There are more issues at shorter focal lengths than at long, partly because of the more difficult design optimisations, but also because Canon has a limited range of fixed focal length lenses below 24mm. In general although the zooms looked at here are good the prime lenses have better performance as well as being smaller.

The zooms (other than the kit zoom) perform very well over the APS-C frame but show much more differentiation over the 35mm format frame coverage where the prime lenses are clearly superior.

Looking at the relative performances the only really bad (relatively) lens is the 18-55mm kit zoom. The other zooms fall into the second tier of performance being 10-20% worse than the primes.

At maximum aperture for many cases the major maximum aperture differentiation for contrast, detail and bokeh are beyond the frame of APS-C sensors.

Generally, all the large aperture primes performed well, even the 28mm f1.8 which had a lacklustre stopped down performance.

Also take a look at:

Photodo / Canon MTF Cross Validation

Contax, Nikkor and Leica R lenses vs Canon

List of Photodo Data Comparisons

Back to TOC

Lens Limited Resolution

It is interesting to find out which lenses are limiting the image resolution, rather than the sensor being the dominant limiting factor. To work this out relies on modelling the lens spatial frequency response as a low pass filter. There are a number of problems with this, firstly the lens frequency response will be a composite of several functions and not a simple function. Secondly, Canon have only provided MTF data for two points, 10 and 30 lp/mm, this is not much to base a trend on.

So the accuracy of this analysis is limited, and at best it perhaps provides an indication of how quickly the lens MTF response decays to minimal contrast, a value of MTF=0.1 is chosen. At worst, the plots may be trying to squeeze to much data out of too little information and so be meaningless.

In these plots spatial frequencies (f) above 80 lp/mm are not plotted as this method is probably grossly inaccurate when the MTF gradient between 10 and 30 lp/mm is low.

Sensor resolution is limited to about 40 to 80 lp/mm for the EOS 20D and EOS 1Ds depending on what assumptions are made in respect if anti-alias filter losses and losses due to geometric orientation.

This is an important factor in terms of being able to get the best out of future sensor developments. Clearly sensor resolution limits are less of a problem for a larger format of a given sensor resolution.

10mm

At 10mm the only information we have is from the 10-22mm zoom, This seems to be mostly above the sensor limit.

17mm/20mm

Here some of the lenses are falling below the sensor limit as the edge of the lens is approached, the 16-35mm zoom seems to be the worst offender.

24mm

Here two zooms show up most significantly as likely to be a limiting factor, the 24-70mm and the 17-40mm.

28mm

Here again the two zooms show up as most significantly limiting, the 24-70mm and the 17-40mm and also the 28mm f1.8.

35mm

Here the most obviously limiting is the 17-40mm zoom.

50mm to 200mm

There are no studied lenses that look likely to be resolution limiting in this range.

Lens Limited Resolution Conclusions

The results of this section are somewhat suspect and should be treated with some caution.

However, it seems this is only an issue at and below 35mm with current sensor technology with the most significantly resolution limiting lenses being zooms down to 24mm focal lengths.

As around 20mm focal lengths, the prime lenses may start limiting very fine resolution at the edges of the 35mm frame.

This does not generally seem to be a major issue for APS-C sized sensors at the lens frequency response remains good up to the edge of the frame.

Note that it is also possible to use this fit method to estimate the MTF at detail resolutions higher than 30 lp/mm to scale the 35mm fine detail contrast standard to the APS-C sensor size, this would be an equivalent resolution of 48 lp/mm for a 1.6X crop factor sensor. This does not tend to provide much information over that obtained directly however, as lens performance is less well differentiated.

Usability, Costs and Reviews

Back to TOC

Times have changed since I was last in the market for a lens system. In those days lenses were manual focus, almost exclusively via linear extension, all lenses indicted the focused distance. Also almost all (can’t think of any exceptions) included depth of field markings, even zooms as most were of the trombone style.

Things have changed, now one has to consider if these features are available or not, plus there is a variety of auto focus actuators (motors).

Below is a table of lens information for the lenses studied. Lenses with a mass 700g and over are highlighted in orange, previous experience shows these become tiresome to use and have around the neck. Anything over 1Kg is highlighted in red.

See IMATEST and DxO Analyzer for a discussion of these lens test methods.

