Camera or lens problem - little help?

[tim]

I did a theatre shoot today, and for the first time I tried using my 50mm F1.8 with no flash, F1.8. I took lots of photos - probably at least 100 with that lens at F1.8, and another 400 with the Tamron 28-75 or at a wider aperture. I was horrified when I got home, lots of the photos were out of focus, it was quite upsetting.

It occured to me to check if the lens or camera had a problem. I set the camera up on my tripod, set a metal ruler on a shelf, and took some shots. The camera distance is written above each shot. The "zero" end of the ruler is 5 inches/12.5cm further back than the "large" end of the ruler. I focused on the middle of the "10" in each shot.

I won't state my conclusion here - what do you guys think's up here?

50mm F1.8 @ F1.8, 75cm/30inches from the ruler.
http://mrwild.co.nz/unprotected/potn/nifty1.jpg

50mm F1.8 @ F1.8, 120cm/47inches from the ruler.
http://mrwild.co.nz/unprotected/potn/nifty2.jpg

Tamron 28-75 @ 50mm, F2.8, 75cm/30inches from the ruler.
http://mrwild.co.nz/unprotected/potn/tamron-close.jpg

Tamron 28-75 @ 50mm, F2.8, 120cm/47inches from the ruler.
http://mrwild.co.nz/unprotected/potn/tamron-close.jpg

I bought both lenses from the states, so if there's something wrong with them it could take a little while to get a replacement, not to mention the shipping costs. At least now I think it's the camera/lens not me, it was upsetting to think I could take photos as bad as they came out!

[tim]

Here's some more images. At this point i'm convinced the 50mm F1.8 is front focusing, and the Tamron 28-75 is working fine. The Tamron does vary a little, but it's usually very close to perfect, if not dead one.

Nifty fifty
http://mrwild.co.nz/unprotected/potn/nifty3.jpg

Nifty fifty
http://mrwild.co.nz/unprotected/potn/nifty4.jpg

Tamron
http://mrwild.co.nz/unprotected/potn/tamron3.jpg

Tamron
http://mrwild.co.nz/unprotected/potn/tamron4.jpg

[RAitch]

were you using manual focus?
If you put the ruler at 45 degrees to the lens, it will exagerate the DOF effect... which should make if even easier to see the focus point.

I've had similar problems http://photography-on-the.net/forum/showthread.php?t=66835

[PhotosGuy]

It's "Clear" to me that the problem is with the 50mm & not the Tamron. :(

[tim]

were you using manual focus?
If you put the ruler at 45 degrees to the lens, it will exagerate the DOF effect... which should make if even easier to see the focus point.

I've had similar problems http://photography-on-the.net/forum/showthread.php?t=66835

Nope, auto focus in one shot mode, 500EX focus assist beam turned on. Looks like I have the same problem as you :(

[robertwgross]

You know, you can take a very good lens and drop it once on the floor. It might not even show a scratch or a dent, but the internal optics can get just a bit out of whack. Off-focus shots can be the result.

---Bob Gross---

[tim]

I've never dropped the lens. It's possible someone picked it up and messed with it or dropped it, but I don't think that happened. Either way, it's going to Canon New Zealand to be calibrated.

[prime80]

I just packed my 50/1.8 in a box to send to B&H for an exchange. I've had the exact same problem. Hopefully the replacement won't.

[tim]

I just got a call from Canon. They said (and i'm paraphrasing):
- The lens was already set for optimal autofocus.
- It's designed more for film cameras.
- The motor in it isn't as accurate as USM and that could have caused my focusing problem. That's not an unreasonable suggestion, I was very close (1m) from my test subject and it was wide open. It doesn't explain why all the pictures I took with it were out of focus - i'm not THAT bad of a photographer.

I guess I have to have a play once it comes back, do a few tests, and see if I agree. The good news is that it was covered under warranty even though I bought it from the states :)

[dng]

Hi Tim i noticed on your original thread that you said you used "no flash" for your theater shots. Could this have contributed to poor picture quality? I recently took pictures at my parents anniversary and found that the built in flash was not strong enough, for sharp picture quality. Also for your tests done with the ruler, I was wondering what focus point your camera was set on? It seems that everything is focused to the right. If you have the right focus point selected or automatic AF point selection on, it could be the problem. You can check this with the EOS viewer utility that came with the camera, it tells you which focus point was used. Anyways, dont know if that helped at all but I hope you sort out your problem.

