IkarusJohn Posted October 30, 2015 Share #1 Posted October 30, 2015 Advertisement (gone after registration) Discussions seem to meander, and I'm getting lost over what was said where. Sean Reid has posted some initial comments about the new zoom, and PCMag posted a review of the lens here; the reviewer gave the lens 4/5 score, but pointed out that the lens was best at 50mm, with 24 being a bit soft and there being resolution issues at 90mm, below f/8! This is less than encouraging ... As we are spoilt for choice of excellent M lenses between 18 & 135, this does suggest that this zoom might not be the best choice for M users in this range. A comparison between the new SL zoom and the R 28-90 lens would be informative. I'm going to get the opportunity to try the camera at some stage in the next few weeks, and will be testing the 15 Distagon, 21 Summilux, 28 Summicron, 50 Summilux, 75 Summilux and 90 Summicron against the 24-90 zoom (if I get the chance. I'll generally be trying each at f/4, focussed on whatever is available at middle distance. Infinity isn't really my thing, but I will see what's available. If there's a fruit bowl about, I'll give it a go! 3 Quote Link to post Share on other sites More sharing options...
Advertisement Posted October 30, 2015 Posted October 30, 2015 Hi IkarusJohn, Take a look here APO-Vario-Elmarit-SL 24-90 zoom. I'm sure you'll find what you were looking for!
Daedalus2000 Posted October 31, 2015 Share #2 Posted October 31, 2015 Discussions seem to meander, and I'm getting lost over what was said where. Sean Reid has posted some initial comments about the new zoom, and PCMag posted a review of the lens here; the reviewer gave the lens 4/5 score, but pointed out that the lens was best at 50mm, with 24 being a bit soft and there being resolution issues at 90mm, below f/8! This is less than encouraging ... As we are spoilt for choice of excellent M lenses between 18 & 135, this does suggest that this zoom might not be the best choice for M users in this range. A comparison between the new SL zoom and the R 28-90 lens would be informative. I'm going to get the opportunity to try the camera at some stage in the next few weeks, and will be testing the 15 Distagon, 21 Summilux, 28 Summicron, 50 Summilux, 75 Summilux and 90 Summicron against the 24-90 zoom (if I get the chance. I'll generally be trying each at f/4, focussed on whatever is available at middle distance. Infinity isn't really my thing, but I will see what's available. If there's a fruit bowl about, I'll give it a go! Thanks. I look forward to hearing your views. Quote Link to post Share on other sites More sharing options...
psss Posted October 31, 2015 Share #3 Posted October 31, 2015 it's never really fair to compare zooms with primes...the only zoom i have ever seen that might be able to compete with primes is the sigma art 18-35mm and it is a crop sensor zoom..... from the files i have and from the very short time i got to spend with the 24-90 i found it great....definitley sharp enough at all lengths i checked, stabilization worked well, size is....ok...it is what it is...and i really like the range....24 is much better then 28 and 90 is a world better then 70.....so i am pleasantly surprised that leica actually announced one lens and i am perfectly happy living with that one lens for a while....add in some M primes for fun and you are good to go....price is pointless to discuss, it is what it is.... i would say it is definitely better then existing canon, nikon, tamron offerings in that range (even if they are all 2.8 for the whole range) and so in all honestly there is no comparable lens/camera combo out there....especially not for sony, the zeiss 24-70 f4 is pretty poor....but i guess a new one is coming.... i honestly don't think that the dpreview files do the lens justice....the copy i got to play with was much better and showed none of the softness....doesn't take care of the banding of the sensor of course.. looking forward to your test, i doubt the zoom will be anywhere near (especially) the corner performance of the primes..... 1 Quote Link to post Share on other sites More sharing options...
IkarusJohn Posted October 31, 2015 Author Share #4 Posted October 31, 2015 (edited) The idea, for me anyway, is not to compare these lenses with each other,necessarily, but to see if they will work for me. If I don't like the performance of my M primes on the SL, that's a problem for me. I also want to see how the zoom performs for me. Rest assured, I won't have a tripod, I won't have Sean Reid's focusing board or his discipline, many here will tell me they're boring, out of focus or just terrible, and they will compare them to each other (which isn't the point for me). However, my plan is to learn something from them and I'll share the raw files. Don't know when this event will be, or how much I'll be able to test. I'll have an APO-Elmarit-R 180/2.8, but not sure I'll have an R-L adapter by then. Edited October 31, 2015 by IkarusJohn Quote Link to post Share on other sites More sharing options...
