canon 10-22 vs. 18-55 test

[jim c]

Here is my test of the 10-22 lens. I sent it back today.... it just doesn't seem right. Attached is the picture I took on my wall. (This has been resized for email purposes. The 2nd picture are 100% crops in different areas of this picture. Canon Rebel was on tripod, remote shutter, aperture @ 3.5. I retook the same picture twice with similar results. While the 18-55 lens was bad? on the left side, the 10-22 started soft near the middle and got progressively worse all the way across the right side. Based on all the positive comments about this lens, it seems like mine is not very good. What do you think?

[JK]

Geez... "Not very good" is an understatement!
Canon better improve their QC - a lens that costs this much should definitely perform better.

[Tom W]

Try them at the same focal length - 10 mm is almost twice as wide as 18 mm. I'd still be skeptical of this particular copy, but comparing two very dissimilar focal lengths isn't going to show the lenses on equal footing.

[Adam Hicks]

Well since you've already sent it back let's see how the next one does on the same tests. You can't expect 50mm 1.4 quality at 10mm. And you wouldn't typically use 10mm for that type of shot. I've been VERY pleased with my 10-22 when I've needed it. Try shooting the same shot with both lenses at 22mm f8 or so. That would be an interesting sharpness test to me.

Good luck with your second copy! It's a lens you'll find you use more than you planned!

adam

[jim c]

This is response to Tom... Attached is test of 10-22 @22mm 4.5 compared to 18-55@18mm 3.5...Still not clear compared to 18-55 lens. I would expect the 10-22 to be much better here since it is at the 22mm end and at 4.5 instead of 3.5 like the kit lens.

[VFTT]

You are shooting at f3.5 at 10mm and you are wondering why things are a little blurred? As Tom W said, shoot them both at 18mm at f8 and see what the results are. I'd be interested in seeing it. I've heard pretty decent things about the 10-22 and I plan on buying one.

- darren

[Hellashot]

And the 18-55 wins! This lens keeps kicking ass in quality tests vs. lenses that cost 6 times more. Maybe this lens needs a nickname, like the 50 f1.8 is the Nifty Fifty.

Anyone want to throw out a name for it?

[Mr. Messenger]

And the 18-55 wins! This lens keeps kicking ass in quality tests vs. lenses that cost 6 times more. Maybe this lens needs a nickname, like the 50 f1.8 is the Nifty Fifty.

Anyone want to throw out a name for it?


The Grad to Gran Dad?

[Adam Hicks]

I have a name for it.... still_crappy.

:)

[RockOne]

Anyone want to throw out a name for it?

Plastic fantastic ? :-) :-) :-)

[Hellashot]

I have a name for it.... still_crappy.

Would that be because you own the lens it beat?

[Adam Hicks]

I own them both. And these tests weren't exactly scientific. Anyone who prefers the 18-55 over the 10-22 is ok by me.

[jim c]

I still want the 10-22...I just want a better one. I think I have a decent 18-55...I just expect a bit more from the quality of the 10-22 based on what I paid

[JK]

Plastic fantastic ?

Damn, you beat me to it!!!

[Adam Hicks]

As well you should Jim... I like the different perspectives this lens gives, see this as a 10mm example...

http://www.fotop.net/adam12hicks/MSR_Feb_2005/Roberto_wide_on_one_car

But at 22mm I think the images are great. This one is unedited. http://www.fotop.net/adam12hicks/MSR_Feb_2005/360_side_22mm_untouched?full=1

Adam

[mbze430]

Plastic Fantastic...that's actually pretty good. but we need to be a little more specific...because there are alot of plastic lens from canon that is fantastic.

[CyberDyneSystems]

Er,.. this "test" is completely flawed..
You all can't see how this test could not possibly be used for an effective evaluation?

[kawter2]

looking at the perspective it looks like you are angled and closer to the right side of the poster

[mbze430]

it might not be scientific, but its definately "real world" test... LOL

[mr.photoguy]

I thought a Real test would consist of two lenses at the same focul lenght, and same F-stop, with multiple F stop tests. from minimum aperature to f8, or so.

[caesars0331]

I thought a Real test would consist of two lenses at the same focul lenght, and same F-stop, with multiple F stop tests. from minimum aperature to f8, or so.

Agreed. Plus, I do not see the point in knowing the relative location on the poster where each image was taken.

[mr.photoguy]

Agreed. Plus, I do not see the point in knowing the relative location on the poster where each image was taken.
I figure 10mm @ f3.5 will give you some seriously rough edges, especially with using the center focus point. Mainly due to Narrow Depth of field.
That could explain why the edges are so rough. With a higher F-stop (by my understanding) more of the image will appear to be in focus.

I would like to see a building shot with the lense, or a landscape shot.

Maybe he sent back a good lens.

[Mike H]

Er,.. this "test" is completely flawed..
You all can't see how this test could not possibly be used for an effective evaluation?

Well, some of us can. If you shoot a fairly close subject wide open the depth of field will be rather shallow. Hence, if the plane of the sensor is not parallel to the plane of the wall, the shot will be sharper on one side than the other, depending on your focus point. You can get the sharpest lenses on Earth to do this.

