The pain of distortion...

[kitacanon]

I've posted it here rather than the photo folders because I think it's an equipment issue...
shot on the 10d/50F:1.8

here's the full frame RAW, no post processing
http://img507.imageshack.us/img507/2148/crw2661pjsmu7.jpg

here's the right corner...nice and sharp...
http://img507.imageshack.us/img507/1061/crw2661prcmu7.jpg

here's the left corner...ouch...
http://img333.imageshack.us/img333/6833/crw2661plcpq0.jpg

worse is I've seen it on the 24-85 on my d30...wierd...

any ideas appreciated...thanx...(I've sent this on to Canon and am waiting for their reply)...

[Ephemeral]

Did you shoot at f1.8? Just wondering if it's a DOF issue.

[kitacanon]

1/640 @ F7.1
the left corner and right corner are the same distance from the camera...and the distortion is not mere softness...it curved lines (blades of grass)...

[SkipD]

I don't understand what you are seeing. Can you mark up the "bad" image with what you are seeing (notes, arrows, etc., written onto the image)?

If this is something you see repeatably in other images, then surely you could find a subject that would be better for a test. You might use, for example, a brick wall of a large building so that you could compare portions of the image that have very similar subject material.

It would also be useful to post 100% crops of the image to make analysis easier. It does not appear to me that your corner crops are at 100% of the original resolution.

[jra]

I would agree that this is a poor test subject for what you are trying to show us. I think I see what you're talking about but I'm not sure. It certainly doesn't appear to be a very big issue in this pic. Maybe if you give us some more examples and show exactly where you're talking about.

[John_B]

kitacanon,
I also don't see much of a difference between your crops, I suggest you follow SkipD suggestion :)

[jra]

Also, keep in mind....pixel peeping can literally drive a person to insanity ;)

[kitacanon]

To answer your questions...
There is no pixelization on any of these pix.
The grass is bowing towards the center on the shot of the left corner.
A shot of a brick wouldn't show the "creep" distortion (evidence of bowing/barreling) as clearly as grass...
Each corner is a 100% crop
Each corner shot is about the same distance from the center...
I just made the left corner large so it'd be easier to see the grass bowing clockwise.

I guess what you can't see won't pain you I guess...

[SkipD]

Make a test pattern target - a fairly large grid of parallel straight lines in vertical and horizontal directions. Place the target in positions that would have it at the four corners and at the center of a photograph at a reasonable distance (same distance for all shots like shooting a large wall with the camera's view perpendicular to the wall). Level the camera (on a tripod, of course) and level the target at each location. Use manual focus for the distance. Set the exposure to manual so it too stays constant. Shoot five photos (or make five targets and do one shot). Then pixel-peep to see if there's any distortion.

[kitacanon]

I realized after posting that I didn't load the full res shots...here are 90% res...that is, at 100% crop the res was 929 high so I reduced them to 800 for viewing here...

http://img72.imageshack.us/img72/4936/crw2661ptrcltr4.jpg


http://img72.imageshack.us/img72/8125/crw2661plclsu6.jpg

The suggestion to use a tripod or other view are not relevant as the image on the right is fine, and on the left is not...if there were a problem with either shake or image shot then the problem would be throughout...and again, if you don't see the distortion you don't feel the pain, and the luckier you are for it...

[SkipD]

If you want to really prove that there's a problem with the lens, you will need a more "scientific" test shot, in my opinion.

Something makes the grass in the "left corner" shot look out of focus, but it does not look distorted to me. In order to prove the distortion theory, you would need some reference lines rather than grass that grows any way it wants to. You should also have a set of targets at the four corners of the photo and the center (at the very least) to be able to diagnose the lens' characteristics.

[kitacanon]

Well, if you can't see the difference between the two photos....

[SkipD]

Well, if you can't see the difference between the two photos....Of course I can see a difference. However, that photograph does not provide any measurable information to truly determine not only IF the lens may be defective or not but HOW the lens may be defective.

What I am saying, is that the grass was not groomed to be perfectly vertical. Any non-vertical positioning of the grass in the real world is an unknown variable at this time. Additionally, nobody can tell for sure on what portion of the scene you had focus lock. It's true that the fast shutter speed should have more or less eliminated camera shake from the equation.

The grass in the lower left of the whole photo appears to be out of focus to me, not distorted (as in 'bowing towards the center'). I would seriously doubt that anybody at Canon's service center is going to try to make a decision as to the condition of your lens (and/or camera) with the grass (and the unknown factors about the grass) as a target and with targets at many different distances from the camera. They would probably want your camera and lens sent in so that they could shoot some standardized test targets and then analyze the results.

