• Please use real names.

    Greetings to all who have registered to OPF and those guests taking a look around. Please use real names. Registrations with fictitious names will not be processed. REAL NAMES ONLY will be processed

    Firstname Lastname

    Register

    We are a courteous and supportive community. No need to hide behind an alia. If you have a genuine need for privacy/secrecy then let me know!
  • Welcome to the new site. Here's a thread about the update where you can post your feedback, ask questions or spot those nasty bugs!

Canon EF-S 18-200mm f/3.5-5.6 IS lens

Doug Kerr

Well-known member
For several years, we have used as the "walkaround lens" on our EOS 40D a Sigma 18-200 mm f/3.5-6.3 DC OS. As you may recall, a while ago, we looked into improving our situation in this regard, the major motivation being to get an increase in the maximum focal length (we were of course actually seeking improvement in "reach", a parameter that considers resolution as well as focal length).

We purchased, tested, and returned two candidates: the Sigma 18-250 mm f/3.5-6.3 DC OS HSM and a Tamron AF 18-270 mm F/3.5-6.3 XI Di II VC LD.

Briefly, we found that in each case, the increased maximum focal length was accompanied by a more than corresponding decline in resolution such that no greater "reach" was actually attained. There were also some other anomalies that, although not major, helped solidify our decision not to adopt either of those lenses.

Recent work done here on AF system behavior helped illuminate some of the shortcomings of the current Sigma lens in the area of autofocus, and there are some irregularities in its optical stabilization behavior as well.

As a result, we again opened the issue of "upgrade" and decided to try out the only remaining contender, the Canon EF-S 18-200 mm f/3.5-5.6 IS USM.

IMG_153000.jpg

Canon EF-S 18-200 mm f/3.5-5.6 IS USM (B&H photo)​

First indications are quite encouraging. Whether its improvement over the Sigma would actually warrant the cost ($595.00 + $31.95 for the hood, B&H) is another matter altogether.

On the image quality front, I find that hand-held at 200 mm, f/8 (shutter speed about 1/1600), with AF, there is a perceptible (but not substantial) general improvement in image sharpness. I have the feeling that contrast is also improved over the Sigma.

Lateral CA is better in some places and worse in others. I have not looked into the geometric distortion story.

The maximum focal length for focus at a distance of 500 feet is about 6% greater than the Sigma (I don't have the actual number yet).

In any case, I haven't done much actual image work with the lens yet. The rest of this initial report will be only on "engineering observations".

The construction of the Canon lens is quite nice. Among other things, even when fully extended (for the top range of focal lengths), there is negligible play between the three telescoping sections.

The zoom ring works smoothly through its range, and has a fairly long throw. The lens will shift in focal length when pointed downward or upward at a substantial angle. There is a zoom lock, but it can only be engaged at minimum focal length.

The autofocus operation seems faster than on the Sigma lens (at least in situations where focus search is not required); this is, however, a difficult property to quantify. In any case, it seems that the focus normally "settles" after only two "rounds" (typical for my other Canon lenses), whereas the Sigma generally took five "rounds" to settle.

Focus search speed is graded with focal length setting, and seems fairly conservative.

The lens uses a DC micro motor, and does not have bona-fide full-time manual focusing. The manual emphasizes that we should not touch the ring when it is turning under motor power nor attempt to turn it by hand with the AF/MF switch in AF (in which case the motor drive is engaged).

Nevertheless, except when the motor is actually moving, it seems quite easy to turn the focusing ring, obviously taking the motor along for the ride. There is clearly no type of braking in effect after the completion of focusing movement (there may well be dynamic braking just at the end of the movement). But I find it hard to believe that turning the ring in this situation, with the switch in AF, actually puts any untoward strain on the gear train. So I am not at all reticent about "tweaking" the focus with the ring with the mode switch still in AF (of course being sure that the motor has stopped being driven).

By the way, the focusing ring is fairly narrow and is well-separated from the zoom ring (which minimizes the likelihood of contact with it when it is moving under motor power. But it is not difficult to grasp and operate when needed (even with the hood in place). There are no distance markings on it. It has a modest throw (I think about 80°) and works well for manual focus (either legitimate or otherwise).

The image stabilization (IS) is well behaved and seems effective. I have no way to quantify the degree of its effect.

The IS seems to "come up" very quickly. It seems as if the IS is shut down very quickly after the shutter release button is fully released. This is probably beneficial with regard to battery consumption.

The lens is 1 mm longer than the Sigma (when collapsed), about 1 mm greater in diameter, and the same weight (21 oz, not including the hood).

The overall conclusion on this lens is that it is a worthwhile upgrade.

I expect to have some actual photographs soon!

