TS-E 90mm F2.8L Macro review
Canon TS-E 90mm F2.8L lens review
90mm macro tilt/shift lens
The Canon TS-E 90mm f/2.8L macro lens is Canon’s ‘short telephoto’ focal length tilt/shift lens.
The lens replaces the 1991 TS-E90mm F2.8 with a lens in Canon’s ‘L’ category. Keith has been trying out the lens on his EOS 5Ds 50MP DSLR. There are also some direct comparisons with the older lens showing what’s changed.
The review covers the image quality of the 90mm and looks at some of the uses for such a lens, making use of the tilt and shift to obtain images you can’t easily get in other ways. You might also want to look over our lengthy TS-E135mm review and TS-E50mm review which have more examples of what makes these lenses so special.
This is the third of the new TS-E lenses from Canon and rounds off my reviews of the 50/90/135mm ‘L’ series tilt shift macro lenses.
As with my other reviews, there is much more consideration of how and why you use such lenses, as opposed to more formal ‘lens testing’ Most of the sample images can be clicked to see enlarged versions and there is a gallery of many samples at the foot of the review.
- EF-Mount Lens/Full-Frame Format
- Aperture Range: f/2.8 to f/45
- Maximum Tilt: +/- 10°
- Maximum Shift: +/- 12mm
- Minimum Focusing Distance: 39cm
- Maximum Magnification: 0.5x (1:2)
Full specs below, but as a long time user of the original TS-E90mm it’s immediately a lot heftier. I’ll leave comparisons with the old lens until later, but there’s no easy mistaking of which is which.
The lens lets you tilt the front by up to ten degrees on an axis running through the lens.
This is up by 2º from the original TS-E90.
The lens can also be shifted by up to 12mm along a plane parallel to the mount (this is up from 11mm for the old lens).
The shift direction can be smoothly rotated a total of 180º with click stops at 30º steps, whilst the direction of the tilt axis can be rotated relative to the shift plane with stops at 45º and 90º.
The upshot of this is that you get considerable flexibility in positioning the direction of shift and axis of tilt.
These two examples show combinations of movement.
Above, the shift and tilt axes are at 90º, but the whole lens has been rotated by 30º
This is a big change from the original 90mm since it only had the lens rotation movement, and if you wanted to change the tilt to shift alignment it was a trip to Canon Service (or you could modify the TS-E90 yourself with care).
You can rotate the lens by just a few degrees if you need to set the plane of focus away from the click stops.
This makes for much easier precise control of the new lens, particularly in close-up photography.
The movement of the optical elements of the lens can be seen in this view from the rear.
I should note that whilst many people call these lenses ’tilt/shift lenses’, this only really refers to specific movements.
Shifting left/right and tilting up/down:
Rising and falling with the lens swinging left/right:
I’ll stick with the newer (and less precise) terms tilt/shift, so my apologies to people with many years experience of using camera movements…
As with the other new TS-E lenses, you can set (lock) the movements to zero, and you’ve got yourself a very high quality manual focus EF mount lens.
The lens front element is deeply recessed.
As you focus closer, the front of the lens extends, and at shortest distances, the centre comes out further.
The lens comes with a clip on lens hood, that with the recessed front element works very well to avoid extraneous light.
The lens is on the rather nice Benro GD3WH geared tripod head (review) I was lucky enough to be testing at the same time as the 90mm.