[tim]

It's kinda annoying that they say it's fine. I took photos of theatre at F1.8, using CF4-1, and I was careful, but a lot of the shots were out of focus. I'll have to take another look and see if they're OOF or there's motion blur - I might post a few of the bad shots later to see what everyone else things. They're at home though, i'll do it later.

[tim]

Hi Tim i noticed on your original thread that you said you used "no flash" for your theater shots. Could this have contributed to poor picture quality? I recently took pictures at my parents anniversary and found that the built in flash was not strong enough, for sharp picture quality. Also for your tests done with the ruler, I was wondering what focus point your camera was set on? It seems that everything is focused to the right. If you have the right focus point selected or automatic AF point selection on, it could be the problem. You can check this with the EOS viewer utility that came with the camera, it tells you which focus point was used. Anyways, dont know if that helped at all but I hope you sort out your problem.

No flash doesn't mean poor picture quality, it means you need to be careful to keep your shutter speed high enough. It's possible it dropped too low and i'm seeing motion blur not OOF shots.

For the ruler shot I used the centre focus point. For the threatre shots I used various focus points. I know the centre is meant to be more sensitive, but it shouldn't be that bad. I'll have to do some more testing with that lens when it comes back.

I did some more theatre shots this past weekend of belly dancing (http://www.mrwild.co.nz/Proofs/BellyDancing/index.html), using my Tamron 28-75 this time, again using a variety of focus points and CF4-1. 95%+ of the shots were perfectly in focus, a few were out because of my error, and a number had motion blur, but in general I'm happy with the way they turned out. I'll have to do some more tests on the 50mm F1.8 I guess.

[RAitch]

I found that the focal point matched the dof up to 2.8. Once you went larger than 2.8, the focal point seemed to shift slightly forward. At 1.8 and up close to your subject, it is near impossible to get a focussed image every time. I've used a tripod and stationary object. Manual focus and auto focus.

Now, every test I did was indoors with natural indirect sunlight. Perhaps it has something to do with low light. Maybe I should buy some work lights and retest.
On that same note, I did the same ruler test in ISO 100 and again right away with ISO 800. It was interesting to see the focus shift to the front during my test. I immediately thought something shifted during the shots. But then when I set for 800, it was spot on until 2.8 then shifted forward again. There's something obviously wrong... and I can't see how the problem can be me. I can't bring myself to believe that something is wrong with the lens either.

http://www.photography-on-the.net/forum/showthread.php?t=66835

[tim]

Thanks RAitch, it seems that we're having the same problem. I'll wait until my lens comes back tomorrow and do a whole bunch of tests like you've done on yours. At least i'll know the lens i'm using is fine, after getting it back from calibration. I have to get this sorted soon, i'm doing a wedding soon and I can't be taking photos that are out of focus.

Like you, I have other lenses that work perfectly, my Canon macro and my Tamron 28-75, so it's not my body.

I don't know whether to have this thread locked and move over to the other thread, having two parallel threads on the same problem is a pain.

[Wildewinds]

I have the XT with the same two lenses and I have the EXACT same problem. I'm going to send the 50mm in to Canon and see if they'll fix it.

[tim]

I have the XT with the same two lenses and I have the EXACT same problem. I'm going to send the 50mm in to Canon and see if they'll fix it.

Which lenses are they?

If it's this same problem there's apparently no point sending it to Canon, they can't do anything about it according to what they told me. Then again it can't hurt.

[tim]

My lens is back after repair, and it's exactly the same as before it left. I've done more tests, all wide open at F1.8. From three distances I took 4 shots each, for two of them I prefocused at infinity, for two the opposite. It front focuses most of the time, hits correct focus 10% of the time, and back focuses maybe 10% of the time. When I manual focus the image looks the same as front focusing, though even worse than when I use AF. I repeated the tests with different distances and found the same thing.

At F2.8 I see exactly the same behaviour. Of course the DOF is wider, but it's still skewed towards front focussing.

When I put the Tamron 28-75 on at F2.8 it works perfectly, focusing on the correct spot each time.