vikasmg Posted October 31, 2015 Share #5 Posted October 31, 2015 (edited) ... PCMag posted a review of the lens here; the reviewer gave the lens 4/5 score, but pointed out that the lens was best at 50mm, with 24 being a bit soft and there being resolution issues at 90mm, below f/8! This is less than encouraging ... ... Nonetheless, the review's bottom line does say >>The Vario-Elmarit-SL 24-90mm f/2.8-4 ASPH. is the first standard zoom for the Leica SL system, and it doesn't disappoint.<< And the conclusion says >> The Leica Vario-Elmarit-SL 24-90mm f/2.8-4 ASPH. is a versatile lens. It covers a longer zoom range than other wide zooms that start at f/2.8, delivers acceptably crisp results (especially for a zoom), shows very little distortion, and can focus close. Fall-off darkens corners at wider angles, but that's something that can be corrected with software as needed. This kind of image quality in a zoom lens comes at a cost, and I'm not talking about the luxury price tag that is a fact of life when shopping for Leica products. The 24-90mm is big, and it's heavy. Aside from that, I don't have anything that bad to say about this lens—it's another excellent optic from Leica, and a good companion for the SL (Typ 601). << I'd say that's fairly good. - Vikas Edited October 31, 2015 by vikasmg Quote Link to post Share on other sites More sharing options...
firststream Posted October 31, 2015 Share #6 Posted October 31, 2015 Chromosoft has an interesting blog entry regarding the distortion of the lens at 24mm, and the amount of in-camera correction needed to fix it: http://chromasoft.blogspot.co.il/2015/10/how-much-lens-correction-is-there-on.html Disappointing first of all, that this lens requires in-camera correction. Disappointing too that in the corrected image displayed, the tell-tale meandering verticals of a digitally produced and repaired image, remain. (In the corrected image, it appears as if the camera was tilted in such a way that verticals lean slightly to the left. At the same time, the vertical far-right edge of the sign tilts to the right.) I maintain that the in-camera corrected warp of a digital lens fails whenever the image is shot off-axis simultaneously in the x-y-z planes. It's my only complaint about the Q. Added to the problems under current discussion regarding the SL, is whether or not professional photographers will accept digital-specific lenses requiring in-camera correction if such tell-tale issues eventually become evident to the discerning client or editor? 1 Quote Link to post Share on other sites More sharing options...
ynp Posted October 31, 2015 Share #7 Posted October 31, 2015 Advertisement (gone after registration) To be honest, I don't understand why the in-camera or in-software corrections are so disappointing. In the real world of professional digital cameras it's a norm. Hasselbad uses DAC corrections for many years and it's does not disappoint their customers. DAC or similar corrections allow the manufacturer to design better lenses for digital. I have seen all the reviews and examined the files I was able to find and I cannot say that I am dissatisfied or disappointed. The zoom behave as a zoom and, in my view, many board members are too spoiled with the modern Leica -M lenses and expect too much. IMHO the 24-90 zoom will be a successful general purpose lens, very attractive to the people shooters. Of course, the SL and its zooms are not targeted to the landscape photographers and others who need to look elsewhere. Quote Link to post Share on other sites More sharing options...
firststream Posted November 1, 2015 Share #8 Posted November 1, 2015 Hasselblad lenses are meant to be used both on film and digital bodies. I don't think there are any H-lenses engineered specifically for digital work. The Hassy lens correction software is for tweaking; the SL and Q corrections are used for repairing. It's an exaggeration, but the ChromaSoft image is almost "fish-eye" in its distortions; something's gotta break in the corrections when the image is unwarped. When an image is shot off-axis in all three planes and then corrected, there appears to be what I can only describe as a two-way keystoning effect in the Q and SL. (Sorry - I'm not an engineer and cannot describe it more accurately.) The Q image exaggerates keystoning from what are the normal parameters of a 28mm lens, and I can see it in the corrected SL image cited above. The irrational geometry coming out of these digital only, inferior-to-film lenses doesn't seem to have bothered the camera enthusiast quite yet, and may not, ever. It's the photo enthusiast who will first notice it, and then eventually become bothered by it. For me, it's the new issue in the conversation regarding image quality and will be of concern to the professional photographer's relationship to those that pay him. Quote Link to post Share on other sites More sharing options...
firststream Posted November 1, 2015 Share #9 Posted November 1, 2015 Oops, there are three HCD lenses in the Hasselblad family that are used only on their digital systems. My bad. Quote Link to post Share on other sites More sharing options...
