Starting a few years ago, Canon’s patent applications started showing a totally new way of adding tilt and shift functions to lens designs. Canon’s range of tilt/shift lenses is the largest of any camera manufacturer, although Laowa produces a wide range [some shift only], covering many mounts from APS-C sensors up to the [33x44mm] medium format Fujifilm G mount.
If Canon does bring such new lenses to production, how will they differ from the current TS-E range and how do they bring tilt, shift, image stabilisation and autofocus under camera control? More to the point, how can they be used and what can be done with them?
A short YouTube video I recently published, looking at implications of some related patents [11 mins] – part of my tilt-shift YouTube playlist.
The history
Canon’s first shift lens, the FD mount TS 35mm f/2.8 SSC appeared in 1973, and although updated in 1975, Canon has never got round to offering another 35mm tilt/shift since then. It’s a useful focal length, with my own TS-E50mm F2.8L Macro giving just over, at ~40mm equivalent, when adapted for use on a medium format Fujifilm GFX100S.
The lenses most familiar to people will be the TS-E range of EF mount lenses, starting, in 1991 with the TS-E 24mm F3.5L, TS-E 45mm F2.8 and TS-E 90mm F2.8. The updated TS-E24mm F3.5L II appeared in 2009, along with the introduction of the very wide TS-E17mm F4L.
Comparing the sizes of the TS-E17mm F4L, TS-E 24mm F3.5L and TS-E 24mm F3.5L II. Note the red marks on the tilt scale of the original 24mm [centre] these show settings where a reduced image quality might be possible. This also applied to shift settings. One of the features of the Mk. II version of the lens is an expanded image circle, meaning no red lines on the settings.
In 2017, the TS-E50mm F2.8L Macro, TS-E90mm F2.8L Macro and TS-E135mm F4L Macro were launched, even if the ‘Macro’ designation only went as far as 0.5x. The three lenses are optically excellent and solidly built. As with all TS-E lenses they are fully manual other than aperture control.
For shift, all these lenses rely on having a very large image circle. It’s much larger than needed to cover the sensor.
That means that a physical shift mechanism can shift the lens relative to the mount, to allow different parts of the image circle to be captured.
Using shift
The commonest use of shift lenses is to correct verticals – one reason they are so popular with architectural photographers like myself.
Point a wide lens upwards and this happens to verticals.
It may be dynamic and loved by some in the marketing department, but my clients [the architects] want to see the building looking a bit closer to their designs.
As to the usual cry of “But I can do that easily in software”. Maybe, but the corrected image is trapezoidal and for one like this would need a hefty and unpredictable amount of cropping. You could of course use a much wider lens and crop as needed, but I generally want the full available resolution.
Shift can be in any direction, giving, along with where the camera is pointed, quite a bit of control of the perceived perspective. Even if the only ‘true’ way of changing it is to move where the camera is positioned. This is one reason the Nikon range of tilt shift lenses are called PC lenses – for ‘Perspective Control’.
Using Tilt
Whilst shift moves the lens optical axis relative to the lens mount, tilt ‘bends’ the lens, meaning the optical axis is no longer orthogonal to the sensor. The actual amount of tilt [measured in degrees] is indicated on the side of the lens, where tilt is applied via a suitable adjustment knob, such as with the TS-E45 F2.8 shown here.
Tilting the lens downwards, for example, allows the plane of focus to be run along the ground, or in this example using the TS-E17mm F4L, a bench.
As with shift, tilt can be applied in any direction, with some physical limits from the design of the lens mechanisms and lens mount.
Terminology: Tilt/swing/rise/fall. In my own descriptions I generally refer to tilt or shift combined with a direction. So shift upwards or tilt to the right for example. If you come from a large format camera background, you may well be used to terms such as swing, rise and fall. Most people looking at my explanations come from a ‘small camera’ background and I’ve found that just referring to tilt or shift is easier. Also if you come from a large format background, you should already know all this stuff inside out. It’s also why I absolutely avoid any significant discussions of optics or the maths – there is no quicker way to put off most photographers from reading further.
The plane of focus can be placed along almost any plane you like – if the lens has enough adjustment range. Combining this with a wider aperture allows effects like this.
