24-70L softness normal here? [Archive] - Canon Digital Photography Forums

View Full Version : 24-70L softness normal here?

wibbly

7th of June 2005 (Tue), 13:46

http://camera.fotopic.net/p16007262.html (http://camera.fotopic.net/p16007262.html)

You can see the original & EXIF from there too. 20D, 1/200, f7.1, 24mm

JPG, taken hand held, un-edited. The centre of the frame is much as I might expect before any sharpening/editing.

But if you look at the full size orginal in the corners, (eg bottom left) it gets a bit iffy. Is this normal for this lens, given that the 20D is a 1.6 crop camera so I'm not even at what would be the edge on a full frame camera...

J

Andy_T

7th of June 2005 (Tue), 17:43

7th of June 2005 (Tue), 17:59

I don't own a 24-70L lens, however in my opinion there may be a defect with your lens. To me, the bottom left hand corner looks noticably worse than any other area of the image. Corners shouldn't look like that, especially from L series glass that costs a fortune. Plus you weren't even shooting wide open! If this is considered "normal" for an L lens stopped down to 7.1, Canon are ripping people off!!!

I'd suggest running more tests with your lens and see if you get the same problem recurring in the bottom left corner. If so, then ask for an exchange or repair under warranty (if it's still covered).

Tom W

7th of June 2005 (Tue), 19:55

I'm not an expert here ... but maybe the DOF at f/7.1 is a bit to shallow for that image ?

What was your distance to the subject?

Maybe you try some test shots against a target that's perpendicular to the lens ... e.g. a newspaper taped to the wall or something like this.

Best regards,
Andy

I agree with Andy here - test on a flat, perpendicular subject. Shoot some real-world stuff as well, but you need to control for all variables that can affect the image.

ssim

7th of June 2005 (Tue), 20:28

Even at 1/200th you can start to see camera shake show in an image. I'm not saying that this is it but it could be a contributing factor. I have a 24-70 and absolutely love it. It has been dropped a couple of times (once considerably hard on concrete) and it still is very sharp. What focus mode were you in and did you check to see where you focus points were in the image to see if those particular points were sharp.

As in anything that is manufactured there is the chance that you can get a substandard version. It is certainly worth doing some more controlled testing before trying to return.

Patrick DR

8th of June 2005 (Wed), 15:03

Tom W

8th of June 2005 (Wed), 15:34

I have a threesome photo's on different F's posted here.

http://www.xs4all.nl/~gnista/24-70%202.8/. (http://www.xs4all.nl/~gnista/24-70%202.8/)
Look sharp and clean to me.
There taken @ 50mm, F2.8. F8.0. F22.0.

PAtrick

3 images and the Snickers' bar is still there!! That's willpower - those things are about as tempting as "L" glass, and a lot cheaper.

wibbly

9th of June 2005 (Thu), 01:03

Yup, I can get stuff that looks sharp and 'pops' too when I have shallow DOF, butter like bokeh in the background, and a subject in the middle fo the frame.

And my sample (top of thread) was at f7.1, and I'm prepared to accept as others have said that perhaps DOF was an issue from top to bottom of the tree (though I have to say that the corners don't look just blurred to me, they look like lens resolution/CA problems)

I'm going to take some more samples, but has anyone got samples of stuff where the lens is at it's quality limits -

f2.8
24mm
at the edges
unedited/100% crop

that I can compare with. Of course it's not going to be perfect. I'm just trying to figure if what I'm seeing is what I should expect from this lens... There are wildly varying reports from people with this lens from 'unsuable at f2.8' to an unqualified 'sharp as a tack'. The reality is somewhere in the middle no doubt...

J

Patrick DR

9th of June 2005 (Thu), 11:33

3 images and the Snickers' bar is still there!! That's willpower - those things are about as tempting as "L" glass, and a lot cheaper.

It only took a few seconds to take 3 foto's, and that's how long i could resist the Snickers' bar:lol:

Patrick

CyberDyneSystems

9th of June 2005 (Thu), 12:27

There is no way the top and bottom could be in focus in that image....

And if you take images of backlit tree branches you will allways get CA...

wibbly

11th of June 2005 (Sat), 09:26

cactusclay

11th of June 2005 (Sat), 10:09

Nothing personal, but I think you are chasing your tail. If you want to see if the lens is sharp, then you might try putting it on a tripod, locking up the mirror and shooting some newspaper, phone book ect. If nothing else hand hold it with the flash popped up to freeze any action and that will at the least give you a rough idea of sharpness. I have tested the 24-70 both ways and got the same results. just make sure the film plane and the print are on the same plane. As was said before, alomost any lens will show aberations when the subject is back lit tree branches. Hope that helps.

wibbly

11th of June 2005 (Sat), 10:45

Nothing personal, but I think you are chasing your tail. If you want to see if the lens is sharp, then you might try putting it on a tripod, locking up the mirror and shooting some newspaper, phone book ect.

Yes, I understand. But I'm trying to avoid the pixel peeping ;-) If I read between the lines, you're saying my technique is the limiting factor (and that's fine - something to work on). If I did do some more tests, just how sharp does it get? No lens is perfect. So looking at 100%/f2.8/24mm/the edges on 1.6 crop sensor, what do others see? That's why I was asking for some others' samples. What's a reasonable expectation? Anyone actually posted such shots (newspaper or whatever, 24mm, wide open, etc) as a comparison?

So given the way I'm taking the samples above, do the results look fine, or 'typical'? Or you just can't tell?

And by using this (poor?) technique, I may be getting results equal to what a much lesser lens would produce anyway?

...And is it thus a fair conclusion that to get/see what L gass is capable of (resolution-wise), you need to have tripod-like conditions on a 8Mp camera like the 20D?

This is all about expectation setting for me, and convicing myself that the equipment's capabilities are in excess of what I can get out of it currently. Which would be a Good Thing, as I can then see my leanings result in better pictures. Already I see pictures I take technically and subjectively far superior to that which I was getting out of my G3 :-)

On the other hand, I have a warranty that expires in a few months, so I want to make sure I exploit it if I really ought to. And I don't want all that hassle, if my lens copy's quality is what most people get anyway. If it aint broke I don't want someone trying to fix it. At the price of L glass, many users like me want typical, or better than typical copies. That's all.

J

cactusclay

11th of June 2005 (Sat), 11:39

I never really got as far as the tests you are now doing, I got caught up with the idea that a lens that has macro capabilities in the 1200 dollar range, should be sharp when shooting something like text, or at least that is where I started with my test of five different 24-70's. There was softness on one side or the other or in the middle on the copies I tested and it was at all the aperatures and focal lenghts. All this testing was done in a months time, so I concluded that there must have been a batch of lenses released from canon, that were defective. I can't really see the defects showing up too much in shots like you did, but anytime you have the subjects all on the same plane and filling the frame, I