lens tests 3

Lens test Canon 1D Mark III vs Olympus E510

see also:

My conclusions:

both these cameras are 10mp so images should be reasonably comparable in resolution terms.
OK, I just spent a LOT of money on a Canon 1D Mark III with 85mm f/1.8, 135mm f/2.0L (which is said to be one of their sharpest), the 90mm tilt-shift and the 24-105mm f/4 IS and thought I should get an idea of how they perform at different apertures so I can use them at the best, and at the same time I pitted them against my Olympus kit.
my prime reason for buying these was to use them mainly at their widest apertures, hence this is what I mainly tested.
boy, was I in for a shock!
even the Olympus kit lenses were significantly sharper at their widest apertures compared to the Canon ones at their widest apertures - think I have been conned here.
have a look for yourself - I have now re-tested these with ISO 100, mirror lockup, and self-timer on 10sec, neither of these options made a big improvement over ISO 400, no mirror lockup and self-timer of 2secs BUT what does seem to be the cause of the poor corner performance at least is that even at a subject distance of 3m, the Canon lenses do not have sufficient field flatness to keep and object 0.5m from the centre in focus while keeping the centre in focus.
Even taking this into account the Olympus lenses look sharper in both centre and corner charts wide open - not bad for kit lenses and a much smaller 10mp sensor which demands higher lens resolution.
the EF 135mm f/2.0L lens had a severe AF back focus problem which was resolved by using manual focus - I had to adjust that on the Canon 1D mark III and set backwards focus to 3 to correct the error - not sure what other photographers do when their camera can't make these micro-adjustments - guess they send it back to Canon to re-calibrate?
furthermore, the Canon's WB tungsten setting didn't seem to adequately correct the lighting and the exposure was variable, especially the TS-E lens which was over-exposed - hence I need to repeat those.
and then there is:
- seems that the AF-servo mode on Canon has a few issues so probably won't be using it for that
- the 49 AF points on the Canon are no better than having 1 point for the vast majority of my work
well, at least I can console myself that with the Canon kit:
- lower noise at higher ISO for astrophotography work
- easier to blur the background (although bokeh itself is better on Olympus, even on the kit lenses with their circular apertures)
- get to use the tilt-shift lens
- don't need to go to the gym anymore, just carry this kit around.
I am so glad I decided to stick with the Olympus kit for my ultra-wide angle and macro work, if this is anything to go by - my next lens looks like it will be an Olympus ZD - going to take a bit to convince me otherwise, and I will use my Olympus OM wide angles on the Canon.
I'm also glad that I bought the 1DIII over any other Canon or Nikon because without the Live Preview I would not be able to obtain critical focus and work out why the images were soft and secondly, it allows adjustment of the back focus problem.

My rough lens testing setup:

print 2 copies of test target standard resolution charts printed at 300dpi on A4 sheets and pinned to black cardboard sheet
camera on tripod with lens shaded and self-timer set to 2secs, centre-spot AF (manual lenses were manually focussed with 10x live preview) on centre target.
Exposure set to aperture priority, WB to tungsten at ISO100 (ISO 400 on Canon 1DMarkIII). No exposure adjustment.
Image stabiliser on the E510 turned off as was it turned off on the EF 24-105mm IS lens.
ideally for direct comparisons, the magnification and exposure histograms should be practically identical - I have not achieved either of these precisely below but I think they still give you a ballpark idea of how they compare.
image quality set at SHQ jpg.
no post-processing or resizing but images cropped and converted to jpg with 54% compression.
I have chosen this target as it would be representative at how well the lens would perform for shooting a couple portrait.
camera to target distance adjusted to give same magnification settings to give approx. this image size:

full target image

Examples of lens tests using the full target image distance:

NB. the poor corner performance of the Canon lenses may be due to the auto-exposure on the Canon over-exposing the charts a touch compared to the Olympus and it seems that even mild over-exposure has a dramatic effect on recorded resolution. I have thus re-tested using manual exposure at ISO 100, self-timer 10secs & mirror lock up and results are better but still not matching the Olympus lenses.

NB. you may need to save them to disk then view at 200% to analyse them more closely.