Conclusion: according to Canon the nifty fifty is working exactly as designed, but the limitations of the lens are starting to become obvious. Focusing accuracy is inaccurate, and wide apertures aren't usable because of the focusing problem. The lens design is probaby optimised for cost, not for performance, which makes sense given (I think) it's the cheapest EF lens Canon makes. For the price it's still good value, but if you want a higher quailty lens, go for the USM lens.

btw here's what Canon told me, written exactly as they did. Spelling mistakes are mine, as are things in square brackets. I'm not sure what the" designed for film cameras" remark means, a sensor is a sensor whether it's digital or film AFAIK.

Checked lens AF & found to be set to the best position. Checked resolution & found okay. This lens was designed more for film cameras. The AF drive in this lens may have great[er] stopping errors than a lens with an ultrasonic motor".

Ordering the 1.4 now :)

[JusSmith]

Don't you need to focus on an object which is square to the lens with your ruler set against it? How far does the focus zone extend from the centre focus point? Isn't focusing supposed to find the closest object in that zone? Wouldn't that mean the results from focusing on a ruler at 45 degress would always indicate from focusing?

Here is a test shot I did, focus was well within the box, not on the ruler. In my case I found out it was my 300D that was front focusing (around the 48 mark), same lens on the 20D is fine...
http://www.pbase.com/jussmith/image/42677589/large.jpg

[tim]

Don't you need to focus on an object which is square to the lens with your ruler set against it? How far does the focus zone extend from the centre focus point? Isn't focusing supposed to find the closest object in that zone? Wouldn't that mean the results from focusing on a ruler at 45 degress would always indicate from focusing?

Here is a test shot I did, focus was well within the box, not on the ruler. In my case I found out it was my 300D that was front focusing (around the 48 mark), same lens on the 20D is fine...

I don't think you need to focus on something square to the lens - I can take photos of people that are curved quite easily, so I doubt it. The "focus zone" width depends on the aperture setting, but I was talking about the points most in focus.

I'll get a proper focus chart and print it tomorrow, and have another go on the weekend. Given that my Tamron works perfectly in the same tests, I really think it's the 50mm lens. My Canon 100mm macro works perfectly too.

[RAitch]

LOL...

"OK wait... before I take your portrait, I'm going to have to put something beside your head to focus on."

When manually focussing, shouldn't you get what you see? I think that's our whole point. Sure, maybe AF will work better it locks on a perpendicular object... but that's not practical.
I like that idea though... I think I'm going to try that test too (when I get a chance). With manual focus of course.

[Andy_T]

I don't think you need to focus on something square to the lens - I can take photos of people that are curved quite easily, so I doubt it.

I disagree ... and there are many threads to that effect.

If you want to have valid results from the test, you should do it as JusSmith did it.

You can not be sure that the camera focused exactly at the lighted point, as the sensor area for focus is actually larger.

Best regards,
Andy

[JusSmith]

LOL...

"OK wait... before I take your portrait, I'm going to have to put something beside your head to focus on."

When manually focussing, shouldn't you get what you see? I think that's our whole point. Sure, maybe AF will work better it locks on a perpendicular object... but that's not practical.
I like that idea though... I think I'm going to try that test too (when I get a chance). With manual focus of course.

Very funny, ha-ha! Surely the lens will focus on the nearest part of the subject within the "range" of selected focus point? So unless the lens will focus on just the exact middle pixel of your red square in the viewfinder then you're gonna not get focus on the point you expect on the ruler. If the ruler is angled then for a perfect lens you will seem to get a front focus problem. That's why you focus on a square object when (only) doing your tests. This is also why people moan that portraits are often out of focus. When using extreme shallow DOF, then probably just the nose is in focus, the ears won't be!

Does this not sound plausible? Does to me or perhaps someone should put me out of my misery and explain it differently!

[JusSmith]

I disagree ... and there are many threads to that effect.

If you want to have valid results from the test, you should do it as JusSmith did it.

You can not be sure that the camera focused exactly at the lighted point, as the sensor area for focus is actually larger.

Best regards,
Andy

Thanks Andy.

[Andy_T]

"OK wait... before I take your portrait, I'm going to have to put something beside your head to focus on."


Er ... how did you find out my secret of using my manual focus M42 lenses :eek:

No kidding, when focusing manually (visually, actual), I find it a lot easier if there is a technical object with straight lines on which I can see if the focus is on.