scott kirkpatrick Posted November 1, 2015 Share #10 Posted November 1, 2015 Thorsten Overgaard's http://www.overgaard.dk/leica_lens_compendium.html blog covers the best-known Leica R lenses (designed for film!) including zooms, and you can download the technical spec sheets for many of them. Most include a graph of the barrel or pincushion distortion expected as a function of the distance from the center of the image. It's large at the wide end of the late R zooms. The 35-70/4 had -3% distortion at 35 mm; the 28-90 had -3.2% at 28 and still -1.3% at 35 mm; the legendary APO-70-180 had -3% at 70 mm, dropping to 0 at 100 mm focal length. The Leica designs (the 35-70's are believed to involve Japanese collaborations) show pincushion distortion at their long ends: +1% for the 28-90, +2.2% for the 70-180. The distortion increases monotonically with radius, unlike some of the famous M lenses, which have increasing barrel distortion at moderate radii which then reduces to zero or pincushion at the corners of the image, resulting sometimes in "mustache" corrections. So distortion was once considered simply the least of the evils that a lens could commit, and is now correctible in postprocessing software with only mild side effects (loss of resolution, chiefly, barely visible at 100% enlargement). One result of digital imaging with software as an element that can be designed is that we may no longer see distortion curves in the technical specifications. It is missing in the 24-90/SL document. scott 1 Quote Link to post Share on other sites More sharing options...
cpclee Posted November 1, 2015 Share #11 Posted November 1, 2015 The two-way keystoning effect you referred to is indeed very apparent in these images. The "after" image still doesn't look totally satisfactory. I do wish that they included another shot from a 24 lens not requiring this sort of repair for comparison. In the past Leica argued against designing lenses with intentional distortions only to be fixed in software. This seems to mark a change in their stance. Chromosoft has an interesting blog entry regarding the distortion of the lens at 24mm, and the amount of in-camera correction needed to fix it: http://chromasoft.blogspot.co.il/2015/10/how-much-lens-correction-is-there-on.html Disappointing first of all, that this lens requires in-camera correction. Disappointing too that in the corrected image displayed, the tell-tale meandering verticals of a digitally produced and repaired image, remain. (In the corrected image, it appears as if the camera was tilted in such a way that verticals lean slightly to the left. At the same time, the vertical far-right edge of the sign tilts to the right.) I maintain that the in-camera corrected warp of a digital lens fails whenever the image is shot off-axis simultaneously in the x-y-z planes. It's my only complaint about the Q. Added to the problems under current discussion regarding the SL, is whether or not professional photographers will accept digital-specific lenses requiring in-camera correction if such tell-tale issues eventually become evident to the discerning client or editor? Quote Link to post Share on other sites More sharing options...
firststream Posted November 1, 2015 Share #12 Posted November 1, 2015 The ChromaSoft images as referral once again: http://chromasoft.blogspot.co.il/2015/10/how-much-lens-correction-is-there-on.html When the Q was introduced, it didn't take long to see examples of the uncorrected images coming out of it. Some people figured,uncorrected, the image had an equivalent focal length of about 24mm. Any keystoning effect in a Q image may have the relatively robust characteristics of a 24mm lens relative to a classic 28mm lens, regardless of what is happening with the in-camera correction. With that in mind, the 24mm image linked above may be coming from an uncorrected image of 21mm say, for argument's sake. It could be the distortions familiar in a wider focal length image that we see. For any given focal length setting of the SL zoom, we may have to accept focal length characteristics common to a slightly wider FL. That includes the effect of foreshortening, as well as keystoning. In the comparison shots from ChromaSoft, it doesn't take much imagination to see how the corrected picture has been cropped, chopped, pulled and squeezed. Notice the disappearance of elements from near the frame-edge. I question what the difference is in pixel count between the 24MP of the uncorrected image, and what is remaining post in-camera correction, and shown in the EVF or back-screen? This shouldn't be of any concern to the M and R-mount user. I'm thinking intuitively on this issue. Keeping that in mind, I just cannot compare the in-camera corrections for something like an R-mount lens which has already been design optimised for the film plane, versus the corrections needed for a digital only lens which has no 'responsibility' to the film plane, prior to correction. That is to say, tweaking an R-mount lens' image is different to repairing the image on the SL sensor. As CPCLee suggests, it would settle the debate with comparison images - although I'd add that I'd like to see the test done with the image-plane off-axis on the three planes. It would also be interesting to see a test conducted at 28mm, allowing the comparison of the Q image (also in-camera corrected), as well as a film-plane optimised lens from the Leica family. Quote Link to post Share on other sites More sharing options...