Make	Model	Dia mm	Length mm	Volume mm^3	Mass g	Filter Size	FE Rotation	Focusing	Reviews / Tests
CANON	EF 200mm f/2.8 L USM MK2	83.2	136.2	740481	765	72mm	AN	RFS	[1] [2] [3]
CANON	EF 135mm f/2L USM	82.5	112.0	598709	750	72mm	AN	RFS	[1] [2] [3] [4] [5]
CANON	EF 135mm f2.8 with Softfocus	69.2	98.4	370081	390	52mm	AN	RFS	[1] [2] [3]
CANON	EF 100mm f/2.8 MACRO USM	79.0	119.0	583299	600	58mm	N	IFS	[1] [2] [3] [4]
CANON	EF 85mm f/1.2L II USM	91.5	84.0	552418	1025	72mm			[1] [2] [3]
CANON	EF 85mm f/1.8 USM	75.0	71.5	315877	425	58mm	AN	RFS	[1] [2] [3]
CANON	EF 70 200 mm f/2.8 L USM	84.6	193.6	1088268	1310	77mm		IFS	[1] [2] [3] [4]
CANON	EF 70 200 mm f/2.8 L USM IS	86.2	197.0	1149663	1470	77mm		IFS	[1] [2] [3] [4]
CANON	EF 70 200 mm f/4 L USM	76.0	172.0	780271	705	67mm	N	IFS	[1] [2] [3] [4]
CANON	EF 70 200 mm f/4 L IS USM	76.0	172.0	780271	760	67mm	N	IFS	[1] [2]
CANON	EF 50mm f/1.2L USM	85.4	65.5	375186	545	72mm			[1] [2]
CANON	EF 50mm f/1.4 USM	73.8	50.5	216020	290	58mm		OLE	[1] [2] [3] [4]
CANON	EF 50mm f/1.8 II	68.2	41.0	149776	130	52mm		OLE	[1] [2] [3] [4]
CANON	EF 35mm f/2	67.4	42.5	151635	210	52mm		OLE	[1] [2] [3]
CANON	EF 35mm f/1.4L USM	83.5	77.4	423842	550	72mm	AN	RFS	[1] [2] [3] [4] [5]
CANON	EF 28mm f/1.8 USM	73.6	55.6	236549	310	58mm	AN	RFS	[1] [2] [3] [4]
CANON	EF 28mm f/2.8	67.4	42.5	151635	185	52mm		OLE	[1] [2] [3]
CANON	EF 24 70 mm f/2.8 L USM	83.2	123.5	671434	950	77mm		FFM	[1] [2] [3] [4] [5]
CANON	EF 24-105mm f/4L IS USM	83.5	107	585931	670	77mm	N	IFS	[1] [2] [3]
CANON	EF 24mm f/1.4L USM	83.5	77.4	423842	550	77mm	AN	RFS	[1] [2] [3]
CANON	EF 24mm f/2.8	67.5	48.5	173556	270	58mm	AN	RFS	[1] [2]
CANON	EF 20mm f/2.8 USM	77.5	70.6	333041	405	72mm	AN	RFS	[1] [2] [3] [4]
CANON	EF-S 18-55mm f/3.5-5.6	69.0	66.2	247540	190	58mm		OLE	[1] [2] [3] [4] [5]
CANON	EF-S 17-55mm f/2.8 IS	83.5	110.6	605645	645	77mm			[1] [2] [3]
TOKINA	17mm f/3.5 AT-X PRO (CANON AF) [1]	84.0	57.0	315881	435	77mm		Internal Floating Element	[1]
CANON	EF 17-40mm f/4L USM	83.5	96.8	530076	500	77mm	N		[1] [2] [3] [4] [5] [6]
CANON	EF 16 35 mm f/2.8 L II USM	88.5	111.6	686589	635	82mm	N	IFS	[1]
CANON	EF 16 35 mm f/2.8 L USM (Disc)	83.5	103.0	564027	600	77mm	N	IFS	[1] [2] [3] [4] [5a] [5b]
CANON	EF-S 10 22 mm f/3.5-4.5 USM	83.5	89.8	491744	385	77mm		IFS	[1] [2] [3] [4]

Make	Model	AF Actuator	FTMF	Distance Scale	DOF Scale	Zoom Control	Extends When Zoomed	Max Mag	Closest Focus M	Dec 04 Price UOS
CANON	EF 200mm f/2.8 L USM MK2	USM	Y	Y	Y	N/A		0.16	1.5	£525
CANON	EF 135mm f/2L USM	USM	Y	Y	Y	N/A		0.19	0.9	£699
CANON	EF 135mm f2.8 with Softfocus	AFD	N	Y	Y	N/A		0.12	1.3	£235
CANON	EF 100mm f/2.8 MACRO USM	USM	Y	Y	Mag Scl	N/A		1	0.31	£394
CANON	EF 85mm f/1.2L II USM	USM	Y	Y	Y	N/A			0.95	£1449
CANON	EF 85mm f/1.8 USM	USM	Y	Y	Y	N/A		0.13	0.85	£275
CANON	EF 70 200 mm f/2.8 L USM	USM	Y	Y	N	RR	N	0.16	1.5	£899
CANON	EF 70 200 mm f/2.8 L USM IMAGE STABILISER	USM	N	Y	N	RR	N	0.17	1.4	£1,370
CANON	EF 70 200 mm f/4 L USM	USM	Y	Y	N	RR	N	0.21	1.2	£599
CANON	EF 70 200 mm f/4 L IS USM	USM	Y	Y	N	RR	N	0.21	1.2	£989.95 [Aug 06]
CANON	EF 50mm f/1.2L USM	USM	Y	Y	Y	N/A		0.15	0.45	£1149.95 [Aug 06]
CANON	EF 50mm f/1.4 USM	MUSM	Y	Y	Y	N/A		0.15	0.45	£250
CANON	EF 50mm f/1.8 II	MM	N	N	N	N/A		0.15	0.45	£69
CANON	EF 35mm f/2	AFD	N	Y	Y	N/A		0.23	0.25	£219
CANON	EF 35mm f/1.4L USM	USM	Y	Y	Y	N/A		0.18	0.3	£1,099
CANON	EF 28mm f/1.8 USM	USM	Y	Y	Y	N/A		0.18	0.25	£325
CANON	EF 28mm f/2.8	AFD	N	Y	Y	N/A		0.13	0.30	£140
CANON	EF 24 70 mm f/2.8 L USM	USM	Y	Y	N	RR	Y	0.29	0.38	£930
CANON	EF 24-105mm f/4L IS USM	USM	Y	Y	N	RR	Y	0.23	0.45	£749 Dec 05
CANON	EF 24mm f/1.4L USM	USM	Y	Y	Y	N/A		0.16	0.25	£950
CANON	EF 24mm f/2.8	AFD	N	Y	Y	N/A		0.16	0.25	£315
CANON	EF 20mm f/2.8 USM	USM	Y	Y	Y	N/A		0.14	0.25	£339
CANON	EF-S 18-55mm f/3.5-5.6	MM	N	N	N	REFE	Y	0.28	0.28	#N/A
CANON	EF-S 17-55mm f/2.8 IS	USM	Y	Y	N	RR			0.35	?
TOKINA	17mm f/3.5 AT-X PRO (CANON AF) [1]		Focus Clutch	Y	Y	N/A			0.25	£278
CANON	EF 17-40mm f/4L USM	USM	Y	Y	N	RR	N	0.24	0.28	£529
CANON	EF 16 35 mm f/2.8L II USM	USM	Y	Y	N	RR	N	0.22	0.28
CANON	EF 16 35 mm f/2.8L USM (Disc)	USM	Y	Y	N	RR	N	0.22	0.28	£999
CANON	EF-S 10 22 mm f/3.5 4.5 USM	USM	Y	Y	N	FR	N	0.17	0.24	£555

