Luminous Landscape Forum
Equipment & Techniques => Cameras, Lenses and Shooting gear => Topic started by: torger on November 25, 2015, 03:07:49 pm
-
I've made a brief search but failed to find any pixel peep testing of the TS-E 24 II used on a 5Ds or 5Ds R. Has anyone done that? I'm a bit curious how it performs. I actually own the lens myself, but I don't have the 5Ds body.
I'm not using the Canon system much, but mostly I use a 49x37mm 50MP CCD and Schneider Digitar 35 on a Linhof Techno which gives about the same field of view, so it would be interesting with a pixel peep shoot-out to see where it currently stands.
-
Not much info. I don't have either, though I continue to consider the TS-E 24 for my 6D
http://www.dpreview.com/forums/post/56043838 strange firmware problem?
http://blog.jholko.com/2015/06/18/canon-5dsr-and-canon-24mm-lens-options-what-is-the-best-canon-24mm-lens/ He likes it on the 5Dsr
-
Naturally it was one of the first things I did ;D
I have a large and cluttered bookcase that substitutes for a brick wall.
My lens performs admirably corner to corner on the 5Ds with the exception of one of the rotation positions where it is a little soft in one corner. The contrast is not eye popping unlike the Sigma 50 art or surprisingly the modestly priced Canon 40mm STM. The comparison with my 5D mark II images is a revelation resolution wise, as should be expected: the lens rises to the occasion.
Generally I would concur with the findings of the test the link to which was posted by NancyP: I also have the Canon 16-35 F4.
Exposure Simulation works as it should in live view so I'm not sure what the DPR poster's issue turned out to be.
-
But are there any pixel peep images to look at out there?
I've found some corner pixel peeps of 12mm shifted TS-E 17 on an A7r-II I think it was, and naturally that looked very blurred and smeared, but 12mm shift is extreme, and it's not the lens I was mainly interested in.
-
But are there any pixel peep images to look at out there?
I've found some corner pixel peeps of 12mm shifted TS-E 17 on an A7r-II I think it was, and naturally that looked very blurred and smeared, but 12mm shift is extreme, and it's not the lens I was mainly interested in.
I've not come across any taken with the 5Ds/r.
If I was in your shoes already owning that lens and if I was able I'd borrow or rent the camera for a couple of days and find a convenient brick wall: there are lots of permutations that would need to be explored. :)
-
But are there any pixel peep images to look at out there?
I've found some corner pixel peeps of 12mm shifted TS-E 17 on an A7r-II I think it was, and naturally that looked very blurred and smeared, but 12mm shift is extreme, and it's not the lens I was mainly interested in.
Hi Anders,
The images will have almost the same shifted corner/edge image blur on all sensor sizes, but the per-pixel-blur will scale with sensel pitch. A smaller sensel pitch will allow to extract a bit more MTF micro-contrast and limiting resolution out of the same lens, so overall per-image resolution will be higher.
Cheers,
Bart
-
Unfortunately the possibility to rent gear (camera bodies or lenses) in Sweden is not as good as in the US and I'm not that curious. I just thought that after this many months there should be some pixel peep examples out there.
I think it's hard to extrapolate results from lower resolution to higher, so it would be really nice to have with the 5Ds R rather than 5D mark II/III.
Back in 2012 I did this test myself:
http://forum.luminous-landscape.com/index.php?topic=63265.msg508962#msg508962
by shifting a 7D APS-C sensor into the "full-frame corner" to simulate how the corner would look in a full-frame 45 megapixel sensor. The conclusion back then was that the Schneider Digitar 35mm with a 39x47mm sensor is significantly sharper. That test served as one of the reasons I later bought a second hand MFD system that year, I didn't think 135 cameras would have the lens quality to work well in higher resolutions.
Since then new high resolution lenses have been released, like the Otus, Sigma Art and indeed a few Canon lenses, but still the 24 TS-E II is the king of the hill when it comes to tilt/shift 24mm wide angle lens.
Being in MFD is quite expensive though, and with the CMOS backs the wide angle compatibility has been less good than I desire so I'm constantly interested in what the latest 135 systems can do.
-
I got a pixel peep example in a private message, I can't share it in public though.