IkarusJohn Posted November 2, 2015 Author Share #13 Posted November 2, 2015 (edited) I've looked at those dreadful thumbnails again and again, and what I can't get over is who would take such a shot? It's truly horrible, in either version. Putting that aside, and the fact that the lens has been focused wide open on the centre of a sign, off centre so at best the G in danger, PE in open, RK in parking and H in the second Holland are the only things in focus, the entire frame is made up of out of focus rendering. The uncorrected image has nasty vignetting right in the corners and the verticals are bowed all over the place. The software corrected image shows a mild improvement (the vignetting is gone, hard to say about the verticals as the sign is now trapezoid and what should be vertical isn't - what was vertical in real life, and was the camera held level?). The questions these images raise are as much about what happens if you are careless using any wide angle in such a way as they do about the in camera corrections. I have not a shred of doubt that if I took such a shot with my M60 and 21 Summilux, the result would also be spectacularly off-putting; and I'm pretty happy with my 21 Summilux, but taking images off vertical and what looks like off horizontal is not going to be pretty. On the plus side, the sign is nice and crisp, if poorly focused - who would have a depth of field which did not include the entirety of a written sign ... As CPCLee suggests, it would settle the debate with comparison images - although I'd add that I'd like to see the test done with the image-plane off-axis on the three planes. It would also be interesting to see a test conducted at 28mm, allowing the comparison of the Q image (also in-camera corrected), as well as a film-plane optimised lens from the Leica family. I assume this is to assess the lens? Can you elaborate? An "image-plane off-axis on the three planes" meaning what exactly? At 24mm, if I understand you correctly, this will result in distortion with any lens, won't it? For argument's sake, if I get the camera and zoom lens in my hands, what image would you like me to take, and at what distance and exposure? PS - Sean Reid hasn't published his full test of the lens yet, but in his opening post he says of the 24-90 zoom "Optical performance is excellent"; similarly, Jono comments on the lens: As far as quality is concerned, Leica feel that this lens is the best zoom lens they’ve ever produced, the only compromise has been the variable aperture (f2.8-f4) which allows them to maximise the image quality. Certainly, from my experience it’s a really fantastic lens at all apertures and all focal lengths. Now, I'm not just citing what is complimentary and knocking what isn't, but I do trust that Sean and Jono know what they're doing, and that Chromasoft image does just tend to show what happens if you play with verticals and receding planes off-axis with an wide lens, doesn't it? Edited November 2, 2015 by IkarusJohn Quote Link to post Share on other sites More sharing options...
davewolfs Posted November 2, 2015 Share #14 Posted November 2, 2015 (edited) If the zoom is better than any of the primes (it won't be) I still wouldn't want to own it. Edited November 2, 2015 by davewolfs Quote Link to post Share on other sites More sharing options...
firststream Posted November 2, 2015 Share #15 Posted November 2, 2015 The traditional lens test is done shooting a flat surface. Sharpness throughout the frame, corner softness, pin-cushioning etc. can be quantitatively measured. That's fine for a lens designed to optimise on a flat recording surface such as film, but if an image can be manipulated in-camera to get rid of these weaknesses, then it's logical to take the lens design and manufacturing to a place where the recorded image on the sensor no longer matters, as it's never seen...until it's processed in-camera, and then first viewed on either the EVF or backscreen. Why not make a lens that records a distorted 21MMish field-of-view, and then pull and squeeze it into a 24mm equivalent image, especially if cost reductions can be made? For me, the problem is what happens during the corrections. The image has been corrected - simplistically - almost literally by squaring the circle. From what the Q camera has shown for the most part the corrections work very well, since most images are shot square-on, many are shot off one-axis, fewer on two; hardly any, on three. The corrections work fine under all these conditions, save the last. That's why the reviewers' photo are fine for the planes they are shot in, and why the reviewers themselves are pleased with the workings of the lens. They haven't tested the lens with the third plane off-axis. Testing three planes off-axis would push the corrections to their maximum stress. The proverbial brick wall would be shot quarter angle at 45 degrees, with the camera pointing upward 45 degrees, and the camera rotated 45 degrees. On a tripod, of course. Then, for any standard focal length, an equivalent, non-digitally-optimized prime lens would take the picture in the exactly same position for comparison. This way, all planes simultaneously recede into the other two which will allow us to see if the corrections hold up under the weight of the double keystoning. The foreshortening and keystoning of the 21MMish uncorrected lens could then be compared to the traditional keystoning and foreshortening of the prime. Ditto for the traditional softness, pincushioning, and suchlike. Really, it has to do with establishing different testing conditions and shifting traditional testing practises. Since the correctable lens no longer has to satisfy the two dimensions of the film-plane, testing it in two dimensions (the brick wall), no longer matters. Testing it under three dimensions will prove its worthiness. Three way off-axis images occur on the street, at the war front and under many photo-journalistic, photo-documentary situations. Generally speaking, under moments of spontaneous capture. Something which - as we all know! - the rangefinder particularly excels at. 1 Quote Link to post Share on other sites More sharing options...