Some tip I've heard here on the forum ... When product photographers shoot reflective curved things (like perfume bottles), they often put something contrasty (like a feather) just in front of the object, focus, remove the focusing aid and then shoot :p

Best regards,
Andy

[RAitch]

Very funny, ha-ha! Surely the lens will focus on the nearest part of the subject within the "range" of selected focus point? So unless the lens will focus on just the exact middle pixel of your red square in the viewfinder then you're gonna not get focus on the point you expect on the ruler. If the ruler is angled then for a perfect lens you will seem to get a front focus problem. That's why you focus on a square object when (only) doing your tests. This is also why people moan that portraits are often out of focus. When using extreme shallow DOF, then probably just the nose is in focus, the ears won't be!

Does this not sound plausible? Does to me or perhaps someone should put me out of my misery and explain it differently!

As Andy mentions above, what you're saying would be true for auto focus. I'm not arguing that.

My somewhat sarcastic comment was based on manual focus... and that's my issue.
If I see the eyes in focus after MF and take the shot, I'd expect that to show in the picture. It does not.
AF does have issues as well... and I can see how that can be effected from up close. If the focus zones are larger then the boxes (which they are) and the camera will focus on the closest object (which it does) technically that would explain front focus images where a nose is in focus. I can accept that as an AF explanation (although sometimes I can even get back focus).

That, however, does not explain focussing issues when using MF.

[Andy_T]

What you see through the viewfinder (with diopter adjustment) is not what the sensor sees.

That - and the different magnification of watching a tiny viewfinder or a large image on screen - might provide part of the explanation.

Best regards,
Andy

[RAitch]

If that was the case, why would they put a viewfinder on the camera then? Are we supposed to guess at the DOF and focal points? That doesn't sound to me like photography.

Sure, the light takes a different path to the viewfinder via 2 mirrors... but I'm sure mathematically they have adjusted for that.

Given all that, looking at a smaller "screen" isn't impacting my ability to see the sharp part of an image. Sure, it would be easier on a larger viewfinder (which I'm assuming might be the case with the D20 or MKII) but I'm still able to see where the camera is focussing and where it's not.

[tim]

I disagree ... and there are many threads to that effect.

If you want to have valid results from the test, you should do it as JusSmith did it.

You can not be sure that the camera focused exactly at the lighted point, as the sensor area for focus is actually larger.

Thanks Andy. I'm aware of the sensor area issue. I wasn't focusing on a single point for these latest tests, I was focusing on a dark vertical line I drew on a piece of paper behind the rule, and I drew 6 incersecting horizontal lines. The focus point was in the centre of the ruler. That should be enough for the camera to lock on to the correct place. What confirms it for me is that when using manual focus I found the same thing. Since Canon said my lens is working fine, I have to assume it's a design error.

[RAitch]

Maybe not a "design" error... just a "what do you expect for $100" design.

I'd love it as much as anybody if somebody can proove us wrong. I'm not just "bitching" for no good reason here.
MF is a problem.

[tim]

Ok, it may be a sub-standard or cheap design, or there's still a chance it's just a small sample of lenses that aren't right. I'd like to see tests (http://md.co.za/d70/chart.html) by a few people with this lens, if anyone wants to go to the bother.

[RAitch]

Great document Tim. I'm printing it out and I'll test this weekend. Maybe I'll do tests with my other 2 lenses too (just to compare).

Thanks!!

[DocFrankenstein]

I just did a test and my 50 consistently front focuses by about 7mm...

[CyberDyneSystems]

JusSmith is correct,. the test methodology is unsound.

I'll see if I can dig it up,. but we have had some in depth discussions on how to it correctly.. but do try what JusSmith recommends.

**edit**
Sorry just followed the link,.
I still think a solid object next to ruler is better,. but thats just IMHO.

[CyberDyneSystems]

I just got a call from Canon. They said (and i'm paraphrasing):
- The lens was already set for optimal autofocus.
- It's designed more for film cameras.
- The motor in it isn't as accurate as USM and that could have caused my focusing problem.

On another note,. the third statement I do agree with to a degree... I find that the AF on the 50mm 1.8 to be atrocious,. it's the worst of any lens I have ever used. The only way I ever get anything in focus is by holding the * button in for some time while it jumps around ... if I just shoot the first time it "locks on" I feel I am almost guaranteed an OOF photo.

[RAitch]

JusSmith is correct,. the test methodology is unsound.