In any case my observation from that is that if you have a good copy of the lens you definitely don't need to be ashamed about the performance when shifted. It has a clear gradual sharpness loss of course, but so does the SK35 in my MFD system. I still think the SK35 with 50MP CCD has a little bit better resolution, but I need to make a better "brick wall" test shot myself to be able to make a fair comparison. My guess is that on the extreme end the TS-E/5Ds could win due to field curvature issue of the SK35, while for moderate shifts the SK35/50MP Hassy will win.
In real prints the difference is probably insignificant though, and I think you'd need the Rodenstock Digaron-W 40 or 32 to significantly exceed the performance at this field of view, but then we're talking significant money. The SK35 lens was discontinued a while ago, but you find it on the second hand market now and then for about $1700 so it's quite good bang for the buck (which you can rarely say about anything medium format...)
-
Hi,
I have a brand new 24TSE LII, which I use on a Sony A7rII. I have also tried it on my friends Canon 5DIII. My experience is that my sample is not that great.
When we tried the lens on my friends Canon 5DIII it was a bit behind my friends 16-35/4 at 24 mm. When shifted I feel it is not reasonably sharp. What I also see is that a large part of the issue is lateral chromatic aberration, that corrects well in software.
I have shot a lot on my last three weeks of travel with the TSE 24/3.5 TSE LII, and I was very happy with the results. It works very well for interiors and things like that. What causes havoc is subject with high contrast corner fine detail, like tree tops.
It would be interesting to try out more samples of the 24/3.5 TSE on Canon bodies.
But just to say, in a lot of real world situations it just worked fine. But, it also failed in a lot real world situations. I have mostly used it on Sony A7rII, but also on a Canon 5DIII and I think both cameras were consistent.
I will post some raw samples in the weekend.
Best regards
Erik
I've made a brief search but failed to find any pixel peep testing of the TS-E 24 II used on a 5Ds or 5Ds R. Has anyone done that? I'm a bit curious how it performs. I actually own the lens myself, but I don't have the 5Ds body.
I'm not using the Canon system much, but mostly I use a 49x37mm 50MP CCD and Schneider Digitar 35 on a Linhof Techno which gives about the same field of view, so it would be interesting with a pixel peep shoot-out to see where it currently stands.
-
I found a test I did in september with the 5DsR and the TS-E 24 L II at full shift.
Not the best weather - muddy light makes the contrast of detail very low by default. Converted in Lightroom. Shapening: 50 / 0.7 / 30 / 4
100% crops (click on the image to make sure you see it in 100%). f 8 / f 11 / f 16. near center / medium distance / shifted corner.
Great lens. When I apply shift I usually use f11 (f8 is too bad in the corners). The corners do profit a tiny bit from f 16 vs. f 11. The center does not suffer much from f 16 (which is astonisching). But to get the optimum I sometimes do "aperture stacking" - I blend the corners from the f16 with the rest of the image from the f11 shot.
But as torger said: if you have a good copy. Most ones are good I think. But after 5 years of heavy use I compared mine to the one of a friend - and his one was much better in the shifted corner. I bought a new one which you see in the test here. Sent the old one to CPS for adjustment - it came back with the same quality as the new one. So I had two perfect copys and sold one. You can see how much worse my original lens was in my older test on the A7R: http://forum.luminous-landscape.com/index.php?topic=93273.0. I did not know then, that it neede adjustment.
Marc
-
I assume full shift is 12mm vertical in this case (not diagonal?). In horizontal sensor orientation that puts the sensor corner at 60mm image circle diameter.
If comparing to MFD that would be about 83mm for a 48x36mm sensor and 95mm for 54x41mm.
With that in mind I'd say that the performance of the 5Ds + (good copy) TS-E 24mm II is pretty impressive.
Comparing to my SK35XL I'd say that the TS-E 24 is clearly sharper in the extreme shift, but mid-range and center the SK35XL is sharper. The SK35XL has high image quality inside 75mm image circle (corresponds to about 8mm shift in the Canon system), and then drops off very quickly and drastically outside until hard clip at 90mm. So it's a little bit apples vs oranges, the SK35 is sharper but allows smaller shifts. The Canon system is obviously superior for the larger shifts as the SK35XL does not really cope.