Shooting a standardized target of some sort as I suggested before will provide a lot of information that can be adequately analyzed. From that, an informed decision as to the source of the problem can be derived. My suggestions to shoot something with a known set of parameters is to help Canon figure out your problem without your having to send the whole rig to them.

One needs to think with an analytical process in mind to solve some problems like this.

[ylnad123]

could it just be that you accidently set your af point to the left?

[kitacanon]

Good thought...so I went back and no...the trees to the left are the same distance as the bushes and trees at the right so even if it WERE the case, the left would still be in focus...thanx for the idea...it made me go back to the shot and check out focus of the trees above the ground...

If it's just the left side I will have to remember that when I shoot...but I'm going to shoot the same shot with my 24-85 (when it stops raining) and see if I get the same results...that would mean...
1. both lenses are ill...
2. the sensor is not perfectly on plane...
3. the lens mount is a bit off plane...
4.......

[basroil]

4, your 50mm was not up to specs and one of the elements is off center. i will have to say that it looks a lot like this option, considering how it looks from that photo. i've seen plenty of photos taken with lenses that had to be sent to cannon for realignments. in this case, it's probably cheaper to sell the lens used and buy a new one.

[kitacanon]

4, your 50mm was not up to specs and one of the elements is off center. i will have to say that it looks a lot like this option, considering how it looks from that photo. i've seen plenty of photos taken with lenses that had to be sent to cannon for realignments. in this case, it's probably cheaper to sell the lens used and buy a new one.

Ouch...well, I was thinking of getting a 50/F1.4...
I will take up your suggestion Skip and shoot some perfectly vertical lines to see if/how they distort...thanks for the suggestions, all...

[kitacanon]

From Canon...
1. The CMOS sensor is not perfectly aligned in the plane of focus,

2. One of the elements in the lens is misaligned.

3. It could possibly be a charateristic of this copy of the lens.

The same scene shot with my D30 proves that the issue is the lens not the sensor.
I shot the same scene, same settings, with my d30 and the distortion , (with consideration given to the softness due to the lower resolution of the D30 vs. the 10D) is identical...

Thus the distortion is a lens related issue...but I imagine one that can not be repaired at reasonable cost...
I'll just have to remember that the left/lower edge (on a vertical) is the edge to be trimmed when cropping to an 8x10...
...or get a 50/F:1.4 and hope for better...

[Lester Wareham]

You would be best to shoot a plane test target as Skip suggested, the traditional thing to do is pin some newsprint on the wall, mount camera on tripod, align the lens centre to the centre of the newsprint and adjust distance and position to fill the frame whilst making sure the camera back is parrallel to the target. Use mirror lockup and cable/timer release.

This will allow an unambiguous test that you can take back to the supplier.

If it confirms your earlier findings it is probably some decentering of the optical elements.

All lenses seem to have some degree of decentering, the problem is greatest with fast wide angle lenses, particuly zooms as they have higher complexity. It should become less apparant as you stop down so take your test shots over a range of stops.

How much decentering is acceptable is not defined anywhere, but if it visible at arms length on an A4 print I would say it is clearly a problem. The thing is not to get carried away with expecting perfection.

As an example here are 100% test crops from the top centre edge and bottom centre edge of my 50/1.4 @f1.4, the largest variation I could see from an perfectly acceptable lens. The top is much softer than the bottom. Needless to say this would be invisible in an A4 print at normal viewing distance.

http://www.zen20934.zen.co.uk/photography/LensTests/EF_50mm_f1.4_USM/f1.4/Crops/B.jpg

http://www.zen20934.zen.co.uk/photography/LensTests/EF_50mm_f1.4_USM/f1.4/Crops/J.jpg

Same test areas at f4, the top and bottom are almost equal.
http://www.zen20934.zen.co.uk/photography/LensTests/EF_50mm_f1.4_USM/f4/Crops/B.jpg

http://www.zen20934.zen.co.uk/photography/LensTests/EF_50mm_f1.4_USM/f4/Crops/J.jpg

[kitacanon]

The chart you have here would NOT expose the distortion evidenced in the scenic, but thanx for the effort...
I'll make a test shot that would be more clear how the distortion is not merely lack of focus...

[Lester Wareham]

The chart you have here would NOT expose the distortion evidenced in the scenic, but thanx for the effort...
I'll make a test shot that would be more clear how the distortion is not merely lack of focus...

In that case I think we are not clear on what you see in your shots.