I'll see if I can dig it up,. but we have had some in depth discussions on how to it correctly.. but do try what JusSmith recommends.

**edit**
Sorry just followed the link,.
I still think a solid object next to ruler is better,. but thats just IMHO.

Perhaps unsound for AF... but not MF.
Plus, if AF fails to focus correctly, you should see that in the viewfinder. Keyword "should."

I posted some results http://www.photography-on-the.net/forum/showthread.php?t=66835&page=4

[DocFrankenstein]

Perhaps unsound for AF... but not MF.
Plus, if AF fails to focus correctly, you should see that in the viewfinder. Keyword "should."

I posted some results http://www.photography-on-the.net/forum/showthread.php?t=66835&page=4

If you can see where it's focused in the 300D viewfinder at room lighting, you have a good pair of eyeballs.

[tim]

I just did a test and my 50 consistently front focuses by about 7mm...

Thanks Doc... maybe it's a widespread problem that most people haven't realised about yet.

[DocFrankenstein]

Thanks Doc... maybe it's a widespread problem that most people haven't realised about yet.
Well... the motor is incapable of greater precision by definition.

I can live with it no problem though. When doing portraits, my models seem to lack the cranial capacity to comprened my direction to "stay still" and invariably move between the time I lock autofocus and release the shutter...

[RAitch]

How can you NOT see where it's focussed? I have a Rebel XT/350 and don't know how the viewfinder compares to the 300... but with the XT, it's plain as day. With the DOF so shallow, it's SOOOOO easy to see when something's out of focus.
My eyesight is good, but I don't think you need anything special to see it focus.
Granted... it is MUCH easier to see small detailed focus with more light (e.g. Tom's focus sheet)

How can you live with it if your models move? Do you just not use 1.8?
If that's the case, I could understand. Some people without studio lighting need the 1.8 to pull off some shots without having to crank the ISO... plus, it's a great effect for some shots.

[tim]

Just got my 50mm F1.4, and it seems to behave exactly the same as the 50mm F1.8. The pics below are at F1.4 and F1.8 as labeled, both taken with the F1.4 lens.

[tim]

This is with my new 70-200 F2.8 IS, first at 70mm, then at 200mm. Both are perfect, same as my Tamron 28-75. What's going on here?!

[tim]

Any thoughts? Why would the 50mm lenses haev problems when none of the others do? I'll bring a better focus test chart home tomorrow and try it, but all lenses except the 50mm ones work fine. Maybe it's time to call Canon...

[RAitch]

I'm at the point where I'm guessing the response will be "what did you expect for $150CAD???"

At first I left the 50 on my camera all the time. I'm sad to say that now the kit lens has taken over. I haven't used the 50 since my last testing session.

[tim]

Yeah but I have the 50mm F1.4, which is about $300, not the cheap one! Why would 50mm lenses be out, but no others?

[RAitch]

Good point... considering the kit lens doesn't have a problem... mind you it's nowhere near 1.8.
I wonder if the 85 has the same problem... but honestly, I doubt I'll buy one.

Bummer!!

[Tom W]

For the record, my old focus test from last year with the 10D, 50/1.4 at f/1.4:

http://www.pbase.com/photosbytom/image/43262038/original.jpg

The lens must have something on which to focus. The AF points are larger than the little squares in the viewfinder. Any contrasty boundary that the focus sensor sees is subject to it locking onto that point.

Here's the 10D's AF points in relation to the focus squares in the viewfinder for reference (courtesy of Chuck Westfall, with actual sensor coverage areas colored in red by me):

http://www.pbase.com/photosbytom/image/42250385.jpg

[Tom W]

PS - sorry about the wide post - I wanted the image to be large enough to see the focus scale.

[tim]

Tom, your 1.4 on the 20D looks perfect! Wonder if my camera's at fault, which would be suprising given the 3 lenses that work perfectly.

[Tom W]

Tom, your 1.4 on the 20D looks perfect! Wonder if my camera's at fault, which would be suprising given the 3 lenses that work perfectly.

That was actually on my 10D, which I no longer own. I had a very good copy of that camera. Still have the 50/1.4.

Considering that your other 3 lenses are giving you perfect results, I'd be more inclined to consider that the lens is faulty. But do try a test procedure more like what I've used - you want to make sure that nothing attracts the camera's focus point except for the intended target. And, don't overlook the shallow DOF on these fast lenses at wide aperture.

