RCicala wrote:
I'm not sure it was negative as much as it was 'so you are aware'. This lens has very complex in-lens focus correction (and we're seeing that can be a problem, although it should be an easily fixable problem); along with vignetting and distortion correction as a default.
Is that bad? Nope, it's wringing every bit of performance out of the lens, allowing it to be smaller, especially.
Is it a good thing to know about? Yep! Someday somebody is going to accidentally turn off the default setting, for example, and if they don't know about it going to spend time and money sending their broken lens in for repair when nothing is wrong. Or some firmware upgrade is going to have someone saying 'now my lens vignettes' or antivignettes, etc. ...Show more →
For me it isn't about being made aware which I agree is a good thing but it's about the how the information is presented. The review says: "That being said it is a bit unfortunate that the lens can't really live without auto-correction - after all Canon and Nikon can manage this as well in their offerings." I'd like to see something in these reviews that quantifies what exactly is unfortunate about the Sony results as compared to the Nikon and Canon lenses - maybe side by side images showing how my image will be better with the Nikon and Canon lenses because of the Sony auto-correction. Otherwise, how do I use this information to understand what I'm giving up? It seems like this open ended criticism just allows each person to perceive this to be anywhere from a non-issue to the worst issue in the world. In my view a website that spends as much time as Photozone does complaining about lenses meant to be corrected with a lens profile should spend the time showing the impact to corrected images with comparisons to lenses that don't need to be corrected along with any other tests that can help the reader understand the real impact.
It's strange to see in the verdict section of the review the Nikon and Canon lenses praised for being able to manage to build their lenses without the need for auto correction and finish with praising the Sony lens as superior to the Nikon and Canon versions. Is the reviewer saying they want Sony to design a lens that is no larger and without auto correction and that is still superior to the Nikon and Canon lenses? Or do they want Sony to make the 24-105 larger? I understand that the Sony lens relies on correction and I'm glad the review points that out but the review doesn't help me understand what that correction actually costs me nor does it help me understand where the reviewer actually thinks Sony should have compromised to build the lens without relying on a lens profile.
Yeah, it doesn't seem like the comparison with Canikon is fair here- neither the 24-105/4L I/II nor the 24-120/4G are in the same performance class as the FE 24-105/4 G, competent as they are for what they are.
And I'll reiterate that it's good to point out that correction is going on as this robs absolute acuity. It will likely be applicable at even higher resolutions that we have to assume will be available in the future relative to lenses that are better optically corrected to some degree.
[on counter argument, higher resolution sensors may not be very effective with slower lenses due to diffraction, to the point that the software corrections make no real-world difference in final output versus a better corrected lens when shot at common apertures]
johnctharp wrote:
Yeah, it doesn't seem like the comparison with Canikon is fair here- neither the 24-105/4L I/II nor the 24-120/4G are in the same performance class as the FE 24-105/4 G, competent as they are for what they are.
And I'll reiterate that it's good to point out that correction is going on as this robs absolute acuity. It will likely be applicable at even higher resolutions that we have to assume will be available in the future relative to lenses that are better optically corrected to some degree.
[on counter argument, higher resolution sensors may not be very effective with slower lenses due to diffraction, to the point that the software corrections make no real-world difference in final output versus a better corrected lens when shot at common apertures]...Show more →
"And I'll reiterate that it's good to point out that correction is going on as this robs absolute acuity."
I do understand what you mean to say, and I suspect it's often true, but I find myself wondering whether this is *necessarily, completely* true.
For example, say that the software correction being applied reduces lateral chromatic aberration by transposing the output of the red channel over to a nearby pixel so as to better align colors. This would seem to improve acuity. Now, one might say that the LaCA should have been better controlled optically so that software correction was not needed and that the suboptimally corrected lens, itself, was thereby inferior to one that wouldn't need correction. However, every lens design is a compromise of a multitude of attributes and perhaps giving the lens designer freedom to let LoCA slide a bit (because it was correctable in software) enabled the lens to compromise less on other attributes which also improved acuity (and perhaps weren't easily solved either in software or optically). Or, perhaps the software corrected version was simply indistinguishable from the optically superior approach (at least to everyone except Fred), but the corrected version of the lens gets to weigh less.
Clearly, some (types of) corrections are going to diminish acuity, as you say. But as long as some do not, or at least degrade less than an observable amount for some other benefit, I wonder whether you can fairly evaluate any lens separate and isolated from the software correction intended to be used with it. The final, corrected result of one lens compared to the final, corrected result of another may be the only relevant measure of acuity.
