MayaTlab wrote:
Putting aside sample variation - and that might not be a good idea according to smpetty - that's exactly what I think about this lens' reviews.
That said I'm less and less convinced that Imatest is of any value to accurately describe a lens' resolution performance. How can you even remotely describe the corner performance of the two lenses below with just one number, given the performance difference in terms of fine / large scale contrast, sagittal / tangential orientation, etc. ?
I'd rather take a set of full res aperture runs on various scenes to evaluate resolution than these tests....Show more →
It’s mostly psychology, people want some sort of confirmation that they didn’t draw the short straw with their lenses. These tests allow them to quantify and easily compare with others. I’m tempted too when it comes to very expensive lenses. Who wants to be the proud owner of an expensive dud...
Maybe if manufacturers had better QA in place and a longer, more user friendly warranty period, then we would be buying with more confidence, and wouldn’t rush to do all these tests within the 30 days return window.
A lot of variables can affect Imatest results (for example : were the shots refocused to address focus shift when changing aperture ?) so I don't believe that it's a good idea to compare results across several websites and unless a single website tests several samples with a sound and rigorous methodology I don't think that it is in any way a measure of variance.
I'm sure that there are some very serious websites out there and it's possible that Imatest is very useful in some cases (isn't that what Lensrentals uses internally to test lenses before and after shipping ?) but I'm also pretty certain that a lot of them are using it improperly.
Les numériques, for example, has a consistently rubbish methodology across pretty much everything they "test".
But really for me the main problem is that I don't know what "3267" or "1547" means in terms of what the corners will actually look like. It's even more arcane to me than proper MTF graphs.
The problem with testing the corners is that the lens may have field curvature, which will result in poor brick wall results, but better results if you focus in the centre and shift.
I haven't tested my copy a great deal, but the results are adequate. Indeed, with IBIS, the problem is more one of subject movement than lens sharpness.
So far the main benefit of the lens is its low weight / bulk. If I work out how to get the nice bokeh, that will be a bonus.
We are poorly served, and lack reliable lens performance data for most apertures and even whole ranges of lenses. You can map a close correspondence between Roger's results and Zeiss's wide open MTF, as both test at infinity. FC at close range at/near wide open is not a show stopper, people want pleasant OOF there for portraits, general work.
FC is pernicious at infinity, and it varies significantly by aperture and focal distance, it's a struggle to know what is going on. Even stopdown to 'cover' it leaves the underlying pattern intact, as DOF expands either side of the curved focal plane, not on axis. In a landmark article in 2014, Roger Cicala suggested f8 for reasonable treatment of a set of 35mm lenses.
On this one, Sigma gave the lens a dual role, making it very good at infinity, according to them.
smpetty wrote:
It took me three tries to get a good copy of this lens. The first had mild softness in the top right and left corners. The lens that I exchanged it for, copy 2, was much worse with global peripheral softness and blurring. The final copy, copy 3, the one that I kept, is sharp from f4 and very well centered.
Enri wrote:
Video "Final Review" of the Sigma 45 from Dustin Abbott
The written review is worth a read as well, I think that it more successfully contextualises the lens' strengths and weaknesses than the video. One of the few reviews i've read so far that "gets it".
I have to report the same problem with de-centering. And this was by far the worst case I have ever encountered (and I had Samyangs from the early days, some 10 years ago). The right side was so badly de-centered that it took all the way to f/11 to mask it completely. This, together with the same issue (to a lesser extent) I had with Sigma 35 f/1,2 cooled me off of Sigma for the time being.
One more question.
I found a Elpro 3 55 mm which is 1.6.
Do you think this can be effective to add sharpness in close range of this lens?
I assume it works below 60 cm, which is where the lens degradates.
MayaTlab wrote:
The written review is worth a read as well, I think that it more successfully contextualises the lens' strengths and weaknesses than the video. One of the few reviews i've read so far that "gets it".
I don't think that I've been as excited by a lens in years. I hope that it will be a successful approach and that Sigma will not only fill this lineup but also release it for the R and Z mounts.