For completeness I’ll show the table covering all three new TS-E lenses
|Canon TS-E 50mm F2.8L Macro||Canon TS-E 90mm F2.8L Macro||Canon TS-E 135mm F4L Macro|
|Lens type||Prime lens|
|Max Format size||35mm FF|
|Focal length||50 mm||90 mm||135 mm|
|Lens mount||Canon EF|
|Maximum aperture||f/2.8 (50mm to f/32, 90mm to f/45)||f/4 (to f/45)|
|Special elements / coatings||Moulded glass aspherical and UD elements + SWC and ASC coatings||Moulded glass aspherical and UD elements + ASC coating||Moulded glass aspherical and UD elements + SWC coating|
|Minimum focus||0.27 m (10.63″)||0.39 m (15.35″)||0.49 m (19.29″)|
|Shift||± 12mm||± 12mm||± 12mm|
|Tilt||± 8.5º||± 10º||± 10º|
|0.74 – 0.23||0.64 – 0.15||0.62 – 0.09|
|1.0 – 0.48||0.82 – 0.32||0.77 – 0.2|
|Full time manual||Yes|
|Weight||945 g (2.08 lb)||915 g (2.02 lb)||1110 g (2.45 lb)|
|Diameter||87 mm (3.43″)||67 mm (2.64″)||89 mm (3.5″)|
|Length||115 mm (4.53″)||117 mm (4.61″)||139 mm (5.47″)|
|Filter thread||77.0 mm||82.0 mm|
MTF charts and internal design TS-E 90mm
Data from Canon
Well, I did test the TS-E90 with the sun in parts of the frame and it was quite tricky to produce obvious flare or notable drop in contrast.
The image is quite heavily processed (@ f/7.1) to bring up the spots of light and light spikes around the sun.
At full shift, the spikes around the sun were a bit less well defined, and my laser pointer test star only showed a few pixels worth of coma towards the edges.
A view of the Vijay Patel building (shifted up by ~10mm) and a 100% crop of the top of the building show essentially no chromatic aberration.
As with many images here, click to enlarge or see the gallery at the end of the article.
I’ve not got the equipment for precise quantitative lens testing, but a simple use of my old slide lightbox lets me look for how the lens vignettes at different apertures, and also with it shifted.
A series of photos at different apertures and at full shift are shown here. It’s a screen shot from when I was looking at the RAW files in Adobe Bridge.
The left shot is with the lens used normally, whilst the right hand ones are shifted by the full 12mm.
As you can see, vignetting is hardly obvious at f/2.8, and distinctly difficult to see by f/5.6.
By processing the image with a strong adjustment curve, and posterising the result I can make the vignetting much more obvious.
Any slight asymmetry of the unshifted images is likely due to slight variation in my 20 year old lightbox, not the lens.
The shifted images are obviously showing more vignetting on the right, due to natural fall-off across the image circle of the lens. What is less obvious is the vignetting at the left side of the shifted frame.
This is due to physical vignetting from inside the lens and is something to take into account when stitching images, where you may get a slight mismatch (of exposure) in overlapping areas. My solution to this is to use at least f/7.1 for images I want to stitch.
Tilting the lens more than a few degrees also introduces vignetting to one side of the image. This can be quite pronounced at full tilt and wide aperture. It’s also modified if you’ve aded a bit of shift along the direction of tilt.
It’s there and something you should test for if using much tilt for macro work – it’s also one of the reasons I modified my TS-E90 several years ago, from the default setting of tilt and shift knobs being at 90º to each other.
More effects of movements
Forget auto exposure with any amount of tilt or shift. Moving the lens has unpredictable effects on DSLR camera metering. It can also mess with sensor phase based AF systems if you were thinking of using them to check focus.
I use the camera in fully manual mode with TS-E lenses. Once you get used to it, you’ll find it gives you a much closer feel for light levels and trusting your judgement.
Focusing the lens
The focus ring is good and solid, with a smooth movement.
The lens has just short of 180º throw from shortest focus to just beyond infinity. Beyond infinity, to allow for temperature variation in focus and specifically because some uses of tilt need the lens focussed ‘beyond’ infinity.
The throw of the new lens is reduced at the far end (new vs. old TS-E 90).
My problem with all three new TS-E lenses is that they seem to have been designed with much less attention to the needs of people wanting to focus in the 5-30 metre range. I’ll come back to this later, but given the similarity between all three lenses, it seems to be a deliberate choice, and from feedback I’ve had from some users of the lenses, not a welcome one.
That said, once I’d got used to the very short lens focus throw at the far end, it became more of a mild irritation, than any hindrance to my photos.
There’s some interesting architecture not far from where I live, which gives me a consistent place to test new lenses and equipment. The campus of De Montfort University is just beyond this park (f/6.3 with ~8mm vertical shift).
