Canon EF 100mm f/2.8 Macro USM vs Tamron SP Autofocus 90mm f/2.8 DI Macro Lens

Due to the fact that in many (but not all) markets the Tamron 90mm f/2.8 Macro is substantially cheaper than the Canon EF 100mm f/2.8 Macro USM the question as to how the two lenses compare seems to crop up extremely often on the photo.net boards. What’s more, as an owner of the third party offering I have often wondered how it compares to the Canon lens. Like many others before me I have occasionally toyed with the idea of upgrading one day, however instead I decided to rent the Canon lens and do a side by side comparison. The tests were carried out with my EOS 10D and all the samples shown are unprocessed and straight from the camera.

Both lenses have excellent reputations however the Canon lens is generally considered to be the better product, so I was very curious to see just how much there is between them.

Handling

The Canon lens is the larger and heavier of the two, and it’s front element is exposed right at the front of the lens, whilst the Tamron’s front element is very deeply recessed inside the lens barrel. This presumably has the effect of shielding it from flare without the supplied lens hood being fitted, however it does reduce the effective working distance from the subject a little since there is an extra couple of centimetres beyond the front element.

The Canon should always be fitted with its optional (and expensive) lens hood, since the front element has absolutely no protection from stray light without this.

In terms of handling, the Canon is the obvious winner since it has USM and FTM. Autofocus is quick and accurate, although not quite as fast as with my 300F4 L. This relative (and slight) slowness is probably down to the enormous focusing range, from 1:1 to infinity. Remarkably, the lens seems to autofocus right down to it’s minimum focusing distance fairly reliably, meaning that it is feasible for AF to be used for macro work. Obviously in most cases manual focus will still be best when working at high magnification, but I can see where AF might be useful with moving subjects.

The Tamron’s AF performance, while not on a par with the Canon, is also pretty good. AF is fast but slightly noisy (no USM motor here), however it is only effective at longer ranges, at close range the AF hunts a lot and often fails to lock on. For macro work with this lens manual focus is the only viable option. The Tamron also does not have FTM, instead it uses a clutch mechanism on the focusing ring itself to switch between AF and MF. This system works well enough but is not as easy or ergonomic as Canon’s full time manual focusing.

Both lenses have excellent manual focusing rings, which are easy and pleasant to use. I personally prefer the smaller build and lighter weight of the Tamron lens, but overall the Canon is a much better mechanical design.

Sharpness

For this test the lenses were mounted on a tripod and a cable release was used. I was very careful to check for accurate focus before each shot.

The lenses have different focal lengths, in this test though I did not try and compensate for this by moving the camera forward, it seemed more logical to keep it in the same spot and simply ignore the difference in magnification. It is still perfectly possible to judge the performance of both lenses despite this difference.

This is what the full image looks like:

Below is a 100% crop from the centre portion of the frame, in each case the Tamron sample is on the left, the Canon is on the right. No post processing has been done, these images are straight off the camera.

F2.8

F5.6

F8

F16

Next up is a sample from the corners, again the Tamron is on the left, the Canon is on the right:

F2.8

F5.6

F8

F16

It seems to me to be a dead heat, in this test at least. Both lenses perform very similarly, they are very slightly soft and dull at F2.8, but sharpness and contrast improve by F5.6, F8 doesn’t seem any different but by F16 there is some softness creeping back in. The corners are slightly softer than the centre, but performance seems to still be excellent in both cases.

I have a feeling this result may well upset a few Canonites, but there it is. Both lenses are clearly excellent and the Tamron lens can quite easily match its more expensive cousin, even wide open. Results on a higher resolution full frame camera might well be more revealing, but on my EOS 10D at least there seems to be very little between the two in terms of sharpness and contrast.

High magnification

disclaimer: This test was carried out for the sake of completeness, it wouldn’t make sense testing macro lenses without some high magnification samples. However there are many problems with any test done at a magnification close to 1:1, simply because it is extremely difficult to focus with 100% accuracy at this level of magnification and depth of field is so extremely shallow that the slightest focusing error would render the test meaningless. In the light of this I will state that this test should only be viewed as representative of my own efforts and should not be held as being a guaranteed benchmark.