Don't think I will be adding any more Canon lenses to my kit for a while until I am convinced otherwise. Fortunately, apart from the 24-105mm lens, the other lenses were purchased for a specific purpose which does not rely on corner sharpness - just a pity the centre sharpness is not up to a standard I expected in such big, heavy & expensive lenses.

the charts are done as centre/corner pairs:

Lens	centre	corner
Canon EF 24-105mm f/4 IS lens at 85mm with IS turned off. f/4:	initial test above at ISO 400, no mirror lock up, self timer 2sec, AF on centre re-test at ISO 100, mirror lock up, self timer 10sec, AF on centre, exposure less to reduce any flare issues same as above but MF using live preview 10x on the corner chart - my conclusion is the lack of field flatness, although the degree of loss of focus seems more than I would expect at f/4, 88mm	this corner is VERY soft hmm... not much difference same as above but MF using live preview 10x on the corner chart, now that's a bit better, but now the center chart on the left is soft
Canon EF 85mm f/1.8 lens f/1.8:	MF on the corner chart now leaves the centre chart soft showing that the focal plane is not flat at 3m subject distance	MF on the corner chart shows it can get reasonable resolution there
Canon EF 85mm f/1.8 lens f/4:	not bad	still a bit soft at f/4
Canon EF 135mm f/2.0L lens f/2.0:	surely this can't be that bad! I'm going to have to re-check the focus on this one. re-tested and still awful with AF ahhh... now this is better - the difference? This was MF on the center using Live Preview 10x so the problem with this lens is back focus calibration problem - now to learn how to set an adjustment for that! MF on the corner chart didn't make much difference on the center chart AFTER AF micro-adjustment with setting put to 3 on backwards focus correction.	corner much the same as center ahhh.. even the corners look OK MF on the corner chart didn't make much difference, suggesting this lens indeed is not bad at all and has relatively flat focal plane - might be great for astrophotography wide open after all! now I am a happy boy, although its a touch soft in the corner but I can live with that.
Canon EF 135mm f/2.0L lens f/4.0:	AF above shows AF back focus error much nicer with MF even though MF is on corner chart	just terrible, surely Canon could have their systems calibrated a little better than this? what a difference correct focus makes
Canon TS-E 90mm f/2.8 tilt-shift lens all settings on zero. f/2.8:	OK, it's not bad in the center on my original test but nothing to write home about re-test with mirror lock up, self-timer 10sec, ISO 100 and less exposure - not much difference really - EXCEPT, this one was MF on the corner chart	what's going on here? now we are talking, much better result - can't explain why the initial test was so poor - maybe the lens was still in a slight amount of tilt?
Canon TS-E 90mm f/2.8 tilt-shift lens f/4:
Olympus Zuiko Digital 14-42mm kit lens: wide open at 42mm = 84mm effective
Olympus Zuiko Digital 40-150mm ED kit lens at 50mm = 100mm effective f/4.2:	a bit soft
Olympus Zuiko Digital 40-150mm ED kit lens at 50mm: f/5.6:	much nicer - incredible result for such a zoom lens only 1 stop down from wide open!
Olympus Zuiko Digital 50mm macro lens: f/2.0:		a bit soft wide open in the corner
Olympus Zuiko Digital 50mm macro lens: f/4:

For some comparison:

Lens	centre	corner
Olympus C8080 at 140mm equiv. and f/3.5, 50 ISO this was taken at higher magnification, but even taking this into account, this is one sharp lens! Not to mention its appears to have minimal flare or fringing even wide open. This easily beats the Canon below, even though the Canon is at f/8. see my tests with the Olympus C8080
Olympus C8080 at 140mm equiv. and f/3.5, 50 ISO with the TCON14D 1.4x teleconverter making it 196mm equiv. Fantastic - almost no loss of resolution!
Canon 350D with Canon EF-S 17-85mm f/4-5.6 IS digital lens at 60mm and f/8 Note that this image was taken at higher magnification and thus one needs to allow for this when comparing. Taking this into account, it does not appear to be more sharp than any of the above OM mount lenses at comparable f/8.