Canon TS-E 24mm f/3.5L II - test images

[Michael Fontana]

Thanks a lot, Bart

yeah, that looks pretty impressive, I agree!
Did you worked the vertical lines ? - it looks pretty good - low distortion at the center - while having some clearly visible (but not to bad) at the corners with the shift of 8.
The shifts of 4 - 6 looks be pretty good, while at 8 it starts to become nasty. We did see it on your other shots, too.

Low distortion is important for that type of photography - as with both - shift and tilt, its hard to correct in post.

Good idea to have the tilt going paralell to the boxes face - is the tilt good visible and controllable in the viewer or do you need live-view?

Your cam position would be about 30 m in height, with the scale of your maquette arround 1/500. When going lower - what the architects often want:

off course more shift is required. On that example, the glas-wall at the left is about 5 - 6 cm in height.

That example was taken with the Perspectar:

it's nose diameter is about 35 mm, so you can put it everywhere, while bigger lenses require to demolish some buildings at the modell for allowing a good position.

With a bigger modell, that problem - size of the lens - will become minor.


After that excursion in architecture model shots, back to the TSE-24:
The difference between 11 and 16 in your DOF-gif is quite obvious, I think the tilt helped a bit for that, too.

Can you change the tilt orientation, (vert/hor) as well as with the older TSE-lenses ?

Beside the use as a shift for architecture photography, it can be used in modell photography as well, for sure not in all situations, but for some.

 

[Asher Kelman]

off course more shift is required. On that example, the glas-wall at the left is about 5 - 6 cm in height.

That example was taken with the Perspectar:

it's nose diameter is about 35 mm, so you can put it everywhere, while bigger lenses require to demolish some buildings at the modell for allowing a good position.

With a bigger modell, that problem - size of the lens - will become minor.

Here's our previous discussion of this unusual lens. For completion, a link to a review and a store would be great.

Then back to the TS-E

Asher

 

[Michael Fontana]

Here's our previous discussion of this unusual lens. For completion, a link to a review and a store would be great.

Then back to the TS-E Asher

I agree, Asher

the example of the Perspectar shows that the new TSE 24 has its limit vs a dedicated °modell° lens, for modell shots - even if the TSE, as Bart has shown, does better in IQ.
You can't buy that Perspectar anymore.

 

[Bart_van_der_Wolf]

Thanks a lot, Bart

yeah, that looks pretty impressive, I agree!
Did you worked the vertical lines ? - it looks pretty good - low distortion at the center - while having some clearly visible (but not to bad) at the corners with the shift of 8.
The shifts of 4 - 6 looks be pretty good, while at 8 it starts to become nasty. We did see it on your other shots, too.

The verticals were vertical because I placed the camera back vertical (parallel to the box edges). No other post processing was done to correct for keystoning.

Good idea to have the tilt going paralell to the boxes face - is the tilt good visible and controllable in the viewer or do you need live-view?

I have the Ec-S groundglass in my camera, so the viewfinder is a bit darker when using a lens like this f/3.5 . It is however very accurate and allows to judge the tilted sharpnes (and I've AF microadjusted the lens, so I get a focus confirmation (beep) if the focusposition is covered by one of the AF sensors). It can beused like that when not shooting from a tripod. However, when shooting with a tripod, Live View is so much more accurate that I use that a lot. Shooting tethered is of course another option which also offers a larger preview.

... That example was taken with the Perspectar:

it's nose diameter is about 35 mm, so you can put it everywhere, while bigger lenses require to demolish some buildings at the modell for allowing a good position.

Yes, that's a perfect tool to get inside the streets of a scale model or inside a building if the roof is removed. The TS-E 24mm is too bulky for that (82mm filter diameter! The Pola filter was expensive), and it's a staight shooter, not at 90 degrees without a mirror. Might be an interesting DIY experiment, constructing a (front surface) mirror gizmo.

------
The difference between 11 and 16 in your DOF-gif is quite obvious, I think the tilt helped a bit for that, too.

Yes the tilt, but also the deconvolution sharpening. I used more sharpening than needed at f/8 or wider, to compensate for the diffraction at the pixel level. At a 300 PPI output size, it would be hard to see the difference between f/11 and f/16 (apart from the DOF gain) in a print. But when planning for larger output, the trade-off becomes more difficult.

Can you change the tilt orientation, (vert/hor) as well as with the older TSE-lenses ?

