Pages: 1 [2]   Go Down

Author Topic: Tilt Shift Lenses on Sony A7RII  (Read 14484 times)

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Tilt Shift Lenses on Sony A7RII
« Reply #20 on: October 08, 2016, 03:18:22 am »

Hi M,

I cannot talk about the Sigma 24 Art as I don't have that lens.

The 24/3.5 TSE LII was the first lens I bought for the Sony A7rII I had it long before the I bought my HCam Master TSII. I mostly use it for convenience.

I would probably not buy a 24/3.5 TSE now.

Best regards
Erik

Erik,

What, if any, are the IQ differences between the Canon TS and Stefan Steib's HCam V2 mounted with a Sigma 24 Art alternative ? The only 'inconvenience' with Stefan's solution is having to preset the aperture before mounting it - personally, I can live with that.

M
Logged
Erik Kaffehr
 

Manoli

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2296
Re: Tilt Shift Lenses on Sony A7RII
« Reply #21 on: October 08, 2016, 05:17:47 am »

Is there something wrong with your TS-E 24L II? That lens is super-sharp when unshifted. Even when shifted, it's sharp for something with such a wide angle of view. I just wish it had either 7 or 9 aperture blades instead of 8, as well as some lens correction profiles available to get rid of CA.

Checked the Lenscore website and their results corroborate Erik's experience - the 16-35/4 outresolves the Canon 24 TSE. The Sigma outresolves them both.

Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2839
Re: Tilt Shift Lenses on Sony A7RII
« Reply #22 on: October 08, 2016, 06:38:12 am »

Checked the Lenscore website and their results corroborate Erik's experience - the 16-35/4 outresolves the Canon 24 TSE. The Sigma outresolves them both.

The Lenscore numbers show so much incongruence with real-world experience that I find them impossible to believe. Furthermore, they do not publish their testing methodology. Do they compare all lenses wide-open (i.e. a wide-aperture lens would be tested at f/1.4 vs f/4, or even f/5.6, for some lenses), or all at f/6.3, or some other number that every lens can reach? Do they compare them at multiple apertures and focal lengths and average, them, or just take the best one? Which end of the focal length of zooms are they testing? Are they comparing centres, corners, edges or what?

This is a website that says that the Canon 100-400 (the original Dust Trombone) is sharper than the Canon 85/1.8 or 400/5.6, and even the Canon 24-70L II (which is renowned for sharpness and often compared to a 'bag of primes'). It rates the awful Canon 14/2.8 as sharper than both the 400/5.6 supertele prime (almost impossible to believe) and the 24-70/4.0 zoom, and even the Zeiss Distagon 18/3.5 (which, although Zeiss' worst lens, is not exactly a slouch). It rates the Canon 135/2.0 (well-regarded for sharpness) well below the Canon 50/1.2 (well-known as a less-than-razor-sharp lens) and the 16-35/2.8 II, with its near-unusable corners, far higher than the critically-sharp 85/1.8 and even the TS-E 24L II (whose corners on a full-frame sensor are still close to the middle of its image circle). It even rates Nikon's much-maligned, horribly-soft PC-E 24/3.5 as sharper than the Nikon 85/1.4G, Zeiss Distagon 21/2.8 and Canon 70-200/2.8 II (again, often regarded as a 'bag of primes'), and almost 300 points higher than the Canon TS-E 24L II (which, by every other account, is a far superior lens). The fact that they don't publish their methodology or their measured data would render them suspect even if the numbers correlated with real-world experience, but, in many cases, they don't even do that.

Photozone.de has a far better set of numbers for resolution, where they test lenses at multiple apertures, multiple focal lengths and different parts of the frame and publish the raw results, in lp/mm, at each and every point. You sometimes need to convert the lp/mm measurements to equivalents to deal with test sensors of different resolutions and be able to recognise when sensor resolution, rather than lens resolution, is the limiting factor in a particular test, but the raw data gives you a much better idea of performance than Lenscore's arbitrary and derived numbers.
Logged

BobDavid

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3307
Re: Tilt Shift Lenses on Sony A7RII
« Reply #23 on: October 08, 2016, 10:24:54 am »

Clyde Butcher uses Cambo Actus on his Sony A7 with medium format lenses.