jaapv Posted November 2, 2015 Share #16 Posted November 2, 2015 I find it weird bordering on the hilarious that reviewers are only discovering that modern lens design incorporates digital corrections now that Leica is applying the technology. Other high-end manufacturers have been doing this for quite some time, for instance Hasselblad/Zeiss. From Hasselblad's website: Digital Lens Correction Hasselblad’s modern lens design has been optimised for digital perfection, including full automatic correction of colour aberration, distortion and light fall off. Phocus makes use of its detailed knowledge of the lens design and calculates the optical corrections for every shot at the given distance and aperture setting, providing perfect images, and an ideal basis for optimal image rendering and further processing. Hasselblad digital lens correction technology works automatically with all Hasselblad H System lenses, even with tilt/shift movements and it works with all the Carl Zeiss lenses from the classic V System, using manual settings. 2 Quote Link to post Share on other sites More sharing options...
scott kirkpatrick Posted November 2, 2015 Share #17 Posted November 2, 2015 The sample pictures available at this point with .dng files (none yet from Jono and probably never from Sean) are pretty off-putting, but I suspect they were all taken at an opening show or within five minutes' walk of a Leica store. However, they do expose the digital corrections that Leica embeds in each .DNG file. With some hints from Sandy McG, I extracted them as a function of focal length on the 24-90. What they suggest is that the distortion seen at the 24mm end is about the same as in the best R "film" lenses of a decade ago at these focal lengths. The distortion corrections are gone by 50mm, and a small lateral chromatic aberration correction remains at the long end of the zoom settings. Details are in the comments at the Chromasoft blog. I see digital corrections as optional improvements over already excellent lenses. If you want a little extra corner resolution, and were careful not to put a doorframe around the edge of your picture, turn them off. Most raw developer programs have a way to do that. scott 1 Quote Link to post Share on other sites More sharing options...
firststream Posted November 3, 2015 Share #18 Posted November 3, 2015 Scott, it's an interesting discussion indeed, even if beyond my non-mathematical background. My intuition, when expressed, is nothing more than non-quantifiable opinion, but could the expected and suspected exaggeration in keystoning and foreshortening as a result of the corrections be measured mathematically? Or does one have to resort to actual two lens comparisons? Also: Were you able to count the change in the number of pixels pre- and post the correction, along the zoom range studied? Quote Link to post Share on other sites More sharing options...
scott kirkpatrick Posted November 3, 2015 Share #19 Posted November 3, 2015 (edited) I think your question boils down to whether or not straight lines in the 3D object space, which get curved when they are mapped onto the image plane by a lens with some radial distortion, will be straightened out by the corrections applied in 2D at the image plane, even if they don't lie in a plane parallel to the chip. I think they do, but I'm doing some experiments to see if anything funny happens. There's a convention for this correction, known informally as the "Adobe way." The DNG file contains a set of parameters which can be used to calculate the distance r' from the center of the chip, where a ray of red, green, or blue light will strike, if the ray should have struck the imaging chip at a distance r from its center. The DNG specification tells how to calculate r' as a function of r and these parameters for each color, and also prescribes how to reverse the distortion by mapping the actual pixel data into a larger effective pixel array with the same spacing as the actual pixels. Then the Adobe prescription is to crop the effective pixels down to the same number, 6000x4000 in this case, as you started with, so that no one is the wiser. BUT, in Capture One, at least (YMMV), I can see what is happening in the "lens profile" window and shift the crop lines out to recover some of the expanded information. With a Q file, the largest rectangle you can recover is 6463x4102 pixels (26.5 MPx) and for the SL Vario-Elmar at 24 mm focal length, it is 6476x4159 pixels (26.9 MPx). That's like a 5 to 6% shortening of the effective focal length. Or you can think of it as giving you 2.5 million pixels of extra resolution! scott Edited November 3, 2015 by scott kirkpatrick Quote Link to post Share on other sites More sharing options...
edwardkaraa Posted November 3, 2015 Share #20 Posted November 3, 2015 I certainly hope Leica will leave out the software correction in the primes. I don't really mind what they do with the zooms, they're already a compromise by definition. Quote Link to post Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.