As 135 is 3:2 and MFD is 4:3 FOV is not easily compared, I cut the 135 frame to 32x24 (40mm diagonal) here to be comparable:
If we compare to more recent and more expensive MFD systems we have say 60 MP full-frame (54x41mm) using 40mm Rodenstock, which is about 23.6mm 135-FOV, and the 32mm which is 19mm 135-FOV. The Rodenstock 40mm is thus a good comparison in terms of FoV. However the Digarons are hard-limited to 90mm image circle which limits vertical shift with horizontal sensor to only 15mm, corresponding to about 9mm in 5Ds/TS-E 24mm terms. So to match the FoV coverage of the Canon system you need the 32mm. How well does that do near the edge? Considerably better than the 5Ds/TS-E 24mm system.
-
Hi Anders,
Here are some more or less real world samples from my 24/3.5 TSE.
http://echophoto.dnsalias.net/ekr/Articles/Shoots/TSE24_examples/20150829-_DSC0170.jpg
Two of the shots are on Canon 5DIII and the rest on Sony A7rII. Raw files included. Quite possible that I don't have a good sample.
I did some testing in my living room on test targets and they very pretty OK. Fine detail with high contrast is problematic. There is some lateral chroma, but it corrects well in Lightroom.
I have taken a lot of images I am perfectly happy with that lens.
Best regards
Erik
-
I edited and added some further comparisons with MFD alternatives in my previous post. What's really impressive with the 5Ds/TS-E combo is that you have so large shift range, MFD tech cams actually has some difficulty to compete there. For smaller shifts also the older MFD lenses have the edge.
(With the smaller 44x33mm format MFD gets more relative shift range, almost exactly the same as 5ds/TS-E, but those newer sensors like the Sony CMOS one has some angular response issues making it unfeasible to use the full image circle of the wides anyway.)
As I do landscape rather than architecture I don't have the need of very large shifts on the wides, I avoid them as the perspective distortion gets a bit too large and disturbs the composition, so the SK35XL smaller shift range is fine with me. If I would do architecture where you're from time to time forced to employ large shifts the SK35XL would feel limiting though. I could get the SK28XL, probably the most extreme digital lens ever made due to it's symmetric design and short focal length; when shifted basically only works with light-shielded sensors like the Kodak 50MP sensor I have on my Hassy back, it doesn't work well with any of the recent Phase One backs. I haven't personally tested the SK28XL but from test shots I've seen my indications is that it doesn't have the same sharp falloff like the SK35XL, so more of the 90mm image circle is actually usable.
I don't know for sure, but I think it's likely to see more issues with lens sample variations here on the wide end with large shifts involved due to very high parallelism requirements.
-
I use the 24tse on a Sony A7r2 and i use as adapter for the canon to the emount a mirex shift adapter, which results in a second shift layer, which can be set to horizontal or vertical shift. It allows 20mm of shift and i can compose much better, using one shift layer for the horizontal movements and the other shift layer for the vertical movements. The lens is still sharp with 20mm of shift, stopping it down to f13 or 16. Very impressive.
-
I use the 24tse on a Sony A7r2 and i use as adapter for the canon to the emount a mirex shift adapter, which results in a second shift layer, which can be set to horizontal or vertical shift. It allows 20mm of shift and i can compose much better, using one shift layer for the horizontal movements and the other shift layer for the vertical movements. The lens is still sharp with 20mm of shift, stopping it down to f13 or 16. Very impressive.
Hi Rainer
Very interesting setup! To control horizontal and vertical shift with 135 was always a compromise.
I was on the Mirex website, but could not find an adapter from Canon to the e-mount. Could you please
post a link to this adapter?
Thanks,
Ben
-
I could get the SK28XL, probably the most extreme digital lens ever made due to it's symmetric design and short focal length; when shifted basically only works with light-shielded sensors like the Kodak 50MP sensor I have on my Hassy back, it doesn't work well with any of the recent Phase One backs. I haven't personally tested the SK28XL but from test shots I've seen my indications is that it doesn't have the same sharp falloff like the SK35XL, so more of the 90mm image circle is actually usable.
Torger
I have the SK28XL since 2 weeks now and it's killer! It's a completely other league than the SK35XL (that I also like).