Characters	Key
Y	Yes
Y?	Probably but Canon Data Contradictory
N	No
AN	Assume No
OLE	Overall linear extension
RFS	Rear focusing system
IFS	Inner focusing system
FFM	Front-focusing method
R EFE	Ring extending front element
RR	Rear Ring
FR	Front Ring
MM	Micro Motor
AFD	Arc Form Drive
USM	Ultrasonic Motor
MUSM	Micro Ultrasonic Motor

Notes	Explanation
[1]	Canon Version still has aperture ring, no sign of how this is operated on the EF mount.

Lens Characteristic	Explanation
FE Rotation	Front Element Rotation: A problem with polarizing or graduated filters. Note Front group rotational extension
Focusing	The focusing method used in the design. Internal, floating and rear focusing tends to be preferred and allows more opportunity for the designer to reduce aberrations across the focusing range,
AF Actuator	The autofocus motor type. These vary in quality, speed and noise level. General order of preference from least to most desirable is MM, AFD, MUSM, and USM.
FTMF	Full-time manual focusing. This is for me a major feature, available only with USM lenses but not all of them.
Distance Scale	A major problem if not supported as it is then impossible to calculate hyperfocal distances or depth of field. (see below).
DOF Scale	Depth of Field or Hyperfocal distance markings. An essential feature for landscape work. Not a major problem if unavailable for 100mm focal lengths and above as the DOF is very small anyway. Highly desirable for 50mm and below. As long as the lens has a focusing distance scale then tables can be used, although this will be fiddly in the field.
Zoom Control	Almost all seem to be ring control, either at the front or the back. It is not clear from Canon’s data if the front element extends or rotates with a front ring control, in all likelihood it does not.
Max Mag	The maximum lens magnification in terms of focusing distance, and if applicable zoom adjustment.
Closest Focus M	The closest focusing distance, this may be different at different focusing distances for zooms.

We can see from the lens data table that only the two short zooms weigh in under the 700g limit and only two longest primes are just over this limit. In terms of upper body strength requirements the 24-70mm and two 70-200mm zooms make significant demands. I am not sure I could hold the 70-200mm f2.8 zooms at the ready for very long.

Trades

Back to TOC

To get a feel for the basics here are some zoom and prime based systems with totals for volume, mass and cost. There are lots of combinations and variations not covered but this gives an idea. Sum values for volume (mm^3), mass (g) and cost (GBP). Neglecting the 10-22mm and 100mm Macro lenses which are specialised.

Zoom System 1

CANON	EF 70 200 mm f/2.8 L USM IMAGE STABILISER	86.2	197.0	1149663	1470	£1,370
CANON	EF 24 70 mm f/2.8 L USM	83.2	123.5	671434	950	£930
CANON	EF 16 35 mm f/2.8 L USM	83.5	103.0	564027	600	£999
	SUM			2385125	3020	£3,299

Zoom System 2

CANON	EF 70 200 mm f/2.8 L USM	84.6	193.6	1342324	1310	£899
CANON	EF 24 70 mm f/2.8 L USM	83.2	123.5	671434	950	£930
CANON	EF 16 35 mm f/2.8 L USM	83.5	103.0	564027	600	£999
	SUM			2577785	2860	£2,828

Zoom System 3

CANON	EF 70 200 mm f/4 L USM	76.0	172.0	780271	705	£599
CANON	EF 24 70 mm f/2.8 L USM	83.2	123.5	671434	950	£930
CANON	EF 16 35 mm f/2.8 L USM	83.5	103.0	564027	600	£999
	SUM			2015733	2255	£2,528

Zoom System 4

CANON	EF 70 200 mm f/4 L USM	76.0	172.0	780271	705	£599
CANON	EF 24 70 mm f/2.8 L USM	83.2	123.5	671434	950	£930
CANON	EF 17-40mm f/4L USM	83.5	96.8	530076	500	£529
	SUM			1981781	2155	£2,058