For this test I mounted the camera on a Gitzo Explorer tripod, set up a 550EX to the side of the subject and connected the camera to my computer via USB. To ensure the best possible focus I used angle finder C to manually focus and then took a series of test frames with focus bracketing which were then compared on the computer screen at 100%, and the focus was fine tuned until it was as good as I could get it. Mirror lockup and a cable release were used, although using the 550EX as the lightsource should help to minimise vibration problems anyway.

I also tried autofocus but at this magnification it wasn’t reliable enough to give results that could be used usefully in a lens test. While the Canon lens autofocused far more reliably at this magnification than the Tamron ever could, manual focusing proved to be much more accurate still (AF was always slightly soft in comparison). I presume that the tolerances for AF just aren’t fine enough for such demanding conditions as are encountered at 1:1.

This is the full frame, a coin taken close to 1:1.

These are 100% crops from the centre of the image:

F2.8

F4

F5.6

F8

F16

Interestingly, although the Canon lens appears to be very slightly sharper wide-open, once the lenses are stopped down a little the Tamron seems to have far better contrast, colour and detail. Contrast from the Canon lens doesn’t really catch up until F8, which seems to be the sweet spot for that lens, while the Tamron actually seems best at F5.6. At F16 both lenses lose some sharpness due to diffraction although the Tamron still appears slightly sharper.

While these results can’t be held as being conclusive, it does appear as if the Tamron is the better lens for macro work. One caveat being of course that lens sharpness is one of many variables for this type of work, and indeed typical macro apertures will be around F16 where both lenses have lost some resolution. At this level most shooters will have enough problems with depth of field and focusing to lose too much sleep over minor differences in lens sharpness.

To illustrate this point here is a 50% sample of the F16 files with unsharp mask applied, as you can see the differences are still visible but very subtle, and sharpness is very good. The Tamron lens does undeniably show more detail though.

F16 @ 50% USM applied.

I would also have liked to show some corner samples to give an idea of corner sharpness and the flat field ability of each lens, but unfortunately this wasn’t possible as I don’t have the means of guaranteeing that my samples were parallel with the film plane. At these magnifications there is no room for error, and the only way I could guarantee that what was being tested was the lenses as opposed to my ability to line up the subjects with the film plane would be with a precision made copy stand – which is an item I don’t own.

Bokeh, colour and flare

Here is a close up sequence comparing both lenses at varying apertures, again the Tamron is on the left and the Canon on the right:

F2.8

F4

F5.6

F8

Some general notes: in both cases colour and contrast visibly improve by F4, but the Canon lens has a warmer colour balance and more saturated colours. Bokeh seems very similar with both lenses in this particular test. It’s smooth and pleasing, the Tamron’s may be ever so slightly softer.

This following sequence was designed to test both bokeh and flare resistance:

F2.8

F4

F5.6

F8

F16

F32

Unfortunately there appears to be about 1/3 of stop difference in exposure between these two sequences, which makes colour differences unreliable. Having said that the Canon does seem to hold detail better in the brightest parts, beneath the flame of the nearest candle.

Bokeh is again very similar with both lenses, wide open both have a pleasing rendition of the highlights. Stopped down the Canon has a nicer star pattern on the highlights, with fewer points arranged in a neater pattern.

Wide open the Canon has a tiny bit of flare by the nearest flame, however stopped down this seems to disappear. The Tamron however seems to suffer from very visible flare artefacts when stopped down. This is the only really disappointing result so far from the Tamron lens, and gives an indication of why that front element is so deeply recessed. The Canon is a definite and dramatic improvement in terms of flare control.

I have also tried a couple of other tests pointing the lenses straight at light sources and both do actually seem to resist flare pretty well in most circumstances, these are full frames:

Sun @ F32

Lightbulb @ F32

Working Distance, focal length and background control

For this test, I measured the working distance from the front of the lens housing (not the front element itself or the film plane like in many tests) to the point of focus at maximum magnification. The Canon has 15cm (just under 6 inches) free between the front of the lens and the subject, while the Tamron only has 11cm (just over four and a quarter inches). The Tamron lens is actually 2.5cm (1 inch) inside the housing, which could have yielded more working distance had the lens been designed differently, but obviously Tamron decided flare control was a more important issue than working distance.