[tim]

Yeah sorry, I read 10D and wrote 20D by default. I know it's not narrow DOF, i'm not saying it's out of focus, i'm saying it's most in focus to the right of the single centre focus point I used.

I have that same focus test chart printed and i'll try it this evening. I've left a message with the Canon lens repair department, maybe they'll have an idea what's going on. Strange how the 50mm lenses are the only ones playing up.

[Tom W]

Well, the Narrow DOF in conjunction with the micro-motor focus mechanism is not going to provide quite the same precision as the ring-type USM focus system (which the 50/1.4 should have, but doesn't). The two 50's are the longest fast lenses without the ring-type USM system, and are thus the most likely to show less precision partly due to the shallow DOF and partly due to the inherent "play" in the geared focus drivetrain. Yes, the kit lenses have geared drivetrains as well, but they don't need to be nearly as close to perfect as the fast 50's.

I would like to see what Canon has to say as far as what the specification of focus precision is - and how it applies to various lenses. Hopefully, they will give you a suitable answer. The 50/1.4 is somewhat of a test standard for several camera characteristics, so I would hope that Canon uses it for its precision measurements as well.

[tim]

The 50mm F1.4 isn't real USM? What a ****ty trick by Canon calling it USM! I got it because the Canon technician told me the USM lenses were much more accurate, and I didn't think to check what flavour of USM they were!I'm tempted to return both the 50mm lenses and get a real USM lens - though 35mm is too short and 85mm is too long. Damned lenses!

Thanks for the info Tom. Do you have a link to a list of what lenses are real USM and which aren't?

[Tom W]

The 50/1.4 has an ultrasonic motor lens, but not a ring-type ultrasonic motor. It is a micro-USM drive unit, which operates on a similar principle (both have the Piezoelectric drive that is characteristic of the USM), but has an output gear that turns the focus barrel, as opposed to the ring-type that is a direct-drive unit.

The 50/1.8 has no ultrasonic motor at all, but instead relies on a DC microdrive motor. The 35/2 uses the Arc-Form drive, which according to Canon is a brushless "circular deformation" drive motor. Both are geared to the focus barrel.

[tim]

I'd pay more for a direct drive ring type USM. It seems faster and better than the 1.8, but not $300 better. I'll do more through tests this evening.

[ron chappel]

It's impossible to do accurate focus tests on a tilted object -like that ruler.

You MUST focus on a flat object with the tilted ruler adjacent to it as jusSmith showed -otherwise the edge of the focus sensor can catch on closer marks on the ruler and (incorrectly) show occasional front focus errors.

[robertwgross]

The ruler images at the top of this page... did you notice the ghost numbers in there?

By any chance do you have a UV filter on the front of a lens?

---Bob Gross---

[RAitch]

Tom, thanks for the comments. Of course, that's true if you're using AF. Most people seem to use AF and not pay attention to what's in focus in the viewfinder... trusting the camera.
What about the problem where what you're seeing in the camera is perfect... but when you take the shot, it turns out like trash?

Even if the focus ring isn't USM... I'm puzzled why that would matter. If you're waching the viewfinder to see what's in focus, why would USM matter (other than comfort)? If the picture looks in focus, does it matter if the lens was USM or not? Sure, it might be easier to get there.... but does it really matter for picture quality?

When I was doing my MF tests, I could easily see the focus/DOF point change. Sure, USM would feel better... but finding the center point wasn't too difficult at all in nice bright light.

[tim]

It's impossible to do accurate focus tests on a tilted object -like that ruler.

You MUST focus on a flat object with the tilted ruler adjacent to it as jusSmith showed -otherwise the edge of the focus sensor can catch on closer marks on the ruler and (incorrectly) show occasional front focus errors.

My slanted ruler test works perfectly with the 100mm macro, the 70-200 F2.8 IS, and the Tamron 28-75, and only fails with the 50mm lenses, so I don't believe that it's impossible. I will do the test with a flat object just to give it a try.

The ruler images at the top of this page... did you notice the ghost numbers in there?

By any chance do you have a UV filter on the front of a lens?

Nope, no UV filter. That shadow's because the paper I taped to the back of the ruler is 1-2mm behind the ruler, because I didn't tape it tightly enough. I'll use a proper focus test tonight, which won't have this problem.