Klaus is simply doing his job even pointing out these data; many reviews will ignore the RAW data and report post-correction. You want to know because if it is possible to turn it off, you want to know the consequences of doing so, esp for jpgs, which may not be easily recoverable in post. Depending heavily on s/w correction is not exactly out of the lens design best practice manual, and many of the best lenses owe a fair degree of their excellence to superb distortion control (think top macros). Many short teles are similar in this respect.
CZ have long held that 2% is the threshold above which distortion becomes troublesome and they try to stay under it, which is why even the wide Distagons are just above that figure in a waveform barrel. The Canon 24-105II is incredibly low in distortion from ~28mm - ~90mm even though the extremes don't really indicate this. in this range it's around 1% - a figure the Loxia 85mm has. The Sony 24-105 goes where none have dared, not much of an exaggeration for a high end zoom lens, maybe. Distortion is a healthy 3% plus under ~30mm and over ~65mm to 105mm.
There will IQ losses but count on the design staff to have very carefully run the numbers; you'll get the very best trade-off here, the best IQ given their working brief. Wide range zooms are lens design hell, and this one is the best in class. The Sony approach will either take off and they will all do it, or they will be on their own. It obviously works, and the 24-105 is almost the equal of the GM 24-70 at landscape apertures, with better CA control. That one is $900 more and the weight of a small prime heavier.
Parariss wrote:
For example, say that the software correction being applied reduces lateral chromatic aberration by transposing the output of the red channel over to a nearby pixel so as to better align colors. This would seem to improve acuity.
I'm definitely talking about distortion as philip_pj points out below; lateral CA is non-destructive unless it's really bad. Few lenses are bad enough to notice, but many newer lenses have higher amounts of CA simply because that aberration can be left to software while others are addressed instead (if there is a compromise to be made).
Distortion is another that is being left less corrected, as is vignetting, as resolution and dynamic range increases have allowed more leeway here.
And we've generally seen better correction of longitudinal CA which cannot be automatically or even fully corrected in software.
philip_pj wrote:
Distortion is the big one, in terms of affecting the image appearance.
Yup, and thanks for the data comparison!
And I agree that Sony's compromise here makes sense: they made the lens as small and light as Canon's original 24-105/4L, but managed to produce an optic with very high resolving power despite the need for software correction- and in the end, while not absolutely perfect, produced an extremely effective tool, certainly best in class!
I found my 40-70 GM f2.8 quite heavy and the 88mm lens size costly for filters etc. Add to this the smaller zoom range and it starts to stack up in favour of the 24-105 G f4 lens. I took it back for a 24-105mm G f4 lens which I have had to send back to Sony (181.. SN). I used the 24-105mm f4 on my old Canon 5D and found it a great workhorse of a lens.
Klaus is simply doing his job even pointing out these data;
The thing is he didn't just point out the data. He offered a number of conclusions about the data without in my opinion backing it up with anything useful and because he didn't provide anything else you can't include it here either. I'm sure if the review provided the data to show how it affected the image appearance as compared to the Nikon and Canon lenses you would have included it but as far as I know it doesn't exist.
Maybe the missing link is that those with experience in this area already know how to translate the above into meaningful data and if so it would be great if they would share the formula. Do these numbers mean that my images at 24 and 105mm will be worse with the Sony lens over the others shown? If so, in what ways and by how much?
Or, does the reviewer care more about the roads chosen to get to the finish line vs who won the race?
Not sure whether this is a new development resulting from this problem, but B&H and Adorama both have them on backorder, so they won't need to pull anything out of stock. It's much cheaper for Sony to fix this upstream, so you can bet dealers are being told of the recall.
Parariss wrote:
Not sure whether this is a new development resulting from this problem, but B&H and Adorama both have them on backorder, so they won't need to pull anything out of stock. It's much cheaper for Sony to fix this upstream, so you can bet dealers are being told of the recall.
I don't think that there is a "recall" yet, is there?
When I worked for a major electronics products manufacturer X% of each incoming component was checked for tolerance to spec. Any reject count over the specified limit and the whole incoming batch was rejected. Statistically it seemed quite accurate.
timballic wrote:
(Pity about being poor with IR hotspot, but thanks for the info) When you say "reasonable" cf the 24-70/4, do you mean it feels only a little larger/heavier, or a lot?
You can definitely feel the difference in size and weight. I'd venture to say it feels a lot heavier and it is a fair size larger. The 24-70 f/4 is tiny and light the 24-105 isn't :-)
Although I'm guilty of trying to use the Canon 100-400 lens on a NEX-3 (the original one) once. 