I use it with Kodak P25+ and Dalsa P40+ up to 20 mm shift with good results. (CF is necessary)
Regards,
Ben
-
Hi Rainer
Very interesting setup! To control horizontal and vertical shift with 135 was always a compromise.
I was on the Mirex website, but could not find an adapter from Canon to the e-mount. Could you please
post a link to this adapter?
Thanks,
Ben
I phoned them, asked for it and ordered it directly. Dont know why not listed on their site.
Its a new feel to work with two shift layers with 35mm.
I bought as well kippon shift adapters to canon mount for my pentax 645 55&75mm lenses and a canon adapter for my contax35mm pc lense. So i have now a range of shift lenses from 17 (tse) over 24/35/55/75 up to 120 mm ( hasselblad lense ) , all with horizontal and vertical shifts. Its just a great setup ! In the bag is as well the great and tiny voigtländer 12mm and a canon 80-400 lense with metabones4 adapter. Not much things i cant do with this set. Disadvantage but good doable is that the canons have to be used in stopped down mode, i use them with f13.
-
I phoned them, asked for it and ordered it directly. Dont know why not listed on their site.
Its a new feel to work with two shift layers with 35mm.
I bought as well kippon shift adapters to canon mount for my pentax 645 55&75mm lenses and a canon adapter for my contax35mm pc lense. So i have now a range of shift lenses from 17 (tse) over 24/35/55/75 up to 120 mm ( hasselblad lense ) , all with horizontal and vertical shifts. Its just a great setup ! In the bag is as well the great and tiny voigtländer 12mm and a canon 80-400 lense with metabones4 adapter. Not much things i cant do with this set. Disadvantage but good doable is that the canons have to be used in stopped down mode, i use them with f13.
Ranier - Does the 55mm Pentax have a manual aperture and have you compared the Pentax to the Hassleblad 50 ? Thanks
-
i didnt compared them directly ( mostly i am too lazy for pixelpeeping ...) cause the pentax is very good and the hassy as well.... so i take the lighter lense which is the pentax.
-
i didnt compared them directly ( mostly i am too lazy for pixelpeeping ...) cause the pentax is very good and the hassy as well.... so i take the lighter lense which is the pentax. Both have manual aperture.
-
This is one file. I haven't had time to go through the various frames of the scene I have so this might not represent the best of the lot. The crops are 100%. The Detail setting in ACR was set to 80, Sharpening Amount to 30 with Radius 1. No other corrections were imparted to the file (except a slight leveling of the horizon).
5DS + TS-E 24L II, @ f/8. Some Tilt on the lens (how much, I don't remember).
-
Centre crop.
-
Torger
I have the SK28XL since 2 weeks now and it's killer! It's a completely other league than the SK35XL (that I also like).
I use it with Kodak P25+ and Dalsa P40+ up to 20 mm shift with good results. (CF is necessary)
Regards,
Ben
I'm looking into getting the SK28XL too at some point (I have test shots and I see that it has a larger high quality image circle than the SK35XL despite being even wider angle). I'm not shooting architecture much though (I do mostly landscape) so the 75mm high quality image circle of the SK35XL serves me fine for the moment. The problem with the SK28XL is that due to it's symmetric design it's not compatible with that many sensors, but it is indeed compatible with my H4D-50 back (the Kodak KAF-51000) so it would be nice to get that lens at some point. Some still use it with the 6um Dalsa sensors (P65+, IQ160 etc) but then maximum shift is only a few millimeters until color cast becomes unrepairable, but for those full-frame 645 sensors you can still get more shift with the SK28XL than the Rodenstock Digaron-S 28mm which has its image circle hard-limited to 70mm...
Medium format tech cams have several excellent wide angles, but if you are used with the huge shift ranges available from TS-E 24 and TS-E 17 with say a 5Ds you may be disappointed. To get most use of the medium format alternative you need to keep to a bit lower amounts of shift (corresponding to 7-8mm on the TS-Es with a 5Ds). Within that range you will get higher resolution from the MFD tech cameras, assuming you have more megapixels on the sensor that is.