Prime System 1

CANON	EF 200mm f/2.8 L USM MK2	83.2	136.2	740481	765	£525
CANON	EF 135mm f/2L USM	82.5	112.0	598709	750	£699
CANON	EF 85mm f/1.8 USM	75.0	71.5	315877	425	£275
CANON	EF 50mm f/1.4 USM	73.8	50.5	216020	290	£250
CANON	EF 35mm f/1.4L USM	83.5	77.4	423842	550	£1,099
TOKINA	17mm f/3.5 AT-X PRO (CANON AF)	84.0	57.0	315881	435	£278
	SUM			2610810	3215	£3,126

Prime System 2

CANON	EF 200mm f/2.8 L USM MK2	83.2	136.2	740481	765	£525
CANON	EF 135mm f/2L USM	82.5	112.0	598709	750	£699
CANON	EF 85mm f/1.8 USM	75.0	71.5	315877	425	£275
CANON	EF 50mm f/1.4 USM	73.8	50.5	216020	290	£250
CANON	EF 35mm f/2	67.4	42.5	151635	210	£219
TOKINA	17mm f/3.5 AT-X PRO (CANON AF)	84.0	57.0	315881	435	£278
	SUM			2338602	2875	£2,246

Prime System 3

CANON	EF 200mm f/2.8 L USM MK2	83.2	136.2	740481	765	£525
CANON	EF 135mm f/2L USM	82.5	112.0	598709	750	£699
CANON	EF 85mm f/1.8 USM	75.0	71.5	315877	425	£275
CANON	EF 50mm f/1.4 USM	73.8	50.5	216020	290	£250
CANON	EF 28mm f/1.8 USM	73.6	55.6	236549	310	£325
TOKINA	17mm f/3.5 AT-X PRO (CANON AF)	84.0	57.0	315881	435	£278
	SUM			2423516	2975	£2,352

Prime System 4

CANON	EF 200mm f/2.8 L USM MK2	83.2	136.2	740481	765	£525
CANON	EF 135mm f/2L USM	82.5	112.0	598709	750	£699
CANON	EF 85mm f/1.8 USM	75.0	71.5	315877	425	£275
CANON	EF 50mm f/1.4 USM	73.8	50.5	216020	290	£250
CANON	EF 24mm f/1.4L USM	83.5	77.4	423842	550	£950
TOKINA	17mm f/3.5 AT-X PRO (CANON AF)	84.0	57.0	315881	435	£278
	SUM			2610810	3215	£2,977

Prime System 5

CANON	EF 200mm f/2.8 L USM MK2	83.2	136.2	740481	765	£525
CANON	EF 135mm f2.8 with Softfocus	69.2	98.4	370081	390	£235
CANON	EF 85mm f/1.8 USM	75.0	71.5	315877	425	£275
CANON	EF 50mm f/1.4 USM	73.8	50.5	216020	290	£250
CANON	EF 24mm f/1.4L USM	83.5	77.4	423842	550	£950
TOKINA	17mm f/3.5 AT-X PRO (CANON AF)	84.0	57.0	315881	435	£278
	SUM			2382182	2855	£2,513

To summarise:

System	Volume mm^3	Mass g	Dec 04 Price
Zoom System 1	2385125	3020	£3,299
Zoom System 2	2577785	2860	£2,828
Zoom System 3	2015733	2255	£2,528
Zoom System 4	1981781	2155	£2,058
Prime System 1	2610810	3215	£3,126
Prime System 2	2338602	2875	£2,246
Prime System 3	2423516	2975	£2,352
Prime System 4	2610810	3215	£2,977
Prime System 5	2382182	2855	£2,513

Summarising relative to the maximum volume or mass or cost and highlighting the minima and maxima:

System	Volume	Mass	Dec 04 Price
Zoom System 1	91%	94%	100%
Zoom System 2	99%	89%	86%
Zoom System 3	77%	70%	77%
Zoom System 4	76%	67%	62%
Prime System 1	100%	100%	95%
Prime System 2	90%	89%	68%
Prime System 3	93%	93%	71%
Prime System 4	100%	100%	90%
Prime System 5	91%	89%	76%

So unsurprisingly, we can make the most savings in volume, mass and cost if we maximally compromise quality.

Clearly there are many combinations including hybrid zoom and prime combinations.

Light Gathering

Back to TOC

So in this section I want to take a look at maximum aperture as this is normally thought of as a major advantage for prime lenses. As we have seen, there are some quite fast zooms these days.

A major advantage of wide aperture lenses is not only their use in low light, but arguably more importantly the ability to limit depth of field to isolate subjects.

We can see from looking at the equations that hyperfocal distance is inversely proportional to the f-stop.

In the following section the relative lens speed and depth of field tabulated against f-stop for each focal length of interest are presented. Two depth of field measures are provided. The first the near point of acceptable sharpness when the lens is focused at the hyperfocal distance. The second is the size of the zone of acceptably sharp focus when the lens focus is set to 2M, 5M or 10M depending on the focal length, all worked out for an APS-C sensor.

There is only one image stabilised lens in this study, this is because the IS feature tends to be provided with very long primes or slow zooms. The IS feature is very valuable for handholding or photographing from moving platforms, such as a car or boat. It allows the photography to use two or more stops slower shutter speed and remain handheld. However, this does not help freeze subject motion or limit depth of field. So for this part of the study it is considered of “no consequence”.