A 100% crop gives a good feel for how well this lens performs.
Note: Most of the images here have had a small amount of sharpening included in the web images, when processing from RAW or resizing to show in this review.
If I’ve the chance, I often prefer to take two shots, with the lens shifted by different amounts. These are simple to stitch together in Photoshop and give me images around 80MP.
Of course, if I know the images are ‘just for web use’ it may be easier to crop what I need from a shot with a wider lens, However, you don’t need a client calling up and asking if you’ve a version of an image ‘with more detail’ too often to err on the side of higher resolution…
Some stitched versions of images around the location above, give a feel for the use of a longer lens.
Taken from the same point as above, I checked this one very carefully for stitching errors and simply couldn’t find any.
If you work around f/7-11 you can still get quite a usable depth of field if you are careful with focussing.
This is the student health centre at the university.
As part of my testing the Benro geared head, I shot this short video, showing me levelling the camera and then adding upwards shift at the end, to get the composition for a single shot.
Nearby, at the back of the Queen’s building, this shot at f/7.1 has virtually no rectilinear distortion and in its resolution of detail convinced me that the new TS-E lenses are designed for full frame sensors well above 100MP.
Some student housing nearby, with the sun just out of shot.
No problems with lowered contrast or flare.
This view along the edge of the building might be a candidate for using a bit of tilt, but a quick check of my tilt tables (see later) suggested less than a degree was needed.
If the wall continued on past me I might well have considered it, but as it stands, the shot at f/7.1 is OK.
Using lens tilt
A more obvious candidate for using tilt is if you wanted to include the side of something like this van.
The shot is hand held with around 7º of tilt.
I could have got more of the side of the van in with a bit of up/down shift, or even stitched multiple shots.
How did I guess ~7º? I used my tilt tables for a 90mm lens (from my article focussing with tilt) The tables are reproduced at the foot of this article.
More extreme tilt and we’re into the realm of the ‘model world’ look. These are shots of DMU campus and nearby, taken from the roof of the gallery building.
Looking over the river (f/2.8).
Campus open space (f/2.8).
Two views of the top of the Queen’s building, both at f/2.8.
Full 10º tilt.
Same view, but with no tilt.
The image detail is impressive for a wide open lens – needless to say, if I was actually taking the photo for ‘real’ I’d be unlikely to be using f/2.8.
You can run the plane of focus along a level plane above or below the camera with a simple up/down tilt of the lens.
Here’s an attempt at running it along the plane of the table tops.
I’ve included this ‘not quite right’ version to show the typical issues people come across in setting up such shots.
With the camera level and the lens focussed at infinity the plane of focus is flat, with it’s distance below the camera set by the tilt angle of the lens (I used my tilt tables).
Once this is set, changing the focus of the lens makes the plane of focus tilt up or down.
Look again at the image above (click to open in a new window if need be). You’ll see that the table top nearest the camera is sharp, but in the distance, the plane of focus is at least a foot above the table top. You would fix this by changing the focus, not the tilt.
If you’d like to know more about how all this works I’ve written an article specifically dealing with focus for tilted lenses. Lots of detail and almost no maths!
One nice thing is that the more you use lenses like this the easier it becomes to visualise the plane of focus as a sort of ‘sheet of glass’ that cuts through the space in front of you and can be moved about.
Most times I use tilt though, it’s not at apertures as wide as this, since I’m wanting to give the impression of a deeper depth of focus rather than emphasise oddly out of focus bits of my image.
Model world type shots are curious to look at, but I’m firmly of the belief that they should lose their novelty appeal once you’ve done a few of them.
Moving back indoors from the rooftop shot above, I’ve the mass of crossing staircases inside the VJP building (see my collection of VJP images with lens choices explained for more).
90mm is quite a long focal length for interior photos, since you’re going to be giving the feeling of compressing space somewhat.
This view (@f/7.1) is focussed on the person, and there’s more than enough depth of field.
You can run the plane of focus along a wall or vertical structure with a left right swing of the lens.
