Juha Kannisto wrote:
I also feel that the focus throw for MF is too long especially when the ring is turned at a slow speed and it's hard to pick the ideal focus point that way since focus doesn't visibly change much at all when turning the ring several degrees. Therefore I've been using my lens primarily with AF only. The feel of the focus ring is otherwise very nice but a shorter MF throw would be welcome for my use.
I don't have auto-magnification (when turning the focus ring) enabled on my cameras as I don't like that being triggered always when turning the ring, but AF mode in magnified view with the + marker works fine with this lens on my A7C when I trigger that mode (just need to have the setting for AF in magnified view enabled). I've used the AF in magnified view option with this lens quite much when shooting close-ups of flowers as it's a very accurate focusing mode for focusing on tiniest details....Show more →
Glad I'm not the only one that feels this way about the manual focus experience. Can't believe this behavior got out the door from Sigma.
I'll be giving it another go later this afternoon as the weather is good. I'll try to get the manual focus much more accurate. I'll also try a test with using AF instead to see how well it does compared to manual focusing.
chiron wrote:
Can anyone provide an example of an image that was harmed in its visual impact and actual use by optical correction? I would be interested to see one or more as examples.
No they can’t, because to do that you would need a lens built to almost the same design, but without optical correction.And that will never happen even if you had access to manufacturers prototypes (because the decision is made before the prototyping stage). You can only see the tradeoff in lens designing software.
I think of distortion correction this way. Optical correction of distortion involves streching parts of the image, which can reduce resolution. We know that a teleconverter, which stretches the whole image can incur a resolution loss, so why wouldn’t optical distortion correction also incur a resolution loss? The difference between digital and optical correction might just be if it’s done optically or digitally, but either way you may have a resolution loss.
Wide angles with optical correction of distortion typically end up with a moustache shape distortion, which is bad for architecture, but not distortion free. Then the moustache ditortion is corrected digitally with a profile in post processing. Is that any better than just doing it digitally in one step?
How different are the end results? Just like @DavidBM says, no one can show you the tradeoffs without lens design software. And those are tradeoffs between hypothetical lens design, since only one is ever likely to make it to becoming an actual product that we can buy.
Fred Miranda wrote:
Continuing on distortion, how is it with the 24/2 DG DN? The little sister f/3.5 is virtually distortion free.
In their preview lenstip.com makes an assesment, « It seems the Sigma is close to the performance of the Tamron 24 mm f/2.8 Di III OSD M 1:2 which distortion was on a level of -5%.». Link, Sigma C 24 mm f/2 DG DN – first impressions and sample images
bjornthun wrote:
In their preview lenstip.com makes an assesment, « It seems the Sigma is close to the performance of the Tamron 24 mm f/2.8 Di III OSD M 1:2 which distortion was on a level of -5%.». Link, Sigma C 24 mm f/2 DG DN – first impressions and sample images
"As we've already mentioned distorion – you can notice distinct barrel variant of this aberraion in sample shorts. It seems the Sigma is close to the performance of the Tamron 24 mm f/2.8 Di III OSD M 1:2 which distortion was on a level of -5%. It seems the correction of that aberration really wasn't a priority for Sigma specialists."
That's bad news. So, the 24/3.5 has many advantages: (price, 1:2 macro, smaller size/weight + zero distortion). I wonder if bokeh is also better at f/3.5.
Of course the new 24/2 will offer better subject isolation. (more blur at same distance)
DavidBM wrote:
No they can’t, because to do that you would need a lens built to almost the same design, but without optical correction.And that will never happen even if you had access to manufacturers prototypes (because the decision is made before the prototyping stage). You can only see the tradeoff in lens designing software.
Not at all. I am asking specifically about digital correction of distortion as a process affecting an image.
What I mean is, can you point to a photograph where the effects of digital correction of distortion are visible in diminishing the visual impact of the photograph?
That wouldn't require an impossible comparison of designs (where there would also be a dozen other confounding factors anyway), just a before-and-after-digital correction where >something other< than distortion has been changed to the visible loss of visual impact? (Sometimes the images look better with the distortion left in--that is not the case I mean.)
Since digital distortion of correction alters resolution a little bit in stretched sections of the image, an example might show how this small loss of resolution in parts of the picture altered the visual impact of the photograph. This seems very unlikely to me.
Possibly more likely is an example where the removal of a slight part of the edges because of the digital correction of distortion has reduced the visual impact of the photo. But I would expect such cases to be very few. I don't think I have come across one in my own years of making pictures, but others may of course have done so.
Anyway, if there are such cases, I think it would add some empiricism to the discussion. And if there aren't, that would be an interesting fact.
bjornthun wrote:
I think of distortion correction this way. Optical correction of distortion involves streching parts of the image, which can reduce resolution. We know that a teleconverter, which stretches the whole image can incur a resolution loss, so why wouldn’t optical distortion correction also incur a resolution loss? The difference between digital and optical correction might just be if it’s done optically or digitally, but either way you may have a resolution loss.
Wide angles with optical correction of distortion typically end up with a moustache shape distortion, which is bad for architecture, but not distortion free. Then the moustache ditortion is corrected digitally with a profile in post processing. Is that any better than just doing it digitally in one step?
How different are the end results? Just like @DavidBM@ says, no one can show you the tradeoffs without lens design software. And those are tradeoffs between hypothetical lens design, since only one is ever likely to make it to becoming an actual product that we can buy....Show more →
This past weekend, I was talking pictures of a group of four people with my Sigma 45/2.8i which only has mild distortion. The heads closest to the edge of the frame were elongated/distorted due to pincushion distortion. In post, the distortion was 100% corrected without a noticeable reduction in resolution/contrast.