-
I have the TSE 24mm F3.5 MK2 and have been using it on the 5Ds for a few months now. It’s very good but as has been said before it does not have the same punch as lenses like the Sigma 50mm F1.4 Art. It is at is best between f5.6 and f8.0 in terms micro contrast when used with modest amounts of shift. For heavy shift use you need f8.0 to f11 to ensure good corners.
Focusing with live view is not as easy as it is on the old 5d Mk2 but this I am sure is mainly to do with the line skipping the 5d2 did for live view, made the contrast really pop when you hit focus in live view so really easy to set up manual focus lenses like the TSE. I have found that using tilt requires a very high level of precision with the 5Ds, any error in your technique is ruthlessly shown by the 50mp camera. The precision required is at the very limit of how controllable the tilt mechanism is in practice, movements have to be very considered. While I have not shot many frames at full shift with it I have been impressed with those I have. The crop below is from the top of a frame with 11mm of shift, shot at F9.0. No tilt was used. 100% Crop of top left part of the image shown together with low res version of the whole frame.
Processing in LR CC sharpening 91, 0.6px, detail 26, masking 30, luminance NR 18, Colour 25
Hope this is useful to you all.
Ben
-
As tested from the-digital-pcture.com:
http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=486&Camera=979
-
Curiously, a lot of responses here seem to say "maybe I just don't have a good copy". I would add me to that list. What does that tell you? Buy from company with 30 day return policy or avoid the lens altogether. Test immediately, while you can still send it back. I have the I; maybe the II version is radically improved. Imagine a zeiss or sigma art T/S lens, would be nice.
-
Curiously, a lot of responses here seem to say "maybe I just don't have a good copy". I would add me to that list. What does that tell you? Buy from company with 30 day return policy or avoid the lens altogether. Test immediately, while you can still send it back. I have the I; maybe the II version is radically improved. Imagine a zeiss or sigma art T/S lens, would be nice.
The TS-E 24 Mark I is so much worse than the mark II. That is not a question of a good or a bad copy.
-
I have the TSE 24mm F3.5 MK2 and have been using it on the 5Ds for a few months now. It’s very good but as has been said before it does not have the same punch as lenses like the Sigma 50mm F1.4 Art. It is at is best between f5.6 and f8.0 in terms micro contrast when used with modest amounts of shift. For heavy shift use you need f8.0 to f11 to ensure good corners.
Focusing with live view is not as easy as it is on the old 5d Mk2 but this I am sure is mainly to do with the line skipping the 5d2 did for live view, made the contrast really pop when you hit focus in live view so really easy to set up manual focus lenses like the TSE. I have found that using tilt requires a very high level of precision with the 5Ds, any error in your technique is ruthlessly shown by the 50mp camera. The precision required is at the very limit of how controllable the tilt mechanism is in practice, movements have to be very considered. While I have not shot many frames at full shift with it I have been impressed with those I have. The crop below is from the top of a frame with 11mm of shift, shot at F9.0. No tilt was used. 100% Crop of top left part of the image shown together with low res version of the whole frame.
Processing in LR CC sharpening 91, 0.6px, detail 26, masking 30, luminance NR 18, Colour 25
Hope this is useful to you all.
Ben
Is it my eyes, or is there something pretty odd going on in the first shot? I get the distinct impression of parts swelling outwards compared to areas beside them. Somewhat like looking through a piece of bad glass.
Rob C
-
Is it my eyes, or is there something pretty odd going on in the first shot? I get the distinct impression of parts swelling outwards compared to areas beside them. Somewhat like looking through a piece of bad glass.
Rob C
A known perception problem with t/s lenses. Our eyes/brain do not expect such tall objects to be straight when viewed from the ground, they expect some perspective distortion. Hence the mistake most t/s users make by creating geometrically absolute verticals, when a slight distortion would be sufficient. Ancient Greeks knew that better than modern photographers, so they built their columns accordingly.
-
A known perception problem with t/s lenses. Our eyes/brain do not expect such tall objects to be straight when viewed from the ground, they expect some perspective distortion. Hence the mistake most t/s users make by creating geometrically absolute verticals, when a slight distortion would be sufficient. Ancient Greeks knew that better than modern photographers, so they built their columns accordingly.