The Cambo Actar 24mm lens is a perfect solution for A7X cams and the Cambo Actus. I'm selling my Actar (see the "For Sale") section.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Tilt Shift Lenses on Sony A7RII
« Reply #24 on: October 08, 2016, 10:54:15 am »

Hi,

I would agree that Photozone is a good site. With regard to Lenscore I don't feel it is wise to put a single figure of merit on things.

One issue with Photozone is that they compensate for field curvature. According the Photozone the 24/3.5 TSE LII would be a better lens than the 16-35/4L, but there would not be a lot between.

Another good site for Canon lenses is The Digital Picture.

I use my 24/3.5 TSE LII a lot for shifts, for straight photography I would use the 16-35/4L or even the 24-105/4L.

For me, 16-35 is shift territory and longer than that is tilt territory.

The image below was probably shot with the Canon 16-35/4L at around 20 mm on the HCam Master TSII:


This was probably also  Contax 28-85/3.3-4 on the HCam Master TSII:


And this one is Contax 35-135/3.3-4.5 at the long end:


Best regards
Erik
The Lenscore numbers show so much incongruence with real-world experience that I find them impossible to believe. Furthermore, they do not publish their testing methodology. Do they compare all lenses wide-open (i.e. a wide-aperture lens would be tested at f/1.4 vs f/4, or even f/5.6, for some lenses), or all at f/6.3, or some other number that every lens can reach? Do they compare them at multiple apertures and focal lengths and average, them, or just take the best one? Which end of the focal length of zooms are they testing? Are they comparing centres, corners, edges or what?

This is a website that says that the Canon 100-400 (the original Dust Trombone) is sharper than the Canon 85/1.8 or 400/5.6, and even the Canon 24-70L II (which is renowned for sharpness and often compared to a 'bag of primes'). It rates the awful Canon 14/2.8 as sharper than both the 400/5.6 supertele prime (almost impossible to believe) and the 24-70/4.0 zoom, and even the Zeiss Distagon 18/3.5 (which, although Zeiss' worst lens, is not exactly a slouch). It rates the Canon 135/2.0 (well-regarded for sharpness) well below the Canon 50/1.2 (well-known as a less-than-razor-sharp lens) and the 16-35/2.8 II, with its near-unusable corners, far higher than the critically-sharp 85/1.8 and even the TS-E 24L II (whose corners on a full-frame sensor are still close to the middle of its image circle). It even rates Nikon's much-maligned, horribly-soft PC-E 24/3.5 as sharper than the Nikon 85/1.4G, Zeiss Distagon 21/2.8 and Canon 70-200/2.8 II (again, often regarded as a 'bag of primes'), and almost 300 points higher than the Canon TS-E 24L II (which, by every other account, is a far superior lens). The fact that they don't publish their methodology or their measured data would render them suspect even if the numbers correlated with real-world experience, but, in many cases, they don't even do that.

Photozone.de has a far better set of numbers for resolution, where they test lenses at multiple apertures, multiple focal lengths and different parts of the frame and publish the raw results, in lp/mm, at each and every point. You sometimes need to convert the lp/mm measurements to equivalents to deal with test sensors of different resolutions and be able to recognise when sensor resolution, rather than lens resolution, is the limiting factor in a particular test, but the raw data gives you a much better idea of performance than Lenscore's arbitrary and derived numbers.
« Last Edit: October 08, 2016, 11:08:58 am by ErikKaffehr »
Logged
Erik Kaffehr
 

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2839
Re: Tilt Shift Lenses on Sony A7RII
« Reply #25 on: October 08, 2016, 12:51:08 pm »

Hi,

I would agree that Photozone is a good site. With regard to Lenscore I don't feel it is wise to put a single figure of merit on things.