Better! This lens (just like it's 17mm sibling) also allows to choose between parallel or crossed (or an angle in between) Shifts and Tilts without screwdriver. There is more chance of vignetting when shifting and tilting in parallel to the same direction (the image circle poses a limit).

Beside the use as a shift for architecture photography, it can be used in modell photography as well, for sure not in all situations, but for some.

Indeed. It's a very versatile lens, offering many creative possibilities in addition to its technical problem solving ones (just like a view camera, but in a smaller package). It is expensive and heavy though.

Bart

 

[Asher Kelman]

Not to get too far off-topic, it looks like it has a microscope objective at the end of a tube! What would one use instead?

Asher

 

[Bart_van_der_Wolf]

Not to get too far off-topic, it looks like it has a microscope objective at the end of a tube! What would one use instead?

A modified endoscope, and a lot of light.
Resolution is limited though, but it's nice for video.

Bart

 

[Michael Fontana]

The verticals were vertical because I placed the camera back vertical (parallel to the box edges). No other post processing was done to correct for keystoning.

well, then distortion is really low! I opend your image in PS and verified it.

I have the Ec-S groundglass in my camera, so the viewfinder is a bit darker when using a lens like this f/3.5 . It is however very accurate and allows to judge the tilted sharpnes (and I've AF microadjusted the lens, so I get a focus confirmation (beep) if the focusposition is covered by one of the AF sensors). It can beused like that when not shooting from a tripod. However, when shooting with a tripod, Live View is so much more accurate that I use that a lot. Shooting tethered is of course another option which also offers a larger preview.

Yes, I'd think that livewiew and microadjustement are a big help in these situations, when precision is required.

Yes, that's a perfect tool to get inside the streets of a scale model or inside a building if the roof is removed. The TS-E 24mm is too bulky for that (82mm filter diameter! The Pola filter was expensive), and it's a staight shooter, not at 90 degrees without a mirror. Might be an interesting DIY experiment, constructing a (front surface) mirror gizmo.

I don't want to make that out of the TSE. But I rather triy to understand, how this lens can be used for model shots or close-ups.

Yes the tilt, but also the deconvolution sharpening. I used more sharpening than needed at f/8 or wider, to compensate for the diffraction at the pixel level. At a 300 PPI output size, it would be hard to see the difference between f/11 and f/16 (apart from the DOF gain) in a print. But when planning for larger output, the trade-off becomes more difficult.

The difference 11 vs 16 looked bigger than usual.
I know deconvolution sharpening from RAW Developer, and use it a lot. I agree, the end format plays a important role in sharpening.

Better! This lens (just like it's 17mm sibling) also allows to choose between parallel or crossed (or an angle in between) Shifts and Tilts without screwdriver.....

Thats great: if on location you have to switch from portrait orientation to landscape....

Indeed. It's a very versatile lens, offering many creative possibilities in addition to its technical problem solving ones (just like a view camera, but in a smaller package). It is expensive and heavy though. Bart

Did you already tried to get te NNP? Is it at the front ot in the middle of the lens?
Might be important - my 14-24 has the NNP pretty at the front - it would require a special, aka longer panohead.

 

[Asher Kelman]

Did you already tried to get te NNP? Is it at the front ot in the middle of the lens?
Might be important - my 14-24 has the NNP pretty at the front - it would require a special, aka longer panohead.

For the uninitiated the NNP="Nodal point" or more correctly, "No Parallax point", the entrance pupil of the lens which is that position inside the lens where the incoming rays from the object become inverted and projected towards the recording medium of the camera.

In stitching images, when the camera is rotated about the NNP, adjacent images can be stitched as the projections of overlapping detail will line up perfectly.

Asher

 

[Juergen Koslowski]

(82mm filter diameter! The Pola filter was expensive)

Dear Bart,

Thanks for providing very detailed and useful information concerning this intriguing lens! Since samples created with the 17mm version started to appear, I've been wondering about the probably more useful 24mm version.

Which kind of polarizer did you get? I do have a regular (non-slim) 82mm B+W Kaesemann model that I would like to use on this lens, if possible.

-- Jürgen

 

[Michael Fontana]

A modified endoscope, and a lot of light.
Resolution is limited though, but it's nice for video.