It would be interesting to conduct a double blind test to quantify what is preferable: a picture of a tall building that is 100% "corrected" versus one that has a slight amount of convergence towards the top. Slobodan, do you take this into consideration with your work? You have an excellent eye for architecture.
-
A known perception problem with t/s lenses. Our eyes/brain do not expect such tall objects to be straight when viewed from the ground, they expect some perspective distortion. Hence the mistake most t/s users make by creating geometrically absolute verticals, when a slight distortion would be sufficient. Ancient Greeks knew that better than modern photographers, so they built their columns accordingly.
Yes, that's so, but in this case I think it's something apart from that problem of expectations.
I felt the same with some shots using my old 35mm shifter Nikkor in France. I'll try to find an image that shows what you refer to above. Got it:
(http://www.roma57.com/uploads/4/2/8/7/4287956/3204588_orig.jpg)
The effect is most noticeable near the centre - a kind of bulging, rather than a problem with the concept of verticals not converging. Maybe it's really all the same thing - just feels different depending on the specific shot. I've seen many architectural shots in my day, and the vast majority, done properly, do not betray themselves in this way. Perhaps it's to do with lenses that tilt/shit instead of having the camera perform the tilts/shifts.
My Australian correspondent, Walter, uses a 24mm TS-type Canon, and as far as I remember, he doesn't think much of the tilting ability on such a tiny format; he comes from 8x10... The problem is seeing well enough what you are trying to do. However, shift works well.
Rob
-
... The effect is most noticeable near the centre - a kind of bulging, rather than a problem with the concept of verticals not converging....
I used converging verticals as an example, but there are all sorts of other optical distortions involved. In Greek architecture, it is known as "entasis," if I am not mistaken.
-
... Slobodan, do you take this into consideration with your work? You have an excellent eye for architecture.
The simplest trick is to hide your standpoint (ground), if you are actually standing at the street level. That way you can correct verticals to be perfectly vertical, as eyes/brain do not have a reference point and assume you are viewing it from an elevated standpoint, e.g. through a window on a neighboring building. If I can not do that, I do not mind leaving converging verticals as such.
-
Hi Torger,
I have a couple of photographs similar to the below that are shot with the TSE 24mm and 5dsr and stitched horizontally and then diagonally up (the subject is in the centre frame). I do this when I know that I want to produce an imageof 16:9 or 2:1 proportions and don't wish to crop.
These total approx 6000x12000px and should give you a pretty good idea of the range of the lens (well my copy anyway)
24mmTSE @ F8-Stitched.
(http://www.andrewlatreille.com/wp-content/uploads/2015/12/LAP0171_0294-2_2000px.jpg)
Happy to send through a tiff for your view, or if you'd like I could send you some of the individual raws.
Also a couple of recent examples shot just last week with the 5dsr and respective lenses. Lots of find detail on these facades.
These are pretty much straight out of the camera, so there's a little retouching on each that will occur if the client selects the images.
(http://www.andrewlatreille.com/wp-content/uploads/2015/12/LAP0164_LethHosp_0496-2-2000px.jpg)
This shot with the 24 @ F11
(http://www.andrewlatreille.com/wp-content/uploads/2015/12/LAP0164_LethHosp_1770-2-2000px.jpg)
This shot with the 17 @ F11
Please pm me if your still looking for examples.
Lots of bricks in all of these so pixel peepers should like this.
;-)
AL
-
I always leave a residual keystoning to my architecture shots. At times I do too with interior photography and I do it both ways. When shifting the lens down and when in shifting up. To me the pictures look better that way. Not only more natural but better.
So far I haven't had an architect or media creative to object this (touch wood). Sometimes with certain subjects and/or angles, I use the Wrap Tool to decrease the higher floors. That to me they look much better when needed because to my eyes, it looks kind of moronic to look at a 10-story building with the 10th floor windows looking as tall as the ones at ground level. The effect can be even more pronounced after the brain does its math.
If my 6D files can withstand this "extreme" treatment, any modern digital camera file can.
Eduardo
It would be interesting to conduct a double blind test to quantify what is preferable: a picture of a tall building that is 100% "corrected" versus one that has a slight amount of convergence towards the top. Slobodan, do you take this into consideration with your work? You have an excellent eye for architecture.