For zooms for variable speed the speed relevant at the focal length of interest is appended in square brackets.

Another point to note is that a faster lens will often reach an acceptable level of sharpness before a slower lens as it is being stopped down [ref]. This may mean the faster lens providing superior resolution compared to a slower lens, particularly if the speed difference is 2 or more stops.

17mm

In this range we have available the following lenses:

Make	Lens	Stops from Fastest Lens
CANON	EF 20mm f/2.8 USM	0.0
CANON	EF-S 18-55mm f/3.5-5.6 [3.5]	0.5
TOKINA	17mm f/3.5 AT-X PRO (CANON AF) [1]	0.5
CANON	EF 17-40mm f/4L USM	1.0
CANON	EF 16 35 mm f/2.8 L USM	0.0

f-stop	2.8	3.5	4
Near Point for Set Focus H	2.75	2.20	1.93
DOF Set Focus 2M	1.67	2.29	2.84

One does not normally use a wide angle lens and try to limit depth of field, however it is important to understand how much control one may be loosing with a slower lens. There is a reduction of about 40% in DoF between f4 and f2.8.

There is not a large range of speeds with lenses at this end of the scale, and speed is less of an issue in respect of handholding.

24mm

Make	Lens	Stops from Fastest Lens
CANON	EF 24 70 mm f/2.8 L USM	2.0
CANON	EF 24mm f/1.4L USM	0.0
CANON	EF 24mm f/2.8	2.0
CANON	EF-S 18-55mm f/3.5-5.6 [4.0]	3.0
CANON	EF 17-40mm f/4L USM	3.0
CANON	EF 16 35 mm f/2.8 L USM	2.0

f-stop	1.4	2.8	4
Near Point for Set Focus H	10.97	5.49	3.84
DOF Set Focus 2M	0.37	0.75	1.12

In this range there is a 3.0 stop speed range, this equates to a factor of 3 for depth of field. This focal length is getting close to a normal lens for crop factor 1.6 so both speed and depth of field control is a significant issue.

28mm

Make	Lens	Stops from Fastest Lens
CANON	EF 28mm f/1.8 USM	0.0
CANON	EF 28mm f/2.8	1.3
CANON	EF 24 70 mm f/2.8 L USM	1.3
CANON	EF-S 18-55mm f/3.5-5.6 [4.5]	2.6
CANON	EF 17-40mm f/4L USM	2.3
CANON	EF 16 35 mm f/2.8 L USM	1.3

f-stop	1.8	2.8	4	4.5
Near Point for Set Focus H	11.61	7.47	5.23	4.65
DOF Set Focus 2M	0.35	0.55	0.79	0.90

A 2.6 stop range and a factor of 2.2 for depth of field, excluding the kit lens. This focal length is fairly close to a normal lens for crop factor 1.6 so both speed and depth of field control is a significant issue.

35mm

Make	Lens	Stops from Fastest Lens
CANON	EF 35mm f/2	1.0
CANON	EF 35mm f/1.4L USM	0.0
CANON	EF 24 70 mm f/2.8 L USM	2.0
CANON	EF-S 18-55mm f/3.5-5.6 [4.5]	3.3
CANON	EF 17-40mm f/4L USM	3.0
CANON	EF 16 35 mm f/2.8 L USM	2.0

f-stop	1.4	2	2.8	4	4.5
Near Point for Set Focus H	23.33	16.33	11.67	8.17	7.26
DOF Set Focus 2M	0.17	0.25	0.35	0.50	0.56

A 3.0 stop range and depth of field factor of 2.9 excluding the kit lens. This focal length is exactly the normal lens for crop factor 1.6, so both speed and depth of field control is a significant issue.

50mm

Make	Lens	Stops from Fastest Lens
CANON	EF 50mm f/1.8 II	0.7
CANON	EF 50mm f/1.4 USM	0.0
CANON	EF 24 70 mm f/2.8 L USM	2.0
CANON	EF-S 18-55mm f/3.5-5.6 [5.6]	4.0

f-stop	1.4	1.8	2.8	5.6
Near Point for Set Focus H	47.62	37.04	23.81	11.90
DOF Set Focus 2M	0.08	0.11	0.17	0.34

Only a 1.5 stop range and depth of field factor of 2.1 excluding the kit lens. This is the “ideal” portrait focal length for crop factor 1.6, so this is critical. The kit lens is particularly poor speed wise at this focal length.

85mm

Make	Lens	Stops from Fastest Lens
CANON	EF 85mm f/1.8 USM	0.0
CANON	EF 70 200 mm f/2.8 L USM	1.3
CANON	EF 70 200 mm f/2.8 L USM IMAGE STABILISER	1.3
CANON	EF 70 200 mm f/4 L USM	2.3
CANON	EF 24 70 mm f/2.8 L USM	1.3

f-stop	1.8	2.8	4
Near Point for Set Focus H	107.04	68.81	48.17
DOF Set Focus 5M	0.23	0.36	0.52

A 2.3 stop speed range and a depth of field factor of 2.2. This is a good portrait focal length for crop factor 1.6, so this is critical.