Bart

It's not a endoscope, it's a different construction with 50 mm; if I remember correctly, the image is projected several times in the lens itself towards the camera, like a microscope. With a normal viewer, its hard to see the image, therefore I build some special screen, similar to the ground glas of a view camera.

you can close the f-stop- which is not possible with a endoscope. I had been trying some endocope, but they didn't covered FF. and had much more distortion as well.

Yes, resolution isn't like the best lens, but you can take photos that are not possible otherwise.

 

[Bart_van_der_Wolf]

well, then distortion is really low! I opend your image in PS and verified it.

That's correct. Very little barrel distortion.

The difference 11 vs 16 looked bigger than usual.

That's probably because the MTF drops really fast after f/11, maybe f/13 depending where one draws the line.

Did you already tried to get te NNP? Is it at the front ot in the middle of the lens?
Might be important - my 14-24 has the NNP pretty at the front - it would require a special, aka longer panohead.

Yes, the entrance pupil of this lens is also pretty far at the front of the big lens. It's approx. 132mm in front of the sensor plane marker (on the side of the prisma enclosure), a little behing the red L ring. I use the RRS MPR-CL II nodal slide (a 7.4-inch rail) that comes standard with the RRS Ultimate-Pro Omni-Pivot Package, and it's just long enough for this lens.

There is one caveat though. Tilting will displace the entrance pupil, so for the best results one would need to adjust according to the amount and direction of tilt.

When used without tilt, stopped down to f/8 or f/11, it makes a great lens for a very wide FOV with only a couple of tiles. Just 3 tiles in portrait orientation with approx. 10% overlap in a single row allows a 109 degree HFOV equal to a 13mm lens (some 10500 pixels wide), and a 74 degree VFOV equal to a 16mm lens (5616 pixels high) or almost 59 MP.

Cheers,
Bart

 

[Bart_van_der_Wolf]

Which kind of polarizer did you get? I do have a regular (non-slim) 82mm B+W Kaesemann model that I would like to use on this lens, if possible.

Hi Juergen,

I got the B+W 82mm slim MRC version, but Canon does not state in the manual that a slim version is required to avoid vignetting at extreme shift/tilt positions. The filter diameter is quite a lot bigger than the front lens element, so presumably a regular filter should be okay. I just didn't want to risk it, since the filter had to be ordered for me, and I had no other 82mm ones to test with.

Bart

 

[Juergen Koslowski]

I got the B+W 82mm slim MRC version, but Canon does not state in the manual that a slim version is required to avoid vignetting at extreme shift/tilt positions. The filter diameter is quite a lot bigger than the front lens element, so presumably a regular filter should be okay.
Bart

I was hoping for that, as the regular filter does work on the 21mm Zeiss Distagon.

Concerning the issue of flat-stitching: would it be feasable to mount the lens on a tripod, and then shift the camera? I don't know if the lens offers enough space for some kind of lens collar, and if weight would be an issue (just ordered my 5D Mark II). Maybe something for the guys at RRS to ponder, or for Rainer Burzynski in Germany.

-- Jürgen

 

[Bart_van_der_Wolf]

Concerning the issue of flat-stitching: would it be feasable to mount the lens on a tripod, and then shift the camera? I don't know if the lens offers enough space for some kind of lens collar, and if weight would be an issue (just ordered my 5D Mark II).

Unfortunately there is no space on the lens that's not already occupied by a knob or lock/release knob or pin.

Maybe something for the guys at RRS to ponder, or for Rainer Burzynski in Germany.

While not exactly what you are thinking about, but RRS does have a solution that I used in the earlier stitches in this thread. Unfortunately it's a 2-step approach.

This is what I use for flat stitching, and it's not very bulky, so easy to travel with:

It consists of 3 basic components.
1. The MPR-192 multipurpose rail
2. The B2-FAB/mAS mini clamp package
3. MPR-B, pair: Two index stop bars

Operation involves sliding and locking the camera to one stop bar position, and shifting the lens in the opposite direction. Take a shot, slide and lock the camera to the other stop bar position and the lens in the opposite direction. You can preset the sliding distance by making a spacer of the desired width and place it on one side of the mini clamp package and tighten the miniclamps in position on both sides of the clamp+spacer. If one makes a spacer of 24 mm, then one can simply use the maximum + and - 12mm shift which works reasonably fast for landscape orientation shifts.

If this is the only type of stitching one does, then an MPR-113: 113mm Multi-purpose rail would suffice, because one only needs 90 mm of rail to lock everything into place. That is even more compact to transport.