All that comes with physical movements of the lens relative to the camera. The movements are set manually, so there is no movement data coming from TS-E lenses telling the camera how much tilt or shift has been applied. Nor is there any distance setting information, just the aperture. That means no tilt/shift related data for EXIF, making automatic lens corrections impossible with any movements. No auto-correction ability means that lenses need to be designed with very good characteristics, over a larger image circle. That means lots more glass and expense.
The patents – a new way of doing tilt and shift?
If you have ever used an image stabilised lens, you will already have used a form of electronically applied shift. The IS unit in such lenses has optical elements which are moved to shift the image on the sensor. However, this is done to correct for camera shake.
Canon’s patents and patent applications over the years show a number of developments in their thinking about making use of lens movements [tilt and/or shift]. In fact, there has not been a single ‘traditional’ tilt-shift lens design, which I’ve seen in any patent application for many years.
A 2019 patent application [20190199912] shows a motorised tilt system. Several other examples show the use of lens tilt in security camera applications, placing the plane of focus through a viewed scene so as to maximise useful sharp areas in focus.
Remember: Tilt moves the plane of focus, it doesn’t magically give you more depth of field.
A May 2022 application [20220155653] introduces the idea of moving internal lens elements to produce tilt and shift functions. Lens elements can be moved independently of each other. Note how large these elements are, since to produce appreciable shift or tilt the element needs de-centring by quite a bit. Just how much that tilt and shift is, compared to existing TS-E lenses, is not clear; however, in the text, it refers to the object focal plane being tilted by some 70 degrees at 2 metres for one example. That is similar to what might be expected from some current tilt-shift lenses, but as ever with patents, there is quite a bit of specific ambiguity— one reason patents are not written for easy reading.
The examples also suggest that the image circle is nowhere near as large as needed for the TS-E designs. Back focus distances firmly place these examples as mirrorless designs.
This patent application is one for those of you really into the details of optical design. Specific examples are given: 35mm f/2.8 | 85mm f/2.8 | 135mm f/4 | 230mm f/4.
Moving the elements in opposite directions gives a tilt of the plane of focus. Those of you familiar with tilted lenses will notice aspects of the Scheimpflug principle at work, in the way the various planes relate to each other.
As a user of tilted lenses, one of my observations is that there is no simple relationship between the movement of the T/S elements and the equivalent amount of lens tilt. Presumably, there is a control on the lens or via the camera, which gives a ‘degrees of tilt’ value. If you include shift [both TS elements moved in the same direction], then it becomes clear that some means of setting the tilt and/or shift amount is needed.
Partly as a working photographer and partly from a previous career in usability research, I’m always wondering just how you make functionality available to the user. At the moment, I do make use of the tilt values and shift values in setting lenses, although fine adjustment is often an iterative process, especially for closer work.
A typical close-up use of lens tilt with the TS-E24mm F3.5L II. The lens is tilted downwards, running the plane of focus as shown.
By March 2023 there were reports of actual new Canon tilt/shift lenses out for testing, but over two years later there are still no reliable signs of any new lenses. However, still the patent info comes, and in September 2023 a patent application [ 20230305294 ] showed a cut-through diagram of a zoom tilt-shift lens.
However, the real point of the patent is describing how to set the tilt-shift lens elements to place the plane of focus on an arbitrary plane. This would be like the example with the TS-E24mm F3.5L II above. You can specify a plane with three points, and the patent shows an example of picking three points on a tabletop and getting the plane of focus to run through them.
The three points are selected via the camera’s screen [so much for those of us with poor close-up vision – hopefully it will work far more easily with an updated version of EOS Utility].
The patent info shows the maths behind selecting an arbitrary plane of focus. For my own studio use, I’ve perfected a technique using a card with a finely detailed pattern on it, utilising moiré fringing to find the correct angular rotation of the T/S lens in its mount. It may be simple to use in practice, but I still have an entire chapter of my book devoted just to close-up focusing with tilt.
A recent patent application in the same style looks further at making use of being able to tilt the lens, including examples capturing images of people at different distances and the tabletop – this time with cakes on it.
Canon are obviously trying hard to come up with uses for this technology, such as incorporating it in tracking [20240284046] and more tabletop photography [20240337905]
Of course, no discussion of tilt/shift lenses would be complete without the so-called ‘tilt-shift effect’ or model world look. [20240361569]
In this, the plane of focus is positioned to exaggerate out-of-focus areas and give a look to images suggesting they are models. Control of the effect [angle of tilt and orientation] is via a lens control ring or the rear screen.