Best regards,

Doug
 
Last edited by a moderator:

Nill Toulme

New member
I have a sense that sample variation plays a significant part in the performance of these superzooms. I've seen various comparison reviews with, as I recall, at least one of each coming out on top.

I had a Sigma 18-200 that I liked OK, but I have since traded it for a Canon 15-85 IS. Nice lens.

Nill
 

Doug Kerr

Well-known member
Hi, Nill,
I have a sense that sample variation plays a significant part in the performance of these superzooms.
Oh, I'm sure.

I've seen various comparison reviews with, as I recall, at least one of each coming out on top.

Yes, but one factor is that they all display different "priorities" in their compromises: C/A in various regimes, sharpness, geometric distortion, and the like.

Then there is the matter of curiosities. For example, the Sigma 18-250 does not have a "caging" mechanism for the image stabilization compensating element; when the IS system powers down, it the element just drops to one limit of its travel. Thus, when we half press, the image position jumps significantly. Only one review I saw mentioned this. Some users might find it unacceptable, some merely curious.

A side effect is that, since return of the image to its "element parked" position after a shot might be bothersome to the user, the stabilization system is kept powered up for a long time after full release of the shutter release button, probably not good from a standpoint of battery consumption.

I had a Sigma 18-200 that I liked OK, but I have since traded it for a Canon 15-85 IS. Nice lens.
I've heard it's nice.

Thanks for your inputs.

Best regards,

Doug
 

StuartRae

New member
Hi Doug,

The autofocus operation (of the Canon lens) seems faster than on the Sigma lens........

Could this have something to do with the fact that in theory the 40D shouldn't be able to autofocus the Sigma at the long end where it's f/6.3? As I understand it, the lens has to lie to the camera and pretend it's f/5.6. Of course this doesn't increase the amount of light available.

Regards,

Stuart
 

Doug Kerr

Well-known member
Hi, Stuart,

Could this have something to do with the fact that in theory the 40D shouldn't be able to autofocus the Sigma at the long end where it's f/6.3? As I understand it, the lens has to lie to the camera and pretend it's f/5.6.

Well, I'm sure the "ability" to do AF does not drop off sharply beyond f/5.6. Insofar as the widely-held belief that the camera will decline to do AF for a reported maximum aperture worse than f/5.6, my current understanding (based on scattered reports) is that the camera actually attempt to do AF so long as the reported maximum aperture of the lens is better than f/8.0.

I have no lenses or lens-extender combos here by which I can confirm this.

In any case, I can find no evidence of the Sigma "lying" about its maximum aperture. Of course, that may be reported in different ways, and I may have no visibility of the one that counts! But everything in the Exif metadata (including in the MakerNote, the proprietary part) says f/6.3 (actually reported by my tool as f/6.2 - this may be a result in rounding of the Av, which is how that is actually carried in the metadata).

In addition, unless the maximum aperture is reported in one way for this purpose and another way for metering purposes, I find no discrepancy in metering, which I would expect if the actual maximum aperture were f/6.3 but was reported as f/5.6 (although the discrepancy would be only 1/3 stop, so I might not be able to discern it).

The Sigma seems to focus accurately even at 200mm ("/6.3").

And the need to do five passes to perfect focus seems to still be present at a focal length at which the apparent maximum aperture is, for example, f/5.6.

But there are many mysteries to all this!

Thanks for your thoughts.

Best regards,

Doug
 

Doug Kerr

Well-known member
Here are some shots that illustrate the Image Stabilization feature of the Canon EF-S 18-200mm f/3.5-5.6 IS lens on my EOS 40D. The blue truck was at a distance of about 500 ft.

These shots were all hand held, using what I would describe as "careful but not obsessive" steadying technique.

In each case the focal length was (nominally) 200mm, and the shutter speed 1/20 s (that being four stops slower than the traditional guideline, 1/320 s in this case).

In each series, with and without IS, I took four shots. The ones shown here were "middle of the road" from each series with regard to blurring.

Here we see, for context, the entire frame of the representative shot without IS (ex camera except for reduction in resolution to about 23% of the original):

Canon_18-200_IS_F18295R900.jpg

Canon EF-S 18-200mm f/3.5-5.6 IS, IS inactive, reduced resolution

Here we see a 7.7% x 7.7% crop of the original frame, emphasizing my favorite resolution test target, a license plate:

Canon_18-200_IS_F18295C1.jpg

Canon EF-S 18-200mm f/3.5-5.6 IS, IS inactive, original resolution​

Here we see the entire frame of the representative shot with IS (same circumstances as before):

Canon_18-200_IS_F18291R900.jpg

Canon EF-S 18-200mm f/3.5-5.6 IS, IS active, reduced resolution​

Here we see the 7.7% x 7.7% crop of that frame:

Canon_18-200_IS_F18291C1.jpg

Canon EF-S 18-200mm f/3.5-5.6 IS, IS active, original resolution​

Best regards,

Doug
 

Doug Kerr

Well-known member
Here, for comparison, we see the same scene shot at the "suggested" shutter speed, 1/320 sec. (The ISO sensitivity was set to produce essentially the same aperture as before.) IS was inactive.