Note below how the plane runs out into where the seats and litter bin are.
Obviously I’m exaggerating the effect to show it here, but hopefully you get the idea that not all tilt needs to make your shot look like a model…
Emphasising the wooden cladding (but look carefully and you’ll see that with this hand held shot, I’ve not got the closest part of the wood in focus).
A nearby corridor – up/down shift, stitched.
Looking down the stairs
As with the other new TS-E lenses, I’ve some examples taken with the camera at the top of my stairs, looking downwards.
The idea is that with the right amount of tilt (and focus setting) I can get the plane of sharpness to run along the stair edges.
The camera is tilted, and the lens even more so.
This is one where live view and the geared head are distinctly useful.
On a ‘paying job’ I might even be using the camera tethered to my laptop.
The setup for tilt is explained in this diagram (and the article focussing the tilted lens)
Here are the shots at f/2.8,f/4, f/5.6, f/8 and f/16 followed by 100% crops. These photos are all in the gallery at the end of the article for easier viewing.
The most obvious difference at this size is the strong vignetting at wider apertures and full tilt (~10º).
The 100% crops suggest that with a lot of tilt, finest detail is peaking from f/5.6 to f/8, but if you are using a lot of tilt then there’s nothing to be scared of of f/16 (I’ll push things to f/45 for the macro side of things).
Remember that the tilted plane of focus is best thought of as a wedge, where its thickness is related to the depth of field.
It’s thinner closer to the camera.
The new TS-E 90mm gets the word ‘Macro’ added to its name, and focuses closer than the old lens, giving a maximum magnification of 0.5x (1:2) rather than the approximate 1:3 of the older lens. The new lens has magnification ratios marked as well as focus distances (1:2 | 1:3 | 1:5).
I’ve used the same basic test setup as with the other two ‘Macro’ TS-E lenses.
For focusing and image capture I’m once again using Kuuvik capture – the advantage it gives me is that I can set multiple high resolution live view areas. By setting these across the range of the tilted object I can line it up precisely with the plane of focus.
This example shows a simple split view.
I’ve annotated this view to show how the maximum tilt of the plane of focus increases with distance from the lens.
Remember – the closer you focus, the less tilt of the plane of focus you get, with the same amount of lens tilt.
10º of lens tilt does not mean the focal plane is tilted by 10º
To get to around 1:1 magnification I’ve added 52mm of extension tubes (a 31mm and 21mm).
Tubes will lower your effective aperture (longer exposures) and the maximum amount of tilt of the image plane. I’ve looked at this with the old TS-E90 in a short articles about extension tubes and tilt.
A series of macro shots at full (10º) tilt, of a memory module, give a feel for depth of field and image quality.
Here’s the full set in one large image, with 100% crops below.
That’s my old TS-E90 in the background.
The 100% crops (also in the gallery later) show how finest detail start to drop off noticeably by f/16.
I can see circumstances where the 50MP images here could be downsampled and sharpened to make f/45 usable for web images. Diffraction blurring need not be the threat that some people seem to think it is ;-).
Moving the plane of focus – the offset tilt axis
The examples above have the plane simply tilted by swinging the lens to the left on a vertical axis.
However you can place the focal plane at an arbitrary position in the space in front of you. It’s quite difficult to work out the direction of the tilt axis needed, even if you have a good 3D visualisation ability.
My own preference is to use a flat plane with lots of detail, such as the stripboard shown earlier or even an iPad screen.
I just place the iPad where I want the plane of focus to go, and then place the subject where it is.
Rather than go through the details again, see the discussion in my TS-E135mm F4L Macro review.
Setting up tilt and focus visually can be quite tricky too – for close up work, my preference is to use an iterative tilt-shift focus technique after I’ve found the correct tilt axis. The split view live-view I mentioned earlier really helps with this.
I very rarely do comparative reviews (my reasons) but since I’ve used the TS-E90 F2.8 for quite a few years, I decided to pitch the two against each other for a few qualitative tests.
First up – there are the adjustments of the tilt and shift axes, which are now much easier to set (with a greater range) and can be altered without the need for a screwdriver.