One issue with Photozone is that they compensate for field curvature. According the Photozone the 24/3.5 TSE LII would be a better lens than the 16-35/4L, but there would not be a lot between.

Unless all you shoot is brick walls or your main purpose is copying documents or two-dimensional artworks, not compensating for field curvature is just as bad as doing so.

If you care about what's in the corners (especially if you care about what's in all four of the corners at the same time), you're probably shooting architecture, landscapes or large group portraits, where you want the whole frame to be in focus. Unless it's astrophotography, you're probably shooting stopped down, in order to get everything reasonably sharp at the same time. You need to do so because the things you want to be sharp (i.e. all of it) are at varying distances. Exactly what shape the plane of focus takes within this volume of space doesn't actually matter, because your subjects are scattered in front of and behind it anyway - whether it's flat or curved, many objects will be in front of it and many will be behind it, and you will need to stop down far enough to make everything acceptably sharp anyway.

Obviously, field curvature becomes a problem when it's so bad that corner objects at infinity are out of focus at f/6.3 when the middle is focused well past infinity, but few lenses are that bad. It would be useful to have that information as a separate measurement. But just rolling it into the 'sharpness' score and not compensating for it makes the overall 'sharpness' metric unreliable - it makes the corners seem far less sharp than they actually are. Even with bad field curvature, the corners might be very sharp - just that they won't be sharp at the same time as the centre. Which becomes a problem when shooting large, flat objects, but not a problem if you're trying to shoot a portrait with the subject centred around one of the rule-of-thirds intersections rather than the middle, for instance.

Incidentally, this is another argument in favour of super-high-resolution sensors, that can resolve light at well beyond the optical limits of the lens. This extra resolution allows for more accurate deconvolution, both for defocus and for diffraction, provided a profile is made for the lens at each particular focal length, aperture and tilt/shift setting, allowing for the maximum resolution of the lens to be used at all times and allowing for things such as altered (or even multiple) planes of focus. There's a lot you can do with fast computers and enough data...
Logged

David Eichler

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 823
    • San Francisco Architectural and Interior Photographer
Re: Tilt Shift Lenses on Sony A7RII
« Reply #26 on: October 08, 2016, 01:43:45 pm »


One issue with Photozone is that they compensate for field curvature.


I think it is perfectly reasonable to account for field curvature in the resolution test results, as long as the review notes the degree of field curvature that is present, which Photozone does.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Tilt Shift Lenses on Sony A7RII
« Reply #27 on: October 10, 2016, 01:20:00 am »

Hi,

I have checked out both test chart data and MTF data at the "digital picture com" and it certainly seems to be the case that the 24/3.5TSE would a better performer than the 16-35/4L.

So, I guess that I have a good sample of the 16-35/4 and a less good sample of the 24/3.5TSE LII. The 16-35/4 looses a lot somewhere around 14-16 mm of axis according to the MTF data. I see a sharp drop of performance in the extreme corners at 24 mm.

In the initial tests I made the 16-35/4L was much sharper, but I am quite happy with the 24/3.5 TSE LII when shooting shifted.

There is a significant spread in performance between different samples, see attached chart from LensRentals.

Best regards
Erik


Is there something wrong with your TS-E 24L II? That lens is super-sharp when unshifted. Even when shifted, it's sharp for something with such a wide angle of view. I just wish it had either 7 or 9 aperture blades instead of 8, as well as some lens correction profiles available to get rid of CA.
Logged
Erik Kaffehr
 

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: Tilt Shift Lenses on Sony A7RII
« Reply #28 on: October 10, 2016, 06:20:44 am »

I think it is perfectly reasonable to account for field curvature in the resolution test results, as long as the review notes the degree of field curvature that is present, which Photozone does.

I agree. Obviously, especially for flat surfaces, it is more of a concern than for objects placed in 3D-space. And for subjects with depth of field one will use narrower apertures, which will also mask the largest differences at a given distance.

