Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
next PZ lens test report: Sigma 50mm f/1.4 DG HSM | A on APS-C
The "Well, it is mostly better than the Canon EF 50mm f/1.2 USM L II" line in the review verdict made me chuckle... Time travelling are we? Then again it's nice to know that even though Canon will inevitably come up with the new 50, the Sigma is still going to beat it. :-)
If you're quoting Klaus' verdict, Rover: He said "clearly better" which sounds in my ears a bit different than "mostly better"  Rolleyes Anyway, after all those people going crazy about this Sigma somebody needs to find some scratches in the glory, I guess  :lol:


Interesting list, Scythels. I just checked, I bought my Sigma 8 days later from the 70-200/4 which was quite fresh at the end of 2012, too. Would've save you $600-1000. Roger's article Sigma 35mm f/1.4 Arrives. Announces New World Order. was from 2012-11-21. 


And comparing a non-AF Zeiss 25/2.8 with a still great 14-24/2.8... hm, well I feel them different things. But after all who knows what I would go for today? And in the end it's always better to use the stuff we have instead of keep on waiting, hoping, complaining. Deciding against D600 could've saved you the latter, too.

The deliberation is pretty much "14-24 - zoom + AF" "21 - filters"


I primarily shoot landscapes, so the whole autofocus this is mostly a nonissue for me.  It's nice, and a convenience, but I can't think of a time where I will frequently be shooting than 35mm where I'll need AF.


In the end, I'm going to buy a wide prime/zoom.  It'll happen sooner or later.  I'm more loyal to price than any one brand, and optical quality is #1 if price is close.  It just so happens that canon has the better optics, and the better price by a very significant margin.

I'm using my 24/1.4 mostly inside and am happy that wide open AF sees a bit more than me. Also, the decision between 14-24 and 21ZF.2 was for me about range and a wider angle. I traded in some majestic flares I know I wouldn't see from a Zeiss. But at the moment my wide angle section is quite covered and I tend to use a set of DP Merills for lightweight and landscape.

Quote:Sigma has considerably more variance than canon, at least.  35L varied 7.32% across tested copies, and 35A varied 12.86%... over 50% more.  However, the average 35A is indeed better than the average 35L... though less than people on the internet would want you to think.


Sample data:

<a class="bbc_url" href=""></a>

In another post you told about how old the Canon versions are. If you think about, maybe it would be a good thing to re-check those tolerances after Sigma will have produced these lenses a a long as Canon has. Tighter tolerances are a lot about manufacturing experience. The "oldest" Sigma Art lens is nearly two years young...
That's true, but "pre GV" sigma was much much worse than that (a 10-20 varied 28% (!!!) across the five copies my uni has for the school paper).  Of course those have seen use, but the variation among them is ridiculous.


I feel that the 35L's age works for and against it.  Its construction tech/engineering is 'outdated' which works against it for holding elements in alignment, but its age works for it in that workers have built a gazillion of them, if the 35A is done on canon's human lines.  In ~2011 they moved 70% of their MFG back to Japan in a new robotic factory due to changes in yen value.  I know they have that plant, but nothing about what it does.


One of the big goals of the 16-35/4L IS was improving repeatability/equalness across produced copies.  Hopefully it was met.  It's a lens I'm keenly interested in, and of all the full-res images I've found from various copies (maybe a dozen) only one was a dud, though the the decentering was pretty bad.  Left side was pretty godawful.  I'm sure I can find a link to that specific sample, if you want.

No, thanks. You know as well, picking up 5 copies with 28% variance is in no way representative for batches of hundreds or thousands of lenses. And you know as well, one can come close to perfection at reasonable costs, but getting the last 5% to 100% always super and maximum quality will make the product extremely expensive. I don't demand what I can't afford to pay for.


Other decenterred lenses are not necessarily telling me, my copy will also be one.


Other not-decenterred lenses are not guaranteeing me to get a perfect one. If it happens, I'll send it in.


If Sigma's facing a worse "good copy rate" than Canon, it would be soon enough to make it my problem after one of my four lenses turns out to be a lemon. 6, if one counts the fixed lens ind front of the Merills.


I tested and re-tested a lot of things, but decenterring is nothing I was looking so, simply for the reason, I don't photograph even brick walls and there are other, bigger problems involved if a picture with focus point off centre is not in focus. I don't think about decenterring but I do think about too much seeking a perfect lens can spoil a lot of fun. I'm only talking for myself here.


But here's a link which puts things a bit more into proportions:


As I understood, in short words: Service is not always able to change a bad (decenterred, i.e.) lens into a good one, but sometimes to exchange my copy against another one.


I experienced that with Tamron. Sent it in twice, after that the camera bodies. Problems were not solved. They exchanged the body against a "new" one which had 1174 shots on it's counter when it arrived. I had some funny experiences with service and also very good experiences (Yashica, Zeiss, Mamiya, Pentax, Nikon, Sigma). And seeking for problems will nearly always succeed. 


Assembly lines with robots - why not, for boring, repeating  work? It's a fact humans get bored and the failure rate can increase. Our brain is not made to repeat the same grips all-day, week in, week out. To produce low-cost lenses in huge quantity and even quality, what would be your suggestions to keep costs down? On the other hand, robots are set-up, adjusted, maintained and controlled by humans, they will never produce better quality than they are set up to, usually they are not improving their workflow with experience. Learning robots are slower than their stupid mind machine colleagues.


By adjusting only one lens, there are so many variables, I'm still amazed how good they actually are. The first Zeiss Microscope lenses were made by try and error...


At least - who of us working always 100% free of failures? To me, that'd be a revolutionary experience. Evolution, by the way, uses very often "failures" and mistakes to create a better version of a living organism. It's not a "no, never!" principle. That makes it easier to deal with my own failures and also, to tolerate others.


5 copies with variance? So what? 5000 copies with no complaints, because there was nobody to recognize their flaws. And no one who kept track of statistics - even statistics can have mistakes  ^_^


Forum Jump:

Users browsing this thread:
1 Guest(s)