Canon EF-S 10-22 sharpness and underexposure

Hi,

First time post here...

 

I've had the Canon 10-22 a week now (on the 40D) and tend to notice two things:

1. Sharpness...or the lack off...especially if you get away from the center at

wider apertures but even not very good in the center at smaller apertures.

Everything speaking at 10-12mm. Once you get into the 20-22mm the center is

almost pefect but still some softness in coners.

And that is at 100% view. For normal 5x7, 8x10 - everything looks good.

Is this a concern?

 

2. It seems to underexposure. Not sure because with other two lenses (Sigma 30

f1.4 and 70-200 f4 IS) - exposures are usually very good. But with the 10-22

especially on the wide end - I find in Aperure Priority to have to dial in +1/3

or even +2/3 to get it right. Interesting - in Manual I can get it better...

 

Any thoughts?

 

Thanks for your time

Sorin

Seeing a sample posted here will help us answer your question.

I have had excellent results with the lens. From the reviews it would seem that your experience is not typical. Maybe at 10mm the corners are softer but the rest is great for me.

I have the 10-22mm and I think it's a great lens. It's very difficult to find super wide glass that is sharp wide open. I'm thinking f/5.6 or f/8 is the sweet spot.

Post examples shot wide open and preferably also at f/5.6 or f/8 please. Tell us how the shots were taken, w/tripod, w/MLU, etc...

Most super wides are soft, in varying degrees, away from the center. If you're getting under exposed images, is it only with that lens? Have you tried others? If not, make sure your exposure compensation is set at 0.

Here are some at 10mm - first at f3.5

http://farm3.static.flickr.com/2196/2201334836_7a160376b7_o.jpg

second - same thing at f7.1

http://farm3.static.flickr.com/2359/2201334716_6025b6c10f_o.jpg

This one is at 22mm - f4.5 - anyway better

http://farm3.static.flickr.com/2010/2200540303_3e4fa01404_o.jpg

Another center crop (just aprox - not 100%) at 10mm f16 - should be sharp?

http://farm3.static.flickr.com/2320/2202394062_516268bd30_o.jpg

 

Thanks for your input.

Sorin

Sorin

 

those look pretty bad. there's a ton of CA in the upper right corner. all corners are very soft. upper right is softer than other corners.

 

this is not typical of the 10-22 (or any lens that costs more than $400).

 

you have a bad copy. replace

To Joshua....no, they were all handheld. I will do some tests as you suggested as well (I did actually those with my 70-200 f4 IS)

Thanks

Hi Alan,

That's exactly what I thought...Thanks for the input

In fairness to the lens, this is not a good test. The wall at bottom left is very close, and everything else is much further away. Did you use a tripod? Did you focus manually or autofocus? What did you focus on? Also, the final shot is at f/16, so diffraction will reduce sharpness.

 

However, I do find that I need dial in about 1/3 - 2/3 stop more exposure compensation with my 10-22 that with my other lenses. It doesn't bother me because I always use exposure compensation to ensure that my histogram is as far to the right as possible. But I do notice the difference. I've used two copies of this lens, and both required more exposure compensation than my other lenses.

Ultra wide angles always tend to have some corner softness wide open, that is part and parcel of extreme retro focus designs. So this is normal for this lens.

 

If you stop the lens down the corners will sharpen up quickly and peak around f8 to f11.

To Gerry - no, those were quick handheld tests...during the lunchbreak at work.

However good that I know you also have to overexpose a bit.

 

To Lester - I knew about "some" corner softness...didn't expext THAT much though.

I'll do some more controlled tests tonight or tomorrow.

Thanks

Try focusing manualy it could be something with your focusing points.

I agree with Gerry that the sample images are not made in the best conditions for corner sharpness.

 

Use a tripod on a sunny day with all image content equidistant from the lens. Then decide if corner sharpness is a real problem.

I've had my 10-22 for a couple years and am still happy - the only lens I use. <P>

 

<a href= "http://pages.sbcglobal.net/b-evans/Images25/2008-1-12/">Here's some

pix</a> from last Saturday...

Sorin

 

i'm changing my mind. did not realize the test shots were hand held. at 10mm, hand held, the corners can turn to mush.

 

i vote for another test. use tripod, mlu, timer release (10 second delay, not 2 second), lens hood, and use flat subject. let's take another look

 

the 10-22 is very quirky under bad conditions (as are all ultra-wide zooms)

The first two pictures are taken very close to the brick wall. You need a smaller aperture than f3.5 or f7.1 to get the depht of field from a few feet to infinity....even at 10mm.If you do the same test again at f11 you will get increased sharpness on things that are only a few feet away. For a ultra wide angle this lens performs very well. The more expensive 14mm2.8L prime or 16-35mm 2.8L are not better. Dont know about the newer 16-35mm2.8LII though.

There is nothing wrong with your lens. The 10-22mm does not have corner softness like you think - it has field curvature. Try this - go out and take another shot like your first ones but this time set to manual focus and put the focus on the line at the start of the infinity mark. Now look at the corners and let us know what you see.

I did some other shots, still it was freezing cold...so couldn't use tripod, however, this is a shot taken from an utility pedestal:

http://farm3.static.flickr.com/2357/2208021541_0879e8852a_o.jpg

and here some others - handheld though, AF:

http://farm3.static.flickr.com/2243/2208817736_3f6fc482f3_o.jpg

http://farm3.static.flickr.com/2309/2208022693_ea352d7fdb_o.jpg

http://farm3.static.flickr.com/2137/2208816704_8e87e107e9_o.jpg

f8: http://farm3.static.flickr.com/2414/2208819630_2f1df1484f_o.jpg

f3.5: http://farm3.static.flickr.com/2160/2208024057_cd7763bbd2_o.jpg

 

 

However, I think 22mm is so much better - in my eyes - than 10mm

http://farm3.static.flickr.com/2269/2208023355_bfc1a3702d_o.jpg

f4.5: http://farm3.static.flickr.com/2131/2208024273_baed46bb4b_o.jpg

f8: http://farm3.static.flickr.com/2120/2208820320_cf036595ef_o.jpg

in-house at f6.3

http://farm3.static.flickr.com/2237/2208814336_a684374ce0_o.jpg

 

...or the truth is that I never shot before at 10mm (or 16mm FF equivalent) and perhaps I'm expecting too much?

Or at all - do I need to check everything at 100%? Because if looking up to 50%-66% everything looks pretty good.

I'm not doing posters...yet...the largest I ever printed was 11x14...so for this to be good...what equivalent of screen size (50%, 66%, etc) would I need to check.

 

Thanks for your time

Sorin

What size is your screen? If it's around 11x14" (mine is around 10" x 12" of display space), then just fill the screen.

 

Actually, if you want to see how an 11x14 print looks, get an 11x14 print made. Video screens are different from prints. If you don't want to waste ink and paper, just print a 4x6 section of it.

Thanks Bob,

You're right about wasting ink and paper..:-)...The thing is printing 4x6 or 5x7 gives very good results...My question was if this should be the criteria of evaluating a lens - or only 100% on screen?

 

Sorin

I agree with Bob, the resolution requirements for 10X8 are fairly light, almost any correctly focused lens will look good.

 

100% crop information, or more stringently SRF/MTF testing will show up a lot of small variations in performance that are not significant or noticeable in real world situations.

 

The mistake most people make is they expect perfection from corner to corner, part of this is the psychology of the digital age - everything is expected to be perfect.

 

People looking at 100% crops lack a baseline and don't know what to expect, thus they expect too much in many cases IMHO. In real life situations DOF and not using tripod will limit IQ much more than the glass.

 

In my case I did an interesting exercise of taking 100% crops from various lens tests with known 50% SRF values and printing these at different print sizes. Then evaluating this at normal viewing distances and under a loupe, it helps give me some perspective. http://www.zen20934.zen.co.uk/photography/LensTests/Understanding.htm