135mm

Make	Lens	Stops from Fastest Lens
CANON	EF 135mm f/2L USM	0.0
CANON	EF 135mm f2.8 with Softfocus	1.0
CANON	EF 70 200 mm f/2.8 L USM	1.0
CANON	EF 70 200 mm f/2.8 L USM IMAGE STABILISER	1.0
CANON	EF 70 200 mm f/4 L USM	2.0

f-stop	2	2.8	4
Near Point for Set Focus H	486.00	347.14	243.00
DOF Set Focus 10M	0.41	0.58	0.82

A 2.0 stop speed range and 2 depth of field factor. Depth of field is becoming limited anyway at these focal lengths so this is less of an issue.

200mm

Make	Lens	Stops from Fastest Lens
CANON	EF 200mm f/2.8 L USM MK2	0.0
CANON	EF 70 200 mm f/2.8 L USM	0.0
CANON	EF 70 200 mm f/2.8 L USM IMAGE STABILISER	0.0 (- 2.0 for IS)
CANON	EF 70 200 mm f/4 L USM	1.0

f-stop	2.8	4
Near Point for Set Focus H	380.95	266.67
DOF Set Focus 2.0M	0.26	0.38

A 1 stop speed range and 1.4 depth of field factor. Depth of field is becoming limited anyway at these focal lengths so this is less of an issue

Summary

The most significant trade-offs are in the normal to moderate telephoto range, 24 to 85mm.

Back to TOC

Applications

Back to TOC

Focal Length Group	Type	Applications	Requirements
10mm	Ultra Wide	Landscape, Strong Perspective Abstracts, Strong Perspective Nature, Wide View Architecture	Excellent Detail, minimal flair and distortion. Speed not an issue.
17mm, 20mm	Wide	Landscape, Moderate Perspective Abstracts and Nature, Architecture, People	Excellent Detail, minimal flair and distortion. Speed a secondary issue.
24, 28, 35mm	Normal	Landscape, Still Life, People, General	Good Detail and contrast well controlled flair, speed very significant.
50mm	Short Telephoto	Landscape, Still Life, Portrait, General	Good Detail and contrast, well controlled flair, speed highly significant
85mm, 135mm	Medium Telephoto	Landscape, Still Life, Portrait, General, Nature, Architecture Detail	Good Detail and high contrast, speed highly significant
200mm	Long Telephoto	Landscape, Still Life, Portrait, General, Nature, Architecture Detail	Good Detail and high contrast, speed significant

There is not a tremendous difference in the angle of view from 20mm to 17mm as can be seen from this example framing.

Looking at the way the focal lengths line up based on the core lenses of the 100mm f2.8 Macro and 50mm f1.4:

This suggests that the 28mm lens may be a better choice for an intermediate focal length between 20 and 50mm than 35mm

Back to TOC

EF-S Sanity Check

The aim of this section is to review the stopped down (f/8) performance of the available EF-S lenses not studied against some of the lenses already reviewed.

At the time of writting the first prime EF-S lens has just been released. If a wide range of fast EF-S lenses are produced in the future they may be worth investigating.

EF-S 17-85mm f4-5.6 IS USM Wide

This lens’s contrast is much worse that the kit lens at the wide end except for beyond about 11.5mm from optical axis. There is less separation with detail, although this is much worse than the kit lens and probably worse than the disappointing 16-35mm zoom for detail.

EF-S 17-85mm f4-5.6 IS USM at 60mm and EF-S 60mm f2.8 Macro USM

The 17-85mm is about the same for contrast and detail as the 17-55mm kit lens, i.e. not that good at the edge, both probably the worst lenses out of the group.

The 60mm Macro performs better, keeping up quite well for contrast with the prime lenses out to about 10mm off the optical axis. Detail for this lens is very good at the centre but degrades fast being worse than the 100mm f2.8 macro beyond about 7.5mm from the centre of the optical axis.

Another possible use of the 60mm Macro is as a portrait lens so the next few plots look at the wide open performance of the lens including the MTF influence on Bokeh.

We can see from this that the indications for Bokeh are very good and the wide open contrast and detail performance is better than all the other lenses. However, it must be noted than the f2.8 aperture is much smaller than the 50mm primes. However it is beating the more comparable 100mm f2.8 Macro.

EF-S 17-85mm f4-5.6 IS USM Tele

The performance is better at the tele end keeping up with the prime lenses for contrast to 6 or 7mm off optical axis after which it drops off quickly being the worse performer by far beyond 9mm from the optical axis.

A similar story is seen for detail with the only exception that the detail contrast below 4mm offset from the optical axis is the best out of the lenses, however it degrades rapidly and is the worst lens by a long way by 9mm offset from the optical axis.

EF-S Sanity Check Conclusions

The EF-S 17-85mm f4-5.6 IS USM is disappointing and typical of a wide range zoom, however it is relatively light and compact if performance is not a major issue.

The EF-S 60mm f2.8 Macro USM is better and should be very sharp and high contrast in the centre of the frame although poor at the edge, personally I think the 100mm f2.8 is a better solution in some ways. The EF-S 60mm could make quite a good portrait lens with a reasonable maximum aperture and excellent wide open detail and contrast and good bokeh.

Back to TOC

On Zooms

Zooms are ideal for walking about general photography. It has to be said that the 18-55mm kit zoom probably fills this role well despite its shortcomings.

