at f/2.8 to the 50/2.8 Macro at f/2.8 the 50/2.8 may actually have more pleasing bokeh since it's using a circular aperture there, while at f/2.8 the 50/1.2L's aperture is taking on a more "geometric" shape, despite its curved blades.
Coming back to the 50L, I think at f/1.2 the bokeh on my 50L looks quite strange - in a not very pleasing way. I suspect this has to do with the "oval" highlights people have pointed out which I'm definitely seeing in my photographs.
Sam Bennett wrote: Well, there's also the slightly intangible aspect of bokeh
I have to admit that I was to edit my previous post to add bokeh to the most important aspects of the lens IQ. As a matter of fact it is almost the most important to me. As you have pointed out, 50 f/1.2L does not appear to be "perfect" in this respect either.
I am trying to understand why it is so much more important to have f/1.2 maximum aperture (is it really f/1.2 for digital sensors) than a much better f/1.4 lens.
Could someone please direct me to one of the 50 f/1.2L photos at f/1.2 (or thereabouts) that exhibits the "oval" highlights? I didn't see these on the last set of photos posted.
Thanks!
For some reason, ever since the Canon 50 f/1.2 was announced there seems to be some sinister desire for this lens to be a failure so that the people who predicted this failure even before seeing a single photo can say "see I told you so". I just don't understand what perverse satisfaction some people get out of the failure of Canon lenses to live up to "expectation" -- seems to make their day.
looscanon said: I just don't understand what perverse satisfaction some people get out of the failure of Canon
Quite the opposite - many do not take any satisfaction seeing that a $1600 lens would need (for instance) a calibration. I can speak only for myself but I do want a 50 f/1.4L lens (I mean good build and reliable focus with f/1.4 lens optics). If Canon decided to build a "prestigious" f/1.2 lens, they should at least accept that people would scrutinize every aspect of it before buying and try to tighten up quality control (if something like this still exists). By the way I was more than eager to buy this lens and cancelled my order only yesterday.
CMOS said: Could someone please direct me to one of the 50 f/1.2L photos at f/1.2 (or thereabouts) that exhibits the "oval" highlights?
The images that were showing this effect most vividly, were on the Web site that crashed. I probably have some other saved and can e-mail them to you later.
mfurman wrote:
The images that were showing this effect most vividly, were on the Web site that crashed. I probably have some other saved and can e-mail them to you later.
mfurman wrote:
The images that were showing this effect most vividly, were on the Web site that crashed. I probably have some other saved and can e-mail them to you later.[/quo
vividly
It was my site and I looked at what the heck you guys were talking about and your all wrong.
Those were ceiling lights. They are round and pointed down. If you take a circle and move it up and away from you it looks oval. Its not the lens its just the normal way a circle would look laying up and flat. An optical illusion if you will.
Notice the small light on the back wall is perfectly round.
http://208.112.37.136/gallery2/main.php?g2_view=core.DownloadItem&g2_itemId=519&g2_serialNumber=2
mfurman wrote:
I am trying to understand why it is so much more important to have f/1.2 maximum aperture (is it really f/1.2 for digital sensors) than a much better f/1.4 lens.
Well personally, I don't feel like f/1.2 is important - that's not why I bought the lens. Personally, I would have been happy if Canon had simply done a 50/1.4 MKII that had the build quality (and internal focusing) of the 85/1.8. The f/1.2 is just a bonus in terms of flexibility, and the weather sealing makes it a serious lens that even the most demanding PJ can use in the rain without thinking twice.
With that said, if the lens cannot perform well between the range I bought it for (f/1.8-f/2.2), f/1.2 and weather sealing is not compelling enough of a reason to keep it.
As for oval highlights, I'll try to post some examples this afternoon.
CMOS wrote:
Could someone please direct me to one of the 50 f/1.2L photos at f/1.2 (or thereabouts) that exhibits the "oval" highlights? I didn't see these on the last set of photos posted.
Thanks!
You can also have a look at Gary's (ghuff) gallery No 2. He posted it back at... ... ... let's see... ... here, page 28.
The claim that it's the "shape of the lights" that is the problem is bogus. If you look at all the lights across the entire image it's clear that the shape changes continuously based on how they radiate out from the center of the lens. It may be "normal", but it is definitely odd.
This is also roughly the effect I'm seeing with mine, and I do not see it with the 35L. However, that's a tough comparison since technically I probably wasn't standing in the same place, it's a different focal length, etc. May just be a fact of life with 50mm lenses for all I know. I tried to pick up a 50/1.4 when I got the 1.2 for comparisons, but someone was buying their last one as I was getting the 1.2. Unfortunately the friend I referred to earlier with 50/1.4 has sold his, so I can't borrow for comparisons.
Koivulehto wrote:
Comparing the Christmas tree bulbs from the different aperture shots this is how it looks to me:
- at f/5.6 the aperture seems to have 8 blades, and most of the lights are shaped like regular octagonals
- at f/2.8 the highlights are still almost reqular
- at f/1.2 and partially also at f/1.8 the bottom part of the highlight figure is cut off, which to me looks like the bottom of the sensor (or a retangular frame at the lens mount of 50L f/1.2, if it has a similar frame as in 24-105L) would be visible in the shape of most of the highlights, possibly because of the direction of light from that particular pointy light source
- all this is consistent with the off-axis light direction from the Christmas tree bulbs - its effect is just not visible at smaller apertures, because the perimeter areas of the lens are not utilized then...Show more →
Jonas B wrote:
The shape of the highlights (wide open and nearly wide open, isn't that the effect called "Cat's eyes" or something similar? I have seen this with many very fast lenses and IIRC it has something to do with the size of the lens and mirror chamber. Sometimes my memory doesn't serve me though, I may be mistaken.
Those quotes are from pages 29 & 30 of this thread. I think there was no more discussion about the reasons on those pages, but I look forward to it!
I have no comment on the whys of the phenomenon. My knowledge of optics simply isn't good enough. But even so, blaming it on the shape of the lights themelves doesn't make any sense to me considering the regularity of the pattern, and it's radial nature around the lens' center. If it was dependent on the shape of the lights themselves, each would be a different shape, which they're not.
EDIT: I just looked at madmax's example and I agree with him - that oval effect is the shape of the one overhead light, but it's not the phenomenon being discussed. What I'm referring to is what you see in Ghuff's examples (and my own).
I think we should be able to understand something of this issue by examining the excellent examples Ghuff posted in gallery No 2.
On the wide open shots, many of the highlights have some close-to-linear edge, or even two of them. A very natural explanation to that is a retangular mask of some sort, which is cutting off some the original round or oval shape projected from the light source (round if it's directly towards the lens, oval if it's coming from an angle). A retangular frame in the lens mount is one possibility, and the other obvious retangular frame is the sensor itself.
Why the masking effect of either of these large retangular frames is visible in the small OOF highlights, escapes my knowledge of optics, if I even have any ... One vague possibility is the large size of the wide open aperture vs. the sensor size, but I admit I can't continue from that.
Look near the center of the paged link, where it says "defocused star image".
And, I have seen examples of the 35L doing similar things with radial blurred patterns of lights, especially wide open and really noticeable with Christmas lights at certain distances.
The 50L f/1.2 is not unique in this type of lens aberration, but it may present this tendency more noticeably than other lens, especially the 35L.
I'm not arguing for or against this tendency, but whether this tendency is tolerable or not, is up to you.
One vague possibility is the large size of the wide open aperture vs. the sensor size, but I admit I can't continue from that.