What's more, for the TS-E 24mm f/3.5L II lens in shifted mode, one would use apertures like f/11 or f/16 to balance the sharpness from center to corner. At its sharpest aperture f/4.5, DOF is an issue for many subjects unless it's a more-or-less flat plane that can be focused with tilt.

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

Ghibby

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Re: Tilt Shift Lenses on Sony A7RII
« Reply #29 on: October 10, 2016, 08:21:16 am »

Is there something wrong with your TS-E 24L II? That lens is super-sharp when unshifted. Even when shifted, it's sharp for something with such a wide angle of view. I just wish it had either 7 or 9 aperture blades instead of 8, as well as some lens correction profiles available to get rid of CA.

While 7 or 9 aperture blades would be good you should evaluate the performance of both the 24mm TS-E mk2 and the 17mm TS-E especially at apertures up to apx f8 with regard to sunstars. 

You will find that they will create 16 pointed stars.  I stumbled upon this a while ago and looked at the aperture design of these lenses very closely to try and understand why this was happening. My understanding is that it is caused by the specific curved aperture blade design used by Canon.  The profile of these blades is actually made of 2 distinct curves in a bid to maintain a more circular aperture at more settings.  The junction of these curves on each blade is just sudden enough to cause an extra spike on the sunstar but only at apertures up to apx f8.0. Beyond this the extra curve in the blade is obscured by the neighbouring overlapping blade. I find on my copies it is most noticeable at f7.1 so when I am shooting scenes were I want nice sunstars I use these apertures and focus / tilt with as much precision as possible, much more difficult to do than at f11 where these lenses are so often used. 

It is more distinct on the 17mm than the 24mm lens but this could be copy to copy variation so your milage may differ. You will also see similar behaviour on another popular lens the 24-105F4L IS but it is no where near as distinct as the TS-E lenses. I am sure there are other 8 bladed lenses from canon out there that will behave in this way too if they use similar designs for their aperture blades.

See images of my xmas tree last year when I first noticed this. First is shot at f7.1 second at f16.  You can clearly see the 16 point sunstar a f7.1, see also images of the actual lens aperture closed down to f7.1 and f 16 and you can clearly see the shape of the iris that caused this effect at f7.1 on this lens.

Hope this is useful you you all, it's rather a lovely nuance of these lenses that seemingly flies in the face of convention but yields some fantastic night time opportunities.

You can also control any residual CA very nicely in extreme shifts with these lenses with some careful use of the defringe tool in lightroom, the auto CA correction often makes a mess of CA of high contrast scenes with lots of shift applied towards the edges of the frame.

Ben
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2839
Re: Tilt Shift Lenses on Sony A7RII
« Reply #30 on: October 10, 2016, 09:00:34 am »

While 7 or 9 aperture blades would be good you should evaluate the performance of both the 24mm TS-E mk2 and the 17mm TS-E especially at apertures up to apx f8 with regard to sunstars. 

You will find that they will create 16 pointed stars.  I stumbled upon this a while ago and looked at the aperture design of these lenses very closely to try and understand why this was happening. My understanding is that it is caused by the specific curved aperture blade design used by Canon.  The profile of these blades is actually made of 2 distinct curves in a bid to maintain a more circular aperture at more settings.  The junction of these curves on each blade is just sudden enough to cause an extra spike on the sunstar but only at apertures up to apx f8.0. Beyond this the extra curve in the blade is obscured by the neighbouring overlapping blade. I find on my copies it is most noticeable at f7.1 so when I am shooting scenes were I want nice sunstars I use these apertures and focus / tilt with as much precision as possible, much more difficult to do than at f11 where these lenses are so often used. 

It is more distinct on the 17mm than the 24mm lens but this could be copy to copy variation so your milage may differ. You will also see similar behaviour on another popular lens the 24-105F4L IS but it is no where near as distinct as the TS-E lenses. I am sure there are other 8 bladed lenses from canon out there that will behave in this way too if they use similar designs for their aperture blades.