A major advantage of zooms with a DSLR, is reducing the number of lens changes and thus the chances of dust contamination. Looking at the available zoom ranges (excluding the kit lens):

Available Zoom Sizes	Effective Focal Length on 1.6X Sensor in 35mm full frame terms
70-200 mm	112-320 mm
24-105mm	38mm-168mm
24-70 mm	38-112 mm
17-40 mm	27-64 mm
16-35 mm	26-56 mm
10-22 mm	16-35 mm

I know in general photography on full frame 35mm I tend to want to swap between 28 and 85mm the most. I can see here that the zooms available would not reduce the number of the changes greatly. Below 70mm each zoom is only replacing about two to two-and-a-half prime lenses, so I will still be changing lenses quite a lot.

One tempting zoom is the 17-40mm f4L. This seems to be a very sharp lens with a good reputation which could cover the requirements of the 20mm and 35mm lenses for a similar total weight, cost and slightly less volume. One issue with wide zooms is barrel distortion.

For 1.6X crop factor cameras the excellent EF-S 10-22mm provides a small aperture good quality ultra wide to normal zoom. A good companion to this would be either the 24-70 or 24-105mm zooms providing a two zoom system covering the most used focal lengths. When augmented with the 18-55mm kit lens then overlap is provided around the normal range.

Looking to the future a 24-70 or 24-105 will provide an ideal walk around lens for full frame and the 17-40 will replace the 10-22.

Other reasons for zooms:

i)        Minimize resolution loss due to cropping.

This is not really such an issue due to the improvements in sensor resolution. In most cases it just requires the photographer to walk a little anyway.

ii)      Optimize Framing

Again not really - see above. In fact when I use a zoom I find the zoom is a bit of a distraction to working out the composition and experimenting with different views and perspectives.

iii)     Reduce Weight

This is just not the case. Zoom systems of inferior performance and speed tend to weigh just as much or more as a prime system, they probably cost the same also. Also zooms tend to be heavy lenses as can be seen from some of the comparisons here.

iv)     Allow you to deal with rapid changes of focal length.

This is true, particularly if you are in a photo-journalistic activity. You can get around this to quite an extent by cropping but this is not ideal. Of course a zoom only helps here if you don't have to go beyond the limits of it's focal length. There are some very wide range zooms with advanced optical designs, they are not very sharp though. This does not describe my sort of photography though. (But I do have the little kit zoom; you see this has all been planned!)

Conclusions

Back to TOC

10mm

The only really viable option is the 10-22mm zoom.

17mm/20mm

There is an abundance of riches in some ways.

The 16-35mm zoom is fast but limited in sharpness, however this problem is most significant for 35mm full frame use rather than APS-C format. Another issue with this lens is it may be the limiting factor for sharpness instead of the sensor when comparing the Canon simulated MTF data. Note that some more modern zoom designs may exhibit lower chromatic aberration for full frame DSLRs, the 16-35mm may be an example of this (Full frame with digital and Chromatic Aberration).

Even stopped down the 16-35 has less centre detail than the 17-40 and wide open on full frame the performance is so poor as to be not worth the trouble [1]. An interesting exerisise though was printing the sample prints in this review as if from the 5D or truncated to APS-C. Although to me at A3+ and even A4 the 17-40 @ 17mm was sharper than the 16-35 @ 16mm both f5.6 my wife and son had trouble in telling the difference and in fact picked the 16mm.

The results of [1] for the 5D with 16mm f2.8 at f2.8 and f5.6 were scalled to 5606 pixels width to set the pixel resolution equivalent to a 20D with a full frame sensor using bicubic smoother. Then, from the results of SLRGear, the f5.6 image was given 2, 3 and 4 Blur More operations to simulate the measured effect at the edge of the APS-C sensor of the 20D. All frames were then scaled to give a APS-C sensor sized view at A4 using Bicubic Sharper, given standard output sharpening of dark and light highpass filter layers with a r=1.1 pixels and printed on Matt paper.

The SLRGear results suggest only a 2 DxO BxU variation between the f5.6 corners and f2.8 corners on a 20D. However even the four blur more example was not as bad as the f2.8. This suggests that the example in [1] was a bad copy and the SLRGear a good copy, or possibly the effect of BxU are not calibrated correctly. This is not encouraging as the performance of the SLRGear lens was thought to be poor.

The 17-40mm zoom is sharper (the lens image simulations for this lens are just detectably sharper than the 20mm and 35mm primes at f/8) but of average speed at this focal length.

Both prime lenses are said to have lower spherical aberration levels by a factor of 3 to 4 than the zooms [1] [2] [3] [4] although there is some disagreement [2], this is particularly important for architectural use. Additionally the zooms have no depth of field markings.

The Tokina 17mm may be slightly sharper than the Canon 20mm, but the latter has USM and FTMF.

A 20mm lens angle of view is only slightly narrower than 17mm, and there is the issue of future compatibility that is more problematic with third party products.

It comes down I think to the 17-40mm zoom or the 20mm prime, it is a close thing. For me the advantages of the prime in faster aperture, depth of field markings and weight swing things. The aperture of the zoom at other ranges will be limiting but not excessively so at 20mm.

Also with the recent release of the 5D and the prospect of affordable full frame DSLRs sooner than expected the 17-40's limited aperture is less of an issue in the longer term. One is thus tempted to forgo the 1 stop speed and other small advantages of the zoom for an increase in cost and weight to gain the flexibility of the zoom. A fast aperture lens in the 24-35mm range can still be provisioned.

This lens test is about the most comprehensive and objective lens test I can find and directly compares the 17-40mm f4L with the 20mm f2.8. In this test the 20mm comes in slightly sharper, with lower spherical and chromatic aberration and less vignetting. The 20mm seems to be able to provide reasonable corner full frame sharpness, at least better than the 16-35mm.