However, even with portraits, when inspecting the in-focus subject closest to the edges at pixel level, I could detect a slight resolution/contrast reduction. This would be even more noticeable if the lens was stopped down for optimal corner performance and the very edges were actually in focus (like in a landscape image).
After correction, I see slightly less 'micro' contrast which would definitely affect the lens MTF performance. So, lenses with higher distortion can offer a great trade-off in terms of optical design complexity, size and price but there is no free lunch.
In this case, almost!
Fred Miranda wrote:
This past weekend, I was talking pictures of a group of four people with my Sigma 45/2.8i which only has mild distortion. The heads closest to the edge of the frame were elongated/distorted due to pincushion distortion. After correction, the distortion was 100% corrected without a noticeable reduction in resolution/contrast. However, even with portraits, when inspecting the subject closest to the edges, I could detect a slight resolution/contrast reduction. This would be even more noticeable if the lens was stopped down for optimal corner performance and the very edges were actually in focus (like in a landscape image).
After correction, I see slightly less 'micro' contrast which would definitely affect the lens MTF performance. So, lenses with higher distortion can offer a great trade-off for size and price but there is no free lunch. In this case, almost! ...Show more →
So just a penny's worth of difference and even that only on the very edges at 100 percent? Could you see it in a print at viewing distance?
It might also be that the allowed but correctable distortion is related to that lens's beautiful rendering as well as to its size and price. The final package is certainly a great lens.
DavidBM wrote:
No they can’t, because to do that you would need a lens built to almost the same design, but without optical correction.And that will never happen even if you had access to manufacturers prototypes (because the decision is made before the prototyping stage). You can only see the tradeoff in lens designing software.
I just noticed that I miswrote "optical correction" when I meant to write "digital correction." So I think you correctly answered a question I didn't mean to ask! Sorry for the confusion.
chiron wrote:
So just a penny's worth of difference and even that only on the very edges at 100 percent? Could you see it in a print at viewing distance?
I doubt it. That's why the ill effects of distortion correction can de dismissed in the case of your scenario. Having said that, I find it annoying to shoot with lenses with high distortion that need to be corrected in post. Subjectively I give more praise to low distortion optical designs.
Fred Miranda wrote:
I doubt it. That's why the ill effects of distortion correction can de dismissed in the case of your scenario. Having said that, I find it annoying to shoot with lenses with high distortion that need to be corrected in post. Subjectively I give more praise to low distortion optical designs.
It annoys you because you see the distortion so clearly when you are composing? Given your eye and sensitivity to image characteristics, that would make sense to me. I rarely notice the distortion until I post-process an image.
Dustin Abbott reports a 'minus 11' is needed to correct the pincushion in the 90/2.8, no idea what software was used. But each can decide on this - if turned off in camera, does it show in the EVF for roof lines, wall edges etc?
Here is the image of the basin-like but linear distortion, at 18.09, YT video title:
philip_pj wrote:
Dustin Abbott reports a 'minus 11' is needed to correct the pincushion in the 90/2.8, no idea what software was used. But each can decide on this - if turned off in camera, does it show in the EVF for roof lines, wall edges etc?
Here is the image of the basin-like but linear distortion, at 18.09, YT video title:
Sigma 90mm F2.8 DG DN Definitive Review | DA
If it's Lightroom, "-11" is not mild pincushion distortion.
It's confusing because Cameralabs says "barrel" distortion:
"Before taking a close look at the detail I wanted a quick look at geometric distortion starting with the Sigma 90 f2.8 with Distortion Compensation on the Sony A7 III set to OFF, which is the default. With this setting, the Sigma lens exhibits some barrel distortion, so for my in-camera JPEG tests"
philip_pj wrote:
Dustin Abbott reports a 'minus 11' is needed to correct the pincushion in the 90/2.8, no idea what software was used. But each can decide on this - if turned off in camera, does it show in the EVF for roof lines, wall edges etc?
Here is the image of the basin-like but linear distortion, at 18.09, YT video title:
Sigma 90mm F2.8 DG DN Definitive Review | DA
I think you would see it. My Nikon 80-200/2.8 has a little less pincushioning at 200mm, and I see it in buildings, fence lines, and trees every time I use it. It bugs me, but then I remind myself I brought the zoom because I needed multiple focal lengths and wouldn't have time to change lenses.
Fred Miranda wrote:
It's confusing because Cameralabs says "barrel" distortion:
"Before taking a close look at the detail I wanted a quick look at geometric distortion starting with the Sigma 90 f2.8 with Distortion Compensation on the Sony A7 III set to OFF, which is the default. With this setting, the Sigma lens exhibits some barrel distortion, so for my in-camera JPEG tests"
That's weird. DA was definitely showing pincushion. If they've got test copies floating around with different types of distortion, that would not bode well for quality control. I'm thinking the Cameralabs report is a typo. Or they hired an intern who doesn't know the difference.
freaklikeme wrote:
That's weird. DA was definitely showing pincushion. If they've got test copies floating around with different types of distortion, that would not bode well for quality control. I'm thinking the Cameralabs report is a typo. Or they hired an intern who doesn't know the difference.
Perhaps Cameralabs misspoke because it's not likely a telephoto lens would have barrel distortion. Still we don't know for sure the severity of the pincushion distortion. Worse or better than the 85/1.4 DG DN?