See images of my xmas tree last year when I first noticed this. First is shot at f7.1 second at f16.  You can clearly see the 16 point sunstar a f7.1, see also images of the actual lens aperture closed down to f7.1 and f 16 and you can clearly see the shape of the iris that caused this effect at f7.1 on this lens.

Hope this is useful you you all, it's rather a lovely nuance of these lenses that seemingly flies in the face of convention but yields some fantastic night time opportunities.

You can also control any residual CA very nicely in extreme shifts with these lenses with some careful use of the defringe tool in lightroom, the auto CA correction often makes a mess of CA of high contrast scenes with lots of shift applied towards the edges of the frame.

Ben

I've realised that - good think I rarely shoot past f/6.3 or f/7.1, except when shooting landscapes with telephotos, as diffraction begins to take an effect there with Sony's 42MP and 36MP sensors. But it would be good to be able to produce good sunstars stopped down further as well. IMO no lens should ever be released with an even number of blades - it's such an easy thing to change, yet has such a profound effect on many shots, including almost every shot taken at night in a city.

What defringe tool? I'm still running PS CS3, and never go anywhere near lightroom (I shoot low-volume, high-detail work that needs a lot of masking).
Logged

marc aurel

  • Full Member
  • ***
  • Offline Offline
  • Posts: 150
Re: Tilt Shift Lenses on Sony A7RII
« Reply #31 on: October 10, 2016, 10:05:56 am »

I like the sunstars of the TS-E 24mm II very much. Yes it is an even number of blades, but the special design of the blades that Ben described makes very pleasing sunstars. My favourite lens in that respect.

I added images of comparisons betwenn TS-E 17mm, TS-E 25mm, PC-Distagon 35mm and Canon 24-70mm f2.8 L II at different apertures.
The 24-70mm zoom has 9 blades, but they are so extremely pronounced. I had a customer who complained massively about an image just because of that. Some may like it, but it dominates an image too much in my eyes.

Regards -
Marc
Logged

Ghibby

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Re: Tilt Shift Lenses on Sony A7RII
« Reply #32 on: October 10, 2016, 10:25:06 am »

What defringe tool? I'm still running PS CS3, and never go anywhere near lightroom (I shoot low-volume, high-detail work that needs a lot of masking).

Hi Shadowblade,

Defringe tool takes the 2 typical colours of optical fringing, ie magenta and green, offers a range of hues around these that you can pick with an eye dropper or manually select / alter and then desaturates them on edges in a controllable way.  It works extremely well in most images, it can be caught out in images dominated by green or magenta colours but thats about its only drawback.  I suggest you have a look at a trial of the most up to date process version of ACR / Lightroom if you are still on the version of camera raw that shipped with CS3, image quality and what can be extracted from a RAW file in the latest versions of camera raw are way ahead of what was possible in the CS3 era. Regardless of volume of work it will surely make masking and selecting easier on an image by image basis, worth a free 30 day trial in any case I would say.

Ben
Logged

Ghibby

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Re: Tilt Shift Lenses on Sony A7RII
« Reply #33 on: October 10, 2016, 10:31:50 am »

Hi Marc,

I remember talking about this on a previous forum topic with you about this.  Interesting optical characteristics, I think the 17 is the pick of the bunch for nice but not OTT sunstars.  It is my go to wide angle lens for night time shots as well.  The sunstars on the 24tse look a lot hazier and less distinct by comparison, not really dazzling enough to make a special effort to use an aperture that is not optimal for the scene in my opinion.  I do like the 24-70 sunstars, similar looking to what you get with the 100-400mk2 aswell.  They can look fantastic but also dominate an image too much if not used carefully.