[RAitch]

Maybe there's an issue with fast short lenses in combination with the mirrors to the viewfinder... but I can't understand how that could be.

I hope I don't come across as a complainer... I just have a hard time letting things go without answers.

[tim]

Maybe there's an issue with fast short lenses in combination with the mirrors to the viewfinder... but I can't understand how that could be.

I hope I don't come across as a complainer... I just have a hard time letting things go without answers.

I like to get answers too, and I generally keep asking questions until I get them answered. In this case the question's gone to Canon, hopefully they can come up with a reasonable explaination without me having to send all the lenses and camera in.

Note that the camera is at most one meter from the ruler, the minimum focusing distance is 0.5 meters so that should be fine, but such a short distance is about the most difficult test you can do with the lens - the closer to the object the more precise you need to be. I'll try at various distance tonight.

[tim]

I've done proper tests now. I set up the focus test chart at 45 degrees, with the left side behind the right, the top parallel to the bottom, and the centre of the lens at approximately the same height as the middle of the test chart. I put the camera on the tripod with mirror lockup, used a cable release, and did all but one of the shots with auto-focus. I've come to a conclusion myself but i'd appreciate it if one or two people could take a look at my test shots and let me know their conclusion. The shots can be found here (http://www.mrwild.co.nz/ExperimentalGallery/Tests/FocusTests/index.html). Thanks to anyone who takes the time to do this.

[ron chappel]

My slanted ruler test works perfectly with the 100mm macro, the 70-200 F2.8 IS, and the Tamron 28-75, and only fails with the 50mm lenses, so I don't believe that it's impossible. I will do the test with a flat object just to give it a try

I don't mean that it's impossible.I meant it's impossible to trust such a test.Somewhere on the web there is an excellent article by canon on how to do an accurate focus test (and why it is important to do it that exact way) .
Unfortunately i haven't been able to find the link to it-still looking

In the meantime-test with a flat object with the tilted ruler next to it-otherwise you are going nowhere

[tim]

Ok, i'll give it a go, Ron, thanks. If you find that article i'd be interested to read it - I should learn something from it at least :)

[tim]

Ron, i'm *VERY* glad I took your advice! :) The 2nd set of tests done as you suggest gives a completely different result to my test earlier this evening, and agree with my earlier conclusions. You can see the test shots here (http://www.mrwild.co.nz/ExperimentalGallery/Tests/FocusTests/Set2/index.html). Focus was on the vertical text "Hencho en Japon", which despite how the pictures look are on the same focal place as the top of the 100mm mark on the ruler. The "100mm" mark should be in perfect focus in all the shots.

The Tamron 28-75, Canon 100mm macro, and 70-200 F2.8 IS work perfectly, and focus dead on. Both the 50mm lenses clearly front focus. Hopefully i'll get to talk to a Canon technician tomorrow, and see if they have any thoughts. Either way i'll have to send the two lenses in, hopefully without the camera body, which appears to be working fine :)

I have a mate with the 50mm F1.4 lens and 20D, I might get him to come over some time and repeat the tests.

[Tom W]

Tim, that last test tells the story with little room for doubt - there is some front-focus on the 50/1.4, and a good deal of front focus on the 1.8. These are the kind of images you should show Canon in your coorespondence.

I would like to see what your friend's 50 and 20D do in comparison, and how each lens does on the other person's camera.

[tim]

I'll do the comparison and let you all know what happens :)

[slin100]

Those look front-focused to me, but I have a bad feeling that Canon may claim the performance is within spec.

[Tom W]

Those look front-focused to me, but I have a bad feeling that Canon may claim the performance is within spec.

Indeed, they may, since the specification is based on a circle-of-confusion that relates to a 6" by 9" print viewed at 10 inches. It's hard to relate that to the screen, but the fact is that the 100% crop is here to stay. Digital, as it is used today, is more demanding.

[tim]

I think you could tell this focusing error in a print. I'll call Canon again and see if I can talk to a technician.

[ron chappel]

I'm glad the testing worked out :)-i wasn't sure i had worded myself properly and may have sounded abit blunt (long story-i have an illness that makes pain so grumpiness is abit of an occupational hazard :):))

I too will be interested in what canon decides on the focus error thing.

[tim]

I'm good at blunt too ;) Thanks for your help, i'll post again once Canon have had a look at the lenses.