Bart

 

[Juergen Koslowski]

Dear Bart,

Thanks for sharing your set-up, my order to RRS is out, even if the lens itself may have to wait a little longer.

Would you be able to determine, if the lens is suitable for infrared-photography? I realize that 82mm IR-filters are not so common. With a Heliopan RG 715 I notice a definite hot spot in the center of the old 2.8/16-35 L, while the 2.8/24-70 L doesn't seem to suffer from this problem. This is confirmed at http://www.lensplay.com/lenses/lens_infra_red_IR.html.

-- Juergen

 

[Asher Kelman]

3. MPR-B, pair: Two index stop bars

What a simple but clever addition to the rail! I hope one can order them separately! That certainly would save time looking at the scale as shift can be achieved by touch and feel!

Asher

 

[Michael Fontana]

Hi Bart

I'm sure you'v been using the TSE often, these days; another question rises here, about the amount of shifting:

there's on the TSE-24-II - unlike other shift lens - very little distortion, until a certain amount of shift. By passing that boarder, the correction of distortion by software will be necessary. How much can you shift while keeping the lines straight?

That question is not asked in terms of landscape, but architecture photography, which is more critical in that regard.

2nd question:

Until which amount of shift will the corners be sharp?
My guess - with the help of your examples would be about 6 mm in critical situations, maybe 8 in less critical. How about the curvature of field on flat objects - interiors?

3rd: It looks like CA was a no-brainer even full shifted. Is that correct?


Thanks, Michael

 

[Asher Kelman]

there's on the TSE-24-II - unlike other shift lens - very little distortion, until a certain amount of shift.

That impresses me since with the 24mm 1.4L II the distortion at the last 7% on each side is considerable. A double base instrument and the muscian then becomes 20% wider! What appeals to me is the possibility of using the 24MM II TSE without any need for such correction for people, especially as well as architecture. I wonder how good the original version of the 24 mm TSE is unshifted?

Asher

 

[Michael Fontana]

Hi Asher

well the TSE is kinda MF-lens, so unshiftet, you use the sweet spot only. Must be very good, that's rarely been a problem of a shift lens.

It's the amount of shift with good quality which finally makes a shift lens good or bad.

Using it for people: that lens is not done for fast shooting; it's manual focus and when you shift, exposure will probably be indicated wrongly. With its f = 3.5 wide open, I would think twice before using it for your purpose; it's not made for that, IMO.

Are you - apart from the distortions fine with the 24 1.4?
You could use software to ge rid of that distortion.

I never had one, but the first version was a duck lame, terrible CA and soft when shiftet; well that's what I heard and saw on some samples. The version 2 has a 9 mm larger image circle, that's quite much - and should help a lot in improving quality. The border of the image circle alwith bite - even with a 4/5'!

I' ve the Schneider PC-28 (same as Leica 28) and the zuiko 35 shift, but don't se them a lot.

 

[Michael Fontana]

Concerning the issue of flat-stitching: would it be feasable to mount the lens on a tripod, and then shift the camera? I don't know if the lens offers enough space for some kind of lens collar, and if weight would be an issue (just ordered my 5D Mark II). Maybe something for the guys at RRS to ponder, or for Rainer Burzynski in Germany.
-- Jürgen

Jürgen, this works as well:

It's the macrorail CT 77 from Cullmann. I had added the yellow screws as a automatic stop of the cam's movement. They' ve the same function as Bart's index stop bars.

 

[Bart_van_der_Wolf]

Hi Bart

I'm sure you'v been using the TSE often, these days; another question rises here, about the amount of shifting:

there's on the TSE-24-II - unlike other shift lens - very little distortion, until a certain amount of shift. By passing that boarder, the correction of distortion by software will be necessary. How much can you shift while keeping the lines straight?

That question is not asked in terms of landscape, but architecture photography, which is more critical in that regard.

Hi Michael,

Yes, I've been testing various scenarios. I tend to restrict myself to a single lens from time to time, in order to better learn the limitations/strengths/characteristics under different shooting conditions.

Here's an example of a more 'architectural' subject, with full vertical shift.

This is part of an old building that's going to be renovated/rebuilt, and because it's close to where I live I sometimes use it as my test object for things like distortion testing. I had trouble getting the horizontals exactly right because when I shot it, there was incoming rain and I had a qhole sequence of images I wanted to take at various shifts and apertures. It also sparked my quest for a better leveling tool.