Ben
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2839
Re: Tilt Shift Lenses on Sony A7RII
« Reply #34 on: October 10, 2016, 11:52:27 am »

Hi Shadowblade,

Defringe tool takes the 2 typical colours of optical fringing, ie magenta and green, offers a range of hues around these that you can pick with an eye dropper or manually select / alter and then desaturates them on edges in a controllable way.  It works extremely well in most images, it can be caught out in images dominated by green or magenta colours but thats about its only drawback.  I suggest you have a look at a trial of the most up to date process version of ACR / Lightroom if you are still on the version of camera raw that shipped with CS3, image quality and what can be extracted from a RAW file in the latest versions of camera raw are way ahead of what was possible in the CS3 era. Regardless of volume of work it will surely make masking and selecting easier on an image by image basis, worth a free 30 day trial in any case I would say.

Ben

I've never used Camera Raw. I use Sony's RAW converter - it does a better job of Sony RAW files than any version of Camera RAW or Lightroom. It's a pig-dog-bear crossbreed to use, though - I wouldn't use it if I were processing hundreds of images per shoot instead of a few select ones.
Logged

Ghibby

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Re: Tilt Shift Lenses on Sony A7RII
« Reply #35 on: October 10, 2016, 12:07:43 pm »

Have you tried capture one pro?
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2839
Re: Tilt Shift Lenses on Sony A7RII
« Reply #36 on: October 10, 2016, 12:18:08 pm »

Have you tried capture one pro?

No. Does it give me anything in the way of a better conversion, i.e. more DR, less noise, etc.? Or is it just a better interface to work with?

My workflow usually involves very little time in RAW conversion and hours of painstaking masking and layers work in Photoshop, so a less-clumsy interface doesn't really save me much time.
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Tilt Shift Lenses on Sony A7RII
« Reply #37 on: October 10, 2016, 12:18:28 pm »

Hi Ben,

That green magenta fringing is know as axial chroma and occurs in out of focus areas. It is often masked by stopping down.

There is another for of chromatic aberration called lateral chroma which is often red-green that often occurs at sharp edges and increases towards the corners. Both Lightroom and C1 can compensate for this automagically.

Best regards
Erik


Hi Shadowblade,

Defringe tool takes the 2 typical colours of optical fringing, ie magenta and green, offers a range of hues around these that you can pick with an eye dropper or manually select / alter and then desaturates them on edges in a controllable way.  It works extremely well in most images, it can be caught out in images dominated by green or magenta colours but thats about its only drawback.  I suggest you have a look at a trial of the most up to date process version of ACR / Lightroom if you are still on the version of camera raw that shipped with CS3, image quality and what can be extracted from a RAW file in the latest versions of camera raw are way ahead of what was possible in the CS3 era. Regardless of volume of work it will surely make masking and selecting easier on an image by image basis, worth a free 30 day trial in any case I would say.

Ben
Logged
Erik Kaffehr
 

Ghibby

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 72
Re: Tilt Shift Lenses on Sony A7RII
« Reply #38 on: October 10, 2016, 12:40:05 pm »

Hi Ben,

That green magenta fringing is know as axial chroma and occurs in out of focus areas. It is often masked by stopping down.

There is another for of chromatic aberration called lateral chroma which is often red-green that often occurs at sharp edges and increases towards the corners. Both Lightroom and C1 can compensate for this automagically.



Thanks Eric, I'm no optical expert but I understand the principal causes of the various optical aberrations. However you do get some extreme versions of these aberrations that are not well compensated for with the auto CA option in the adobe 2012 process version for ACR and LR.

In fact with moderate to heavy shifts on the 17 and 24mm TSE lenses you often find that while the auto CA button makes things better in parts of the frame as you get to the extreme edge it can often make things a lot worse and add significant fringing where there was almost none.  In these cases I switch off the auto CA and and use the defringe tool to address the problem.  I know the original design intent of the defringe tool was to remove the fringing caused by axial coma in OOF areas (principally specular highlights) but it works equally well at zapping lateral chroma on sharp contrasty edges, especially when using very targeted hue settings and low saturation levels. As Adobe removed the old style CA controls with process version 2012 its the only way of doing this currently.

Ben
Logged
Pages: 1 [2]   Go Up