Canon_18-200_IS_F18300R900.jpg



Canon_18-200_IS_F18300C1.jpg

For comparison, here is again the shot at 1/20 s with IS on:

Canon_18-200_IS_F18291C1.jpg

Best regards,

Doug
 

Doug Kerr

Well-known member
Hi, Stuart,

Of course this doesn't increase the amount of light available.
Keep in mind that the matter of a minimum required aperture for AF does not involve the matter of "enough light on the AF sensors". If it did, then of course the needed aperture would depend on the scene brightness. (There is a minimum on that, by itself, too, but it does not interact with the aperture requirement.)

In fact, the amount of light landing on an AF sensor is not determined at all by the lens aperture (assuming that it is big enough for the AF process to work at all). That is because the AF sensors have their own apertures, in order that their exit pupils lie (for the two sensors of a pair) on opposite sides of the actual lens exit pupil. Those apertures have a fixed size.

I discuss this matter here:

http://www.prophotohome.com/forum/c...s-vs-release-priority-chuck-5.html#post483506

Best regards,

Doug
 

StuartRae

New member
Hi Doug,

Doug Kerr said:
Insofar as the widely-held belief that the camera will decline to do AF for a reported maximum aperture worse than f/5.6, my current understanding (based on scattered reports) is that the camera actually attempt to do AF so long as the reported maximum aperture of the lens is better than f/8.0.

Thanks.

It's very difficult to get at the truth. The manual for my 50D says:

For lenses whose maximum aperture is larger than f/5.6, with all AF points cross-type AF sensitive to both vertical and horizontal lines is possible.

It doesn't say that for lenses smaller than f/5.6 AF is not possible, but implies that cross-type AF is not. Maybe with apertures less than 5.6 AF is sensitive only to vertical lines? The Canon info-base also implies this, but (I think) only for professional bodies and quotes a definite cut-off point at f/8.

Please excuse me; I'm just thinking out loud.

Regards,

Stuart
 

Doug Kerr

Well-known member
Hi, Stuart,

It's very difficult to get at the truth.
Indeed.

My conclusion after several days of intensive study of all the literature (via interpolation, extrapolation, and triangulation!) is this, for both the EOS 40D and 50D:

a. At each of the nine AF points, there is a "cross-type" detector (V-H). Both its sensor pairs (for the horizontal-line-sensitive aspect and the and vertical-line-sensitive aspect) are said to be workable for apertures of f/5.6 or better. (There are two sensors for the horizontal-line-sensitive aspect and two for the vertical-line-sensitive aspect, thus my word "pair".)

b. At the top, center, and bottom points, the horizontal-line-sensitive aspect sensor pair has a "2-line zigzag" design. Thus supposedly increases the ability of this aspect to make a focus error determination in the face of more severe initial misfocus than otherwise (thus averting the need to do focus search in some cases). There is no implication that these special sensors have an aperture requirement other than f/5.6 (although of course their precise actual behavior in that regard might be a little bit different than the "ordinary" "f/5.6" sensors). (And of course, none of this says that these sensors just give up at f/5.7.)

c. In addition, at the center AF point, there is a second, independent detector, cross-type (but at 45° orientation). Both of its aspects are said to be workable with an aperture of f/2.8 or better. Presumably it is ignored for an aperture worse than f/2.8 by some degree.

My summary of my conclusions of this type for several contemporary EOS bodies is found here:

http://www.prophotohome.com/forum/c...rs/92055-canon-eos-af-point-arrangements.html

***************

Although this strategy is only openly mentioned in connection with some 1-series bodies, it is entirely possible that is followed in a 40D/50D:

With an f/2.8 lens aboard, initial focus error determination is made based on one (or more) of the "f/5.6" detectors. (They will probably make a determination in a more severe initial misfocus situation than for the F/2.8 detector.) Then, after the lens focus cam is moved to the position reckoned as ideal by the lens CPU based on the initial error report, and if the center AF point is in effect, the camera looks at the unique f/2.8 detector at that point and makes its second (hopefully final) error determination from it.

Best regards,

Doug
 

StuartRae

New member
Hi Doug,

Doug Kerr said:
My conclusion......... is this.........

Thanks for your explanation. It makes sense and more or less agrees with what I've been able to find out.

WRT the lens "lying" (maybe an unfortunate choice of word), it can be asked two questions:

  • What aperture are you set to?
  • What is your maximum possible aperture?