Secondly, the increased size of the new lens means that I can’t use the attachment ring (shown unscrewed) to mount the MT-24EX macro flash.
The flash simply clips in place.
The photo above was taken with a KF-150 ring flash attached to the new TS-E90 (KF150 review).
The old lens doesn’t have the 10º of tilt – this shows up in this overhead shot of my macro test setup at the closest focus point of the old TS-E90.
You can see this in these two shots of the memory module at f/8. There is a bit of change in framing because of the camera and lens just resting on my desk (you’ll get precision when someone kindly sends me an optical bench ;-)
The tilt is less, but images look similar.
Let’s go to 100% views.
The new lens certainly has an edge at f/8, not massive, but having tried quite a few shots, it is there.
Now let’s be really unfair to the 1991 lens – this is the same set of photos, but at f/2.8.
Now at 100%.
Ah… this would explain why I hardly ever used my TS-E90mm F2.8 at f/2.8.
A trip outdoors lets me check ‘untilted’. The detail shots are from this garage door, across the street.
First at f/2.8.
Note the coloured fringing and lowered contrast from the old lens.
The old lens is still showing coloured fringing (look at the specular reflection on the lock).
Just one more of the old lens at f/5.6, where many of the aberrations have dropped off and we’re starting to show the detail from the 50MP sensor.
A quick set of 100% views at the corner of this unshifted image.
The new TS-E looked pretty much identical to the centre at f/4 and above, with just the vignetting I showed earlier.
Based on my use of the TS-E90 over the last few years, I’ll broadly stick to a maximum aperture of >f/6.3 and >f/8 if I’m shifting.
I used to think my TS-E90 F2.8 was a sharp lens… Well, I still do, just over much less of a range than the new lens.
So, having sent the TS-E90mm F2.8L Macro back to Canon, do I want to replace my old one?
Without quantitative testing I can’t be sure, but the TS-E90mm F2.8L Macro is probably the sharpest lens I’ve ever tested (it could have been the TS-E135, but I’m not quibbling).
The build quality is superb, and putting it next to the old TS-E90 you really do get a feel for progress in lens construction. It is well up to its ‘L’ designation and red ring.
If I have a complaint, it’s the limited throw of the focus ring, which is simply too short for easy manual focus, especially at non-macro distances.
Optically, I couldn’t fault the new lens, it’s performance at f/2.8 is exceptional, with only vignetting to really take note of.
Where the lens really does show its quality is once you start to add in a lot of tilt and shift, with the sharpness of the razor thin DOF at f/2.8 – 4 contrasting nicely with the well controlled OOF softness. Mind you, at full tilt, the OOF areas take on a very distinctive look that may or may not suit your tastes.
All in all a lens that significantly expands the capabilities of the old version, but you will need to make a serious effort in learning how to use this type of lens to get the best from it.
PS – Will I be buying a new TS-E90mm? Unfortunately not – I just don’t do enough paying work that relies on my existing TS-E90, where it would make a difference. Remember that my kit has to earn me a living, so there is always a disconnect between what I’d like and what our finance and marketing department (aka Karen, my wife) will OK ;-) Thanks to Canon UK for loaning me the three new TS-E lenses for the reviews.
Tilt tables for current TS-E lenses
For a detailed explanation of how these tables can be of help, see my ‘How tilt works‘ article.
Have my reviews piqued your interest in camera movements but you can’t justify a lens like this?
I learned a lot about camera movements by making (for a few pounds) a simple DSLR adapter for a view camera I’d bought from eBay – more tilt and more shift, just not so practical.
Click on images to open at larger size.
Never miss a new article or review - Sign up for our occasional (ad-free) Newsletter
Enjoyed this article?
All the latest articles/reviews and photo news items appear on Keith's Photo blog
We've a whole section of the site devoted to Digital Black and White photography and printing. It covers all of Keith's specialist articles and reviews.
For All about using tilt and shift - articles/reviews about tilt/shift lenses
Articles below by Keith (Google's picks for matching this page)