To answer your question, very little geometric distortion (a tiny bit of barrel distortion is the main component, I estimate 2 or 3 pixels at the fully shifted edge of this example assuming the concrete floors of this building are level). When I stitch images with the regular rotation through the entrance pupil method, I get extremely low amounts (e.g. with parameters like a= -0.001131, b= -0.002484, and c= 0.001508 on non-shifted shots of 5616 x 3744 pixels). These values vary a bit depending on the optimization, but you'll understand how little correction is needed. Since these numbers are multiplied by radius, they will probably not be significantly different at larger radii, but I'll test it later for when the shifted edges are more critical. For now, you can see that the barrel/cushion distortion is minimal for such a FOV.

2nd question:

Until which amount of shift will the corners be sharp?
My guess - with the help of your examples would be about 6 mm in critical situations, maybe 8 in less critical. How about the curvature of field on flat objects - interiors?

Hard to pinpoint it at a given amount. It also (besides aperture used) depends on whether one shifts horizontally on a landscape orientation, or vertically (as in the example above). If we limit it to 6mm horizontal, then we'd have to shift further (by a factor of 3/2 or +50%) in the vertical direction for the edge to reach the same edge of the image circle, so 6 mm becomes 9 mm shift. It's also different for the edge or the corner. The latter is more critical because it's closer to the edge of the image circle, but it may not contain meaningful detail (e.g. clouds, or Out-of-Focus regions to enhance the depth perception).

You are right that curvature of field will be most visible when shooting close up, and there seems to be some going to the back of the scene in the corners, but these are typically scenes where one either uses tilt, and/or a smaller aperture. DOF is adequate to cover for it when it might be noticable with aperture wide open. Since the lens is an internal focusing one, it may also depend on the focus distance, so I'll have to test that separately. For many types of photography this lens will be used at f/ 8 to f/11, which offers 1.13 m to 1.67m total DOF at 2 meters distance with my very critical COC criterion (C=0.0096mm), and 3x more with the traditional criteria (C=0.030mm).

Here is an attempt to subjectively determine a rule of thumb aperture that's a compromise between corner and center sharpness:

From left to right, optical center (bottom center of the above overview), top left corner at full 12 mm vertical shift, and top left corner of an unshifted image (mid left of the above overview).

Besides the same uniform sharpening (based on the f/8 images) of all images, no other postprocessing was done. Notice the low level of CA on the edges of the white painted concrete support beams. The micro detail contrast at the optical center starts to deteriorate from f/4.0, but the most extreme corner with this type of shift will improve upto f/11, thus kind of leveling the sharpness across the image. Sharpness (especially of the corner) can be improved by dedicated sharpening, but sharpening for a stopped down f/11 shot already improves both 'center' and corner sharpness compare to these examples. Beyond f/11 the loss of microdetail accelerates, so for images requiring output enlargement, I'd be carefull with anything less than f/13.

3rd: It looks like CA was a no-brainer even full shifted. Is that correct?

Yes, although there is some, it's very little for a lens with this FOV. With a vertical shift of + and - 12 mm, the VFOV is approx 90 degrees. In many cases, a full 12 mm shift will not be necessary, but even then the CA is easily corrected (and will allow better sharpening).

Bart

 

[Bart_van_der_Wolf]

What a simple but clever addition to the rail! I hope one can order them separately! That certainly would save time looking at the scale as shift can be achieved by touch and feel!

Yes, there are sold (at the huge price of US$ 15 each, or US$ 25 per pair) separately.

Not having to look at the exact shift saves a lot of time (can help with clouds racing across the sky). It doesn't beat shifting the camera body alone, because the lens needs to be shifted as well, but one could settle for the maximum amount there as well, and crop later.

The RRS system is expensive, but one does buy into a system. That means that parts can be repurposed depending on the configuration one assembles (many parts come with hex screw tools for user re-assembly if needed). Even the Ballheads can be ordered without a clamp, so one can combine them with e.g. a full pano package with it's rotating leveling clamp (PCL-1) or a quick release clamp.

Bart

 

[Bart_van_der_Wolf]

That impresses me since with the 24mm 1.4L II the distortion at the last 7% on each side is considerable. A double base instrument and the muscian then becomes 20% wider! What appeals to me is the possibility of using the 24MM II TSE without any need for such correction for people, especially as well as architecture.