Both these measurements were done without lens hoods on, if you fit a hood to the Tamron it will almost touch your subject at 1:1, the Canon has a little more clearance. Working distance is only really an issue with live subjects such as insects, however it is important to many macro shooters and if you are one of those who needs as much distance as possible the Canon lens is the better choice.

There is however one situation where less working distance is actually a good thing, which is working with flash at high magnifications. The closer your flashgun is to the subject, the softer the light. If you have an external flashgun mounted on a bracket over or near your lens, or a unit such as an MT24-EX, every inch closer that the flash heads are to your subject will result in better and softer light.

Below is an example of a photo taken with a 550EX mounted on a cheap bracket to position it very close to the subject, and with an small reflector (cannibalised from a Cokin filter box) tied by elastic bands on the other side of the lens barrel to provide some fill. This is a very low tech and low cost solution to macro flash that produces some very pleasing and soft light. Obviously the further away from this rig the subject is, the less attractive the light becomes.

Macro flash, hand held @ F5.6

Another important consideration with macro photography is focal length: longer focal lengths give you much more control over the background. This is because at any given magnification, a longer lens will include less of the background in relation to the subject. So where a 50mm lens might make it difficult to control what background you have behind your subject without a great deal of repositioning, with a 300mm lens the smallest shift in position will radically change what is visible behind your subject.

I had harboured some concern about the Canon’s internal focusing mechanism which focuses closer by reducing the focal length of the lens. The advantage of this system is that the lens does not change shape when focusing, and remains constant in length all the way to 1:1. The obvious disadvantage is that a shorter focal length results in less background control, however since the Canon is a longer focal length than the Tamron at the outset it should still retain the advantage unless the IF reduces the focal length of the lens very drastically indeed.

The following image was taken at 1:2 magnification, with the remote control unit placed about 20 cm (8 inches) behind the ruler. The aperture was set to F32 to maximise the depth of field. As usual the Tamron sample is on the left and the Canon is on the right.

focal length comparison @ 1:2

It seems that the IF does not shorten the focal length of the Canon lens too drastically, and that even at high magnification the focal length is still clearly longer than the Tamron’s 90mm. The Canon lens includes less of the background than the Tamron does, so for background control the Canon is again at a slight advantage.

It is also worth noting that the Tamron lens focuses closer by extension, which means that the lens barrel lengthens as magnification increases, and it’s length goes from 9cm (3.5 inches) at infinity to 14cm (5.5 inches) at 1:1. This does not affect the working distance mentioned earlier since I have measured this from the front end of the lens barrel, not from the film plane.

Compared to…

One caveat of testing lenses with a digital camera is that the limitations of the camera sensor will affect the test and subsequent results. It’s possible for instance that one lens is actually much sharper than the other but that my camera simply isn’t able to resolve enough detail to demonstrate this. However I know from my own experience that lens quality does indeed show up fairly obviously with the 10D. Whilst I do believe that both the lenses on test here are probably capable of outresolving my camera’s sensor, I thought it would be of some benefit to compare the performance of these two excellent lenses against a potentially inferior model. Here is a shot which compares the two primes against a Canon 28-80 consumer zoom. The difference in quality is very apparent.

Compared with a consumer 28-80 zoom @ F5.6

Conclusion

This exercise has surprised me somewhat in that I knew the Tamron lens was considered pretty good, but I did expect the Canon model to be significantly better. In this test at least, this hasn’t proven to be the case. Both lenses display excellent sharpness and contrast, and the only significant advantage the Canon has is in it’s vastly superior flare control. Other differences are fairly minor: the Canon lens has better saturation, more working distance and better AF – it is clearly a superior product – but the Tamron does very well to come so close (and to even surpass the Canon lens at high magnification) especially when you consider that it is up against one of Canon’s very best.

All in all I would say that in those places where the price of both lenses is fairly close the Canon is probably the better choice, but where there might be a significant price difference (such as in the UK where I live) then the Tamron is certainly worth considering since it delivers oustanding performance to rival and sometimes better it’s more expensive counterpart.