The effect is a simple one, but getting the right viewpoint and subject detail aids the illusion. This is taken with the TS-E90mm F2.8L Macro
Where are my new tilt/shift lenses?
Maybe they are waiting for a new high-resolution R camera [a well known Unicorn], or maybe, as someone from Canon recently said to me, the existing lenses are just fine…
The recent 50/90/135mm TS-E lenses are excellent and even work well on my GFX100S, as well as my R5. The 17mm and 24mm still perform well on the R5, but could likely benefit from updated optics for anything over 50MP. Incidentally, my 1991 TS-E90 2.8 works just fine on the GFX100S
Other manufacturers [apart from Laowa] are not rushing to fill out or create a tilt-shift range, with Nikon’s PC-E range looking decidedly long in the tooth and suffering from the narrow throat of the F mount at larger lens movements. The PC-E 19mm optics in a Z mount lens would be impressive, when freed of the potential physical vignetting of the F mount. Fujifilm’s 30mm and 110mm TS lenses are superb quality [at a price], and interestingly enough, incorporate sensors so as to be able to feed back the amount of shift to the camera, for correction purposes [unfortunately only usable with C1 software at the moment].
It’s interesting that many of the patent applications are concentrating on uses for the lens tilt/shift functionality – ones where I’ve found it much more difficult to explain to people new to such lenses. In order of how quickly the ‘lightbulb moment’ occurs, I’d say it goes [for individual non-stitched shots] from quickest for up/down shift followed by left/right, then diagonal shift, then tilt at a distance [landscape] and finally close-up tilt. Obviously, that is a generalisation, but Canon seems to be working on ways to make it ‘easier’. As to if we will see such lenses any time, that is what CR is for ;-)
References
Photography with Tilt and Shift Lenses, Keith Cooper ISBN 9781785007712
My book looks at the many ways that tilt/shift lenses can benefit your photography from a technical and creative point of view. Recently reprinted and available from good bookshops.
Focusing the Viewcamera, H.M. Merklinger
PDF available via http://www.trenholm.org/hmmerk/#FVC
Well-written practical guide, from where I developed many of my techniques for using tilt tables as an aid to setting lens tilt in practical usage.
My written tilt/shift reviews and articles
My YouTube Channel has a dedicated tilt/shift playlist
https://www.youtube.com/c/KeithCooper
Patent searching
Go to https://ppubs.uspto.gov/pubwebapp/static/pages/ppubsbasic.html and enter any of the patent numbers above. A search on ‘Canon’ and ‘tilt’ will get you started in exploring further.
|
When you purchase through links on our site, we may earn an affiliate commission. Here's how it works. |
Will we ever get the RF TSE 14mm?
Oh, how I sympathize with all the poor creatures waiting for the RF 1,2/35! :cry:
I own 17, 24 II, 50mm and 90.
I use the 24 II the most. It’s a super lens, very sharp and adaptable. 17 is the next most used. 50mm is very sharp but a less useful (for me) focal length.
What I’d love is for the lens to be able to keep rotating. The second would be some sort of preciseness on tilt. I don’t use it much but a little tilt can be useful but it’s hard to be precise currently. Auto focus wouldn’t bother me. There is something about manual focusing and being slow and accurate might be lost.
You can do it easily, ok, but just look at how many sensor pixels you lose this way. And what about tilting???
There's a good reason why TS lenses exist. :)
These architecturally themed rooms are real “models,” built on a 1:12 scale – each room is less than a meter wide. Amazing detail on the furnishings at that scale.
My attachments are out-of-camera JPEGs using an EF 17-40mm lens at 17mm and f/4, close to MFD.
The apparent lack of depth of field relative to perceived scale is real – these are close-ups of models!
As to the Canon patents, I'm curious as to how these lenses will perform - perhaps that's why we haven't seen any yet ;-)
I’d love it if someone made a 35mm TS. I use a Zeiss 35 PC-Distagon at that FL and it’s definitely lacking convenience and functionality.
From what I read, it'll automatically crop but keep the 45mp?
I'm really just curious if there's any benefits to getting a lens designed for crop sensors on a full frame camera. It's seem like the pixel density would be great. (I could have the terminology wrong! )
Hopefully Sigma, Tamron and others can start making RF mount lenses soon!
If I come across it again, I'm going to screen shot it and see what the exact source was. I just read the Google AI answer.