Studio Tests - 35mm full-frame format

The Canon TS-E 24mm F3.5 L II is a lens which we suspect will be used mainly on high resolution full frame cameras such as the EOS-1Ds Mark III and EOS 5D Mark II. We've therefore chosen to test it only on full frame, and not an APS-C body as we usually would. As a start we're showing our conventional test results, covering the central region of the lens's image circle.

The TS-E 24mm F3.5 L II puts in a hugely impressive performance in our studio test. Chromatic aberration is virtually nonexistent (lower than some 50mm primes), distortion is near-zero, vignetting is low and central sharpness impressively high even wide open; the only slight concern is a slight drop off in sharpness towards the corners at larger apertures. Nevertheless it's rather better than the EF 16-35mm F2.8 L II, with sharper corners and less distortion, and is no slouch even when compared to Canon's excellent EF 24mm F1.4L USM, with slightly softer extreme corners but notably lower CA.

Resolution Central sharpness is extremely high wide open, but the extreme corners are just slightly soft in comparison. Stop down to the sweet spot around F8 - F16 though, and the lens provides excellent results right across the frame; by F22 the image has softened noticeably due to diffraction.
Chromatic Aberration Lateral chromatic aberration is to all intents and purposes nonexistent.
Falloff We consider falloff to become perceptible when the corner illumination falls to more than 1 stop less than the centre. As we'd expect for a lens with a large image circle, falloff is very low, at just 1.3 stops wide open.
Distortion Distortion is so low as to be negligible.

Macro Focus

A 24mm tilt and shift lens is probably your very last port of call for close up work, so we've omitted our usual macro test. The lens's specified closest focus distance is 0.21m, and this results in a very limited working distance of just 6cm from the subject to the lens, making lighting problematic. Maximum magnification is quoted by Canon as 0.34x, which should equate to an image area of 106 x 71mm on full frame.

Extended image circle tests

There's little point in testing a tilt and shift lens without measuring its performance at the extremes of shift, so we've extended our usual regime accordingly. However as this data is no longer directly comparable with our normal studio test results, we're chosen to present it in a slightly different form. It's also important to appreciate that this test is at a relatively close focus distance of ca. 1.25m, which is a severe test of field flatness (as the corners of the chart are about 1m further away from the lens than the center).



This graph shows the sharpness profile measured from the lens's optical centre towards the edge of the image area that is used when the lens is fully shifted along the image diagonal. Corner softness now becomes a genuine concern, and while the majority of the image area will be acceptably sharp at F11, it's necessary to stop right down to F16 to get good sharpness in the extreme corners of the frame.

This graph corresponds to an image circle approximately 67mm in diameter, as opposed to the 43mm of FX. In practice, though, full diagonal shift is not a common scenario in normal use, and the image circles used when the lens is fully shifted either vertically or horizontally are slightly smaller. The vertical red lines indicate the extremes of the image circles for various configurations - full horizontal shift (~65mm, 96%), full vertical shift (~60mm, 89%), and for comparison the normal unshifted position (43mm, 64%). These are illustrated in the diagram below, with the lens's image circle in blue and the frame area in white. (Note also that when the lens is shifted, the central sharpest region of the lens will not correspond to the centre of the frame.)

(43mm, 64%)
Vertical shift
(60mm, 89%)
Horizontal shift
(65mm, 96%)
Diagonal shift
(67mm, 100%)


This test represents the recorded full-frame area with the lens shifted by 12mm to the right, and no tilt applied. There's a slight pincushion effect visible at the extreme corners, but it's not huge. It's also worth noting that the camera to chart distance used here is extremely short (less than 1m), and distortion tends to increase at close focus distances, so overall this is a very impressive result.


(2.3 stops)
(2.3 stops)
(1.7 stops)
(1.3 stops)
(1 stop)

Again this is the full-frame image area with the lens shifted 12mm to the right and no tilt applied. Not surprisingly for a lens with such a large image circle and filed of view, there's some fairly strong falloff wide open, but by the time you've stopped down to normal working apertures of around F8-11, there's not a lot to worry about.

In practical use, however, a it's possible to get significant mechanical vignetting from the lens mount and mirror box when tilt and shift are applied simultaneously along the same axis. Canon warns of this in the lens's instruction manual, with a diagram of the combinations of tilt and shift that can be safely used, so it's worth taking note of these. The examples below show exactly what kind of vignetting you get when you ignore Canon's advice and combine full tilt and shift in parallel - it's pretty severe, so needs to be avoided.

Mechanical vignetting, Canon EOS-1Ds Mark III
Right, F3.5
Right, F8
Right, F16
Down, F3.5
Down, F8
Down, F16

Checkerboard crops

As a visual illustration of the image quality across the full image circle, here are our familiar checkerboard crops taken from the center and extreme corner of the image circle at F3.5, F8 and F16 (do bear in mind these are 100% crops from a 21Mp EOS 5D Mark II frame). Central sharpness is excellent wide open and even better at at F8, but the extreme corner is nothing to write home about, and never quite reaches the same level of definition even at F16. Falloff is obviously extremely strong at F3.5, but progressively diminishes on stopping down. Most impressively, though, there's essentially no visible chromatic aberration even in the extreme corner, which is frankly astonishing for such a lens. The very slight diamond shape of the corner crop shows the tiny degree of pincushion distortion right at the extremity of the frame, which is unlikely ever to be a problem in real life.

Lens Center
Extreme Corner