The geometrical (barrel/cushion) distortion is almost negligible at larger distances. Since the lens uses internal focusing, things may vary a bit by focus distance setting (I haven't specifically tested for that). At extreme flat-stitched shifted positions one does have projection distortion as a result of projecting a large FOV on a flat surface. There are software solutions to compensate for that as well as minute barrel distortion.

I wonder how good the original version of the 24 mm TSE is unshifted?

I can't compare personally because I never bought the original version due to it's relative lack of sharpness compared to its 45mm and 90mm siblings. However, I think the new version is hard to beat, and it also offers better glass elements and a newer type of coating. This review suggests that there is a difference towards the corners, if you look at the relative displacement (assuming identical image position crops were taken) closer to the edges.

Bart

 

[Bart_van_der_Wolf]

Would you be able to determine, if the lens is suitable for infrared-photography? I realize that 82mm IR-filters are not so common.

Hi Juergen,

I don't have an IR filter in the correct size, but when I handhold a 58mm diameter Hoya R72 in front of the lens, I don't see an IR hotspot. I do see spots but those may be dust and light coming in around the filter edges. I think this lens behaves well, probably due to the improved coatings. The rear lens surface is pretty flat, so I think that a hotspot would already have been clearly visible.

Bart

 

[Asher Kelman]

The geometrical (barrel/cushion) distortion is almost negligible at larger distances. Since the lens uses internal focusing, things may vary a bit by focus distance setting (I haven't specifically tested for that). At extreme flat-stitched shifted positions one does have projection distortion as a result of projecting a large FOV on a flat surface. There are software solutions to compensate for that as well as minute barrel distortion.

Hi Bart,

What I need is to have to use no more than f 5.0 and to have the greatest possible resolution and enough DOF for the orchestra. At present from where I am placed, I include the entire orchestra and a good 5 feet on either side. Lucky for me, at the moment, the lens does not seem to be immediately available in the USA!

Asher

 

[Michael Fontana]

beside Bart's very helpfull insights, here' s another review of the new 24 TSE, from Bryian C.

 

[Bart_van_der_Wolf]

And here's another review: http://www.dpreview.com/lensreviews/canon_24_3p5_tse_c10/

Bart

 

[Aboud Dweck]

Shifty

Last year I switched from Canon to Nikon for the superior Nikon 24MM PC-E. Of course, just a few months later, Canon announced this 24 II shift lens. I am wondering if anyone has done a side by side comparison of the two lenses and what their impressions were. I seem to notice a little resolution fall-off on the Nikon when used in vertical position with full shift.

 

[Bart_van_der_Wolf]

Last year I switched from Canon to Nikon for the superior Nikon 24MM PC-E. Of course, just a few months later, Canon announced this 24 II shift lens. I am wondering if anyone has done a side by side comparison of the two lenses and what their impressions were. I seem to notice a little resolution fall-off on the Nikon when used in vertical position with full shift.

Hi Aboud,

I'd be surprised if a T/S lens, or any lens for that matter, wouldn't show some resolution fall-off towards the edges/corners. The main point is how much fall-off, and how much of it can be recovered (in postprocessing)?

I don't have a side by side comparison with the Nikon.

Cheers,
Bart

 

[Michael Fontana]

Last week, I had the opportunity to test the TSE 24, vs TSE-17 vs TSE-17 +1.4 TC.

Correct me if I'm wrong, Bart uses the 24 mm much for stitching with a panohead, while I'm interested to extend the FOV of a shot, called flastitching and shifting in single shots.

Here's a test for looking at flares, in backlight; the inner white line shows the unshiftet image; the outer white line a flatstitch, when shifting 5mm, and the entire frame shows a shift of 10 mm.

Analog to what I saw in Bart's tests, a shift of about 7 mm keeps image quality to the required degree; while in the above example the chimney at the left side, beeing at 8 mm wouldn't fit it.

Shifting 5mm to both sides gives a image with a 1: 2-ratio, whith a HFOV of 90 degs, (equivalent to a 18 mm on a prime) while of course the VFOV remains stable.

Sometimes, I don't want to use a wider focal lengts (adding sky and grass) but want to have a long building - like this stadium - contextualised in the surrounding, here the skyscrapers. For that purpose, that TSE 24 works well, if shiftet about 7 mm only.

Did somebody can explain the distortion at the skyscraper under the "unshiftet" line, at the left? Its roof is not not horizontal, even if I set the verticals correct in post.