The first question will be answered truthfully and will be entered in the Exif metadata.
The second question, asked for the purpose of AF, may be answered with a little white lie. Err, about 5.6.

I'm sure I've seen this somewhere, but for the life of me I can't remember where.

However, the fact remains that your 18-200, like the 150-500 I borrowed from a friend, does AF quite well at the long end.

Thanks fro your time. I shall continue to dig and if I come up with anything I'll post it.

Regards,

Stuart
 

Doug Kerr

Well-known member
Hi, Stuart,

Thanks for your explanation. It makes sense and more or less agrees with what I've been able to find out.

WRT the lens "lying" (maybe an unfortunate choice of word), it can be asked two questions:

  • What aperture are you set to?
  • What is your maximum possible aperture?

The first question will be answered truthfully and will be entered in the Exif metadata.
The second question, asked for the purpose of AF, may be answered with a little white lie. Err, about 5.6.
Well, perhaps. But answered where, and how? In the MakerNote part of the Exif metadata, there is an item MaxAperture. For the Sigma, it shows f/6.2.

That is probably a result of this item being actually stated in Av form, perhaps, to a precision of 1/8 unit. The nearest value in that case, expressed as an f-number, would be (to 3 decimal places) f/6.169. (Some other APEX values in the MakerNote seem to be to a precision of 1/8 unit.) In the actual data, the APEX values are expressed as a ratio of integers, and the manufacturer can choose any denominator they wish for any given quantity. But "binary" values (8, 256, etc) are the most popular.

Ah, yes, I poked around a little in the data and that seems to be the situation.

I'm sure I've seen this somewhere, but for the life of me I can't remember where.
Same here.

It may be folklore, or was maybe something that used to happen. (These stories have a long life!)

As of now, I can't see the need, since supposedly the camera is willing to attempt AF so long as the reported maximum aperture if better than f/8.0. (Might not have been so on the EOS D30, for example!)

Thanks fro your time. I shall continue to dig and if I come up with anything I'll post it.
Please do.

Thanks.

Best regards,

Doug
 

Doug Kerr

Well-known member
Here we get to see a basic "architectural" shot with the EF-S 18-200mm f/3.5-5.6 OS. Again, the victim is 438 Charlie Way, 500 feet down the street from the World Headquarters.

This was done on a tripod with IS off, AF, ISO 400, f/8.0, 1/166 s.

Here is the full frame at 23% of original resolution, ex camera (no sharpening, not even after resampling):

Canon_18-200_IS_F18302R900.jpg

Here is an off-center 23% x 23% crop, original resolution, ex camera.

Canon_18-200_IS_F18302C900.jpg

The lateral CA is readily visible in the crop. I suspect it will respond well to correction. I do not consider it debilitating in the full frame presentation.

Best regards,

Doug
 

Asher Kelman

OPF Owner/Editor-in-Chief
Here is an off-center 23% x 23% crop, original resolution, ex camera.

The lateral CA is readily visible in the crop. I suspect it will respond well to correction. I do not consider it debilitating in the full frame presentation.

Best regards,

Doug

Canon_18-200_IS_F18302C900.jpg


The cutout is a good place to find out how good this might be. can you now sharpen this:

Unsharp masking in these 4 ways and then other ways you might like. Fade in "Edit, Fade" use Luminosity and set to 90%.

  • 6 % 60 pixesl

  • 12% 24 pixels

  • .3 pixels 300%

  • .6 pixels 300%

If you can correct CA first, that would be a bonus but we should still get a good idea as the file is.

Asher
 

Doug Kerr

Well-known member
Hi, Asher,

Here is that same extract but from the full frame with the lateral CA mitigated:

Canon_F18302-02C900.jpg

Perhaps you could be so kind as to apply what you feel would be a valuable type of sharpening.

This is what I would have ordinarily applied without any real thought:

Canon_F18302-03C900.jpg

Thanks so much.

Best regards,

Doug
 

Doug Kerr

Well-known member
Here is the extract from the image, at original resolution, with CA mitigated, processed with four combinations of local contrast enhancement and edge sharpening (via unsharp mask) as suggested by Asher:


Sharpen_F18302-ac.jpg

Unsharp mask: 6%, 60 px + 300%, 0.3 px

Sharpen_F18302-ad.jpg

Unsharp mask: 6%, 60 px + 300%, 0.6 px

Sharpen_F18302-bc.jpg

Unsharp mask: 12%, 24 px + 300%, 0.3 px

Sharpen_F18302-bd.jpg

Unsharp mask: 12%, 24 px + 300%, 0.6 px​

All look quite nice.

Thanks, Asher.

Best regards,

Doug
 
Top