It is worth noting there is significant part to part variation, this is most obvious below 35mm focal lengths where the lens design difficulties highlight any poor variation. So for example the 17-40 and 20mm should on average be sharper than the 16-35mm there will be some individuals that have experienced parts with the opposite ordering.

The EF-S 17-55mm f2.8 IS seems to have good MTF performance compared with other lenses in this class. It is quite heavy and expensive for an EF-S lens. However it may be a good choice for those that do not plan to move from crop sensors to full frame.

24mm/28mm/35mm

These are candidate focal length for normal lenses for a 1.6X crop factor camera.

The sharpest lenses are the 35mm primes.

The 28mm primes and the 24mm primes in between in terms of sharpness.

For the primes the slowest lenses in any given focal length are the sharpest at f/8. However, the faster lenses have USM and FTMF, this is a significant compromise for the 28mm f1.8.

The clearly superior lens is the 35mm f1.4L, however this is a weighty and expensive lens.

Of the zooms in this range the 24-70mm is the sharpest, but even so is mostly outclassed by the primes. Additionally it is a very heavy lens. It was worth noting the review [SLRGear] showed that although the APS-C performance at 24mm was very good there was significant degradation in the corners and edges on full-frame wide open.

If you must have a zoom go for the 24-70. If you have the budget and weight is not a problem, go for the 35mm f1.4L. Otherwise consider the 28mm f1.8. If lens speed and autofocus performance is not a major issue go for the 35mm f2. I am inclined to save for the 35mm f1.4L, but it is heavy so the 28mm f1.8 may be better.

A late assessed lens is the 17-40mm f4L. This is a very sharp lens with a good reputation.

If the 50mm and 20mm focal lengths are taken as given then the 28mm lens would be the best intermediate fit.

50mm

This is a no brainer, the 50mm f1.4 is blisteringly sharp, has USM and FTMF and is small light and reasonably priced.

The 50mm f1.8 is good optically but with no distance scale limited. A good value budget lens if build quality is not important.

Of the zooms in this range the 24-70mm is the sharpest again.

The EF-S 60mm f2.8 Macro performs well but not as well as the 50mm primes or the 100mm f2.8 Macro and is good at maximum aperture.

85mm

The 85mm f1.8 is an exceptionally good lens. There is a faster L class lens (the f1.2L), however it is very heavy and expensive so it has been neglected in this study as the f1.8 ticks all the boxes.

Both the 24-70mm zoom and the three 70-200mm zooms are good, but come no where near this lens (or the 100mm f2.8 Macro I already have).

All the zooms are heavy, extremely heavy in the case of the 70-200mm f2.8 pair. I know that these would be exhausting to use without a tripod and the IS version would be needed if only because my arm would start shaking early on.

135mm

Of the two primes the 135mm f2 is much better than the f2.8 softfocus in resolution terms, but at a price and weight.

The 135mm f2 is on a par weight wise with the 70-200mm f4, if you need a long zoom then that is the one to go for.

However, in terms of optical quality the 135mm f2 is well ahead of the zooms particularly in edge detail.

200mm

Here we have the one lone prime against the four zooms. In terms of image quality there is not much between them. In terms of weight the prime is similar to the f4 zoom although this f4 zoom is the less sharp of the group, the IS version is almost the same in terms of sharpness as the non-IS except for f8 30 lp/mm where the non-IS has superior microcontrast across the frame.

For me, the prime has it due to the speed weight trade. However, one of the f4 zoom would be an excellent alternative to the 135mm f2 and 200mm f2.8 if maximum kit weight is a major issue.

A point to remember is the f/2.8 lenses can be used with a 2X teleconverter and still retain autofocus on the 20D (the limit is f/5.6). The f/4 lens can only use the 1.4X teleconverter and retain autofocus. The mass of the teleconverters is 265g and 220g respectively further adding to the mass and length of the lens. However, in probability use will entail a tripod on these focal lengths.

Back to TOC

Final Comments

It seems that the primes are there with the performance. However, the smaller aperture zooms come in well if all up weight needs to be kept low.

The wide aperture zooms are good image wise (with the possible exception of the 16-35mm, however this is not too bad for APS-C sensors) but are very heavy to deal with on camera. If you have the upper body strength this may not be an issue for you.

There is an argument that the 17-40 provides improvement in sharpness over the kit lens only at the normal to long lengths but no speed advantage and so a better selection of zooms might be the 10-22 with either the 24-70 or 24-105mm zooms. These could be augmented with fast primes.

I feel that maximum apertures are well worth the compromises for the creative control the allow.

If budget is unlimited then a dual zoom and prime system gives the best of both world, just choose what you need before you leave the house (in my dreams).

In practice as I have the 100mm f2.8 Macro, this for the moment double for the 85mm and 135mm, allowing me to concentrate on populating the extremes of focal length and perhaps the 50mm and 20mm.

Back to TOC

Last Updated 05/06/2008

All material is supplied as is and without warranty, use at your own risk.

All opinions stated are the authors own.

All quoted information remains the copyright of the respective authors.

Any trade marks mentioned are the property of their owners.

Contact:

Note this email gets spammed a lot so if you want to get your message through please prefix "Holding Tank" in the title

holdingtank@ware.myzen.co.uk