People test lenses by shooting test charts, which are also flat, all the time. It's a legitimate method. But they have to be very careful in setting it up and fully understand how to interpret what they are seeing before declaring a lens a stud or a dud.

It would be useful for us to know the focused distance so that we can calculate the depth of field, and hence, be able to better interpret the test. You have to evaluate your test method as carefully as you evaluate the results.

Mike H

[caesars0331]

Why not test the lenses at the same focal length, for starters.

I really dont see the point of this. There is no control and there is no clear objective of this experiment; just two different lenses shot at different settings.

[caesars0331]

I figure 10mm @ f3.5 will give you some seriously rough edges, especially with using the center focus point. Mainly due to Narrow Depth of field.
That could explain why the edges are so rough. With a higher F-stop (by my understanding) more of the image will appear to be in focus.

I would like to see a building shot with the lense, or a landscape shot.

Maybe he sent back a good lens.

Why did you quote me?? Your response doesn’t pertain to my comments.

[jim c]

Yes, I'm sure the test is "flawed", I don't have a full day to do an in depth evaluation. I only did this after I was dissatisfied with my regular shots taken with the 10-22 lens, but I couldn't put my finger on the problem until I did these shots. (By the way, I also took a shot of a stone wall and had similar results as well. I believe I got a bad lens... Other than that, having a ultra wide lens is really neat! It takes shots that you couldn't normally get....I still want one (a good one).

[jim c]

Here is a cropped shot @100%, 1/200 s, f9....Opinions?

[mr.photoguy]

Why did you quote me?? Your response doesn’t pertain to my comments.
It's in relation to what you quoted me on...

[caesars0331]

photoguy, i was in agreement with your first post. That is why i quoted it. I also added on the part that I didnt see the point of describing the location of the crops in relation to the full size image. Your "answer" to my post does not address my point, whatsoever.

[jim c]

Here is another sample (cropped 100%) thru center of picture of front yard/house. There doesn't seem to be much in focus. 1/200s, f9

[Jon]

Can you post a downsized version of the full-frame so we can see what the camera might have been focussing on? It's sort of hard to tell from a ribbon cut out of the pic.

[Dave C]

I did a test between the 10-22 and my 17-40 on the 20D and at A3 there was very little in it (set the 17-40 @17 and the 10-22 at approx 17. both f8.
Btw sweet spot for the 10-22 is probably f5.6 bearing in mind defraction etc

Slogan for 18-55
See your pictures come alive when you shoot with the 18-55:-)

[jim c]

Here is a resized picture....

[mr.photoguy]

photoguy, i was in agreement with your first post. That is why i quoted it. I also added on the part that I didnt see the point of describing the location of the crops in relation to the full size image. Your "answer" to my post does not address my point, whatsoever.
calm down bro, it wasn't an answer.. just more of an , in addition to. I wasn't discrediting, or disaggreing with you ..

[Jon]

From the look of that, the camera could have picked a focus anywhere from the plant right at the corner of the drive and road back to the roof over the entry (not the porch, but the room), which I'm guessing is about a 30' distance. So you can't be sure that the camera was actually focussing on the spot that you thought it was. Try setting up with a tripod and manually focussing, then seeing what your results are. Also, what are your in-camera processing settings? I realize you were trying to compare two lenses, but if they have slightly different performance (resolving power, inherent contrast), that could make a difference in the appearance.

[mr.photoguy]

Here is a resized picture....
Damn wide angle is awesome.

[jim c]

As far as the focusing point, the distance markers on the lens show you can focus (I don't have the lens to look at it) but I believe it focuses from .8' to 3.5' then the next mark is infinity. What I am trying to explain is after 3.5' or so, it is at infinity, which means everything should be in focus?, although I'm not sure what the DOF is at f9 on this lens.

[Jon]

No - it'll focus continuously between the closest focussing point (9.5 inches) to infinity. It doesn't just click to the marked distances. It will focus on 20 ft, or 50 ft, or 3 ft, as appropriate. But at f/8 the hyperfocal distance is just over 3 ft. depending on your desired circle of confusion. You also need to remember that the circle of confusion, and therefore hyperfocal distance, is figured for viewing of an 8x10 or so print from a normal viewing distance. If you enlarge the image beyond that, you need to use a smaller circle of confusion, which will decrease the DoF markedly. And viewing an image at 100% on a computer monitor certainly qualifies as enlarging the image. Have you tried any prints?

[kawter2]

although I'm not sure what the DOF is at f9 on this lens.


If you were focusing about 35' away (this is f8, DOF master don't got no f9), the DOF will start @ 5.08 feet and continue to infinity

Just for kicks..
Hyperfocal distance is 5.98

[jim c]

1. No, I have not tried any prints.
2. Does this mean that everything from approx.3-5' out would have the same focus.?
3. Should you look at your picture on screen at 100% ? Or should you look at it at print size?
4. I will have to read more about DOF / circle of confusion...thanks...I'd like to understand this.

[Jon]

The camera should focus on the precise distance to your target. Any point closer to, or farther away from, that will be out of focus to a greater or lesser degree. "Circle of Confusion" is the largest blur (how far out of focus) a point can have and still look sharp in the resultant print. But if you set your lens to the hyperfocal distance, everything from half that distance to infinity will appear sharp in a "normal" print. Different people use different definitions of "sharp", so use different sizes for "Circle of Confusion", so quoted numbers may vary.

Should you look at your pictures at 100% on the monitor? Depends on what you're looking for. If you're doing some precise editing, by all means. If you're expecting to make a large enlargement, yes. If your prints at your "normal" size are unsatisfactory, yes. If you want to find every possible problem with your equipment, yes. But if your final product's good enough, don't go inducing ulcers by "pixel-peeping".

[Monito]

This test is complete and utter junk. First of all, it tests apples versus oranges. Second of all, the test conditions are ridiculous, starting with atrocious lighting. Thirdly, it is not Raw unprocessed images, it is processed Jpegs with gross sharpening artifacts clearly visible. I am guessing you used in-camera Jpeg. No shutter speed indicated. Perhaps you were in the 1 sec to 1/15 second danger zone where mirror lockup is advised (though not available on DRebel).

You have (or had) a great lens, as indicated by the outdoor shot you showed. I think some people (not you, but many squeaky wheels) love complaining and putting themselves through a lot of martyr-complex returning of lenses. If testing is not done very carefully, with relentless rooting out of testing artifacts, people put themselves through great unnecessary agony over perfectly good lenses.

If you had read this page: http://www.normankoren.com/Tutorials/MTF5.html you would easily have made a much better test and not subjected yourself and ourselves to this clumsy waste of time and good money. Too much returning of lenses raises prices for all of us.

Sorry not to sugar coat it, but that's how it is.

[jim c]

Thanks for your direct reply... Maybe I shouldn't have called it a "test" in the first place. I don't claim to do or provide scientific analysis of these lenses. However, the shots I took with this lens always seemed out of focus. (I took hundreds of shots on Easter weekend also) That made up my mind to send the lens back. The so called "test" I did only confirmed the other shots I had taken (IMO). Thanks for the info on testing lenses.
I'll look into it.

[Andy_T]

Thirdly, it is not Raw unprocessed images, it is processed Jpegs with gross sharpening artifacts clearly visible.

Heehee, you will have a hard time comparing RAW unprocessed images :wink:

Best regards,
Andy

[Monito]

Heehee, you will have a hard time comparing RAW unprocessed images :wink:

Technically you are correct if you are talking about the 12 bit files directly produced by the camera. However, at the very least, they can be compared in DPP, where the only processing is the conversion. Further, they can be converted in DPP without any sharpening and posted.

[Andy_T]

However, at the very least, they can be compared in DPP, where the only processing is the conversion. Further, they can be converted in DPP without any sharpening and posted.

That is very useful information :D

The question 'can RAW files be compared with no conversion at all' pops up in every second thread about RAW. Think I'll install DPP, after all.

Best regards,
Andy

[Monito]

However, the shots I took with this lens always seemed out of focus. (I took hundreds of shots on Easter weekend also) That made up my mind to send the lens back.

Perhaps you could post some of these shots that convinced you (100 percent crop and resized full frame, so we can see where the crop comes from). You have not said, and it would be helpful to know what focusing technique you used: manual focus or automatic, which automatic mode (I think DRebel has two), and which autofocus points were triggered for the shots (Canon's software can show you). Also make some outdoor comparisons between the two lenses at the same focal length and the same crop, making sure there is lots of vegetative or architectural detail and that the shutter speed is in the 1/1000 sec range and fstop around f/8 to f/11 (about two stops open from the widest aperture). You posted one shot and two mysterious unlabelled crops. The full frame shot you posted looked gorgeous, but it was unlabelled.

Looking at things with 100 percent crop on the monitor closeup is like looking at a 36 x 24 inch print from 12 inches away (based on my monitor and resolutions settings, yours are probably a little different). This is a very severe test. Is the DRebel your first SLR camera, or did you have 35mm cameras and submit them to the same test (huge enlargment and close inspection)? My 20D has revealed flaws in my technique and now I am making better pictures than I ever did.

The Digital Rebel (300D) has 7.4 µm pixels. If your Digital Rebel is the 350D, it has the same size 6.4 µm pixels the 20D has, which is the severest critic of any lens since it has the smallest pixels of any Canon DSLR.

[paulhillion]

Just my 2 pennys worth. I love my 10-22 and it's the main reason I decided not to sell my 20D. Here's a shot with a 100% crop underneath it, nice & sharp! Oh and as you can see, with this lens you have a hard time trying not to include your shadow in the shot!

10mm f8 @ 1/400 sec - handheld.

http://img.photobucket.com/albums/v116/hillion/3.jpg

http://img.photobucket.com/albums/v116/hillion/3a.jpg

[mr.photoguy]

Wide is an sunderstatement with that lens....

I took 2 side by side 18mm shots yesterday of a landscape I wanted yesterday, for a panarama. I be with the 10mm I wouldn't have had to.

[Monito]

I find that my 10-22mm is a great lens optically and practically. Great fun too!