General guide for making a lens review/comparision

[PekkaM]

Could somebody list the important things to include in comprehensive lens comparision. I bought me a Canon 17-40L and would like to create an online comparision of it and my old Sigma 17-35.

So far I know to:
- use a tripod (I have 300D so no mirror lockup in use)
- focus on larger apertures
- have 100% crops of edges and central areas
- use manual settings on A, T and WB
- test shots with sun hitting the front lens

What about raw-conversion, should all sharpening etc be set to zero?
What is the easiest way to create worst possible CA?
What kind of scene to use (obviously lots of detail), what about lighting, cnotrast etc?
Focusing is not very good on Sigma with low light condition, should I shoot landscape-like scene with focus at infinity?

[Olegis]

This page (http://www.normankoren.com/Tutorials/MTF5.html) has a lot of links to lens testing resources - guides, test charts, testing techniques etc. Hope it helps ...

[PekkaM]

Thanks, althuogh that probably goes a bit too far. I was going to do something where you can judge the quality by your eye. I can't even read the MTF-curves myself... :lol:

[ron chappel]

Yes,judging by the images themselves is allways very usefull.In fact i much prefer it to any resolution or MTF tests!


Some of my suggestions:

Allways take two to three pics in each position instead of just one and choose the best from those,this elliminates focus errors and greatly cuts down on the chances of a dud result from a range of other factors

For the flare tests,do some with the sun right in the frame and some with the sun just in/just outside of the frame.The latter is usually by far the worst for aperture shaped highlight flare and the former is usually worst for contrast reducing flare.
In fact you can handhold for these tests- move the lens around until the flare is greatest,it will be in a slightly different position for every lens

"What about raw-conversion, should all sharpening etc be set to zero?"

I've seen tests done at zero sharpening and also done alot myself with sharpening set at 1.It doesn't seem to make a whole lot of difference when judging one lens directly against another
On a related note,i've done many tests to see how much sharpening lenses can cope with.Good lenses can be sharpened alot while allmost any pictures from a crap lens will have very little room for improvement.It's one of the tests i do on a lens if i'm seriously considering keeping it

"What is the easiest way to create worst possible CA?"

I have found that shooting across water towards the sun (lots of small wave reflections with sun out of frame) will allways bring up huge amounts of fringing :shock: :lol:
It might be tricky with a wide lens to keep the sun out of frame but that's just the general idea-anything with lots of highlights should do the trick



"What kind of scene to use (obviously lots of detail), what about lighting, contrast etc?
Focusing is not very good on Sigma with low light condition, should I shoot landscape-like scene with focus at infinity?"

There are two types of general scenes-those such as landcapes plus "equal magnification" tests
For landscapes you probably should focus on the hyperfocal point to get as much in focus as possible.Personally i would avoid that though and just do seperate tests at the most likely focus points.Points where you KNOW you have properly focussed on them.

The "equal mag" test i mentioned is where you take a pic of a subject (usually close'ish,certainly not infinity) then change the zoom on the lens and move back/forward to get the same framing.Why do a test like that?So you can see *exactly* how much better or worse one end of the zoom range is from another.
Honestly though,this test may not be very important for wide lenses-it's very important for telezooms for seeing how much worse a lens gets at it's long end for example.

[PhotosGuy]

- use a tripod (I have 300D so no mirror lockup in use)

I'd add to that, 'Use the self timer or a remote release to minimize vibrations, even at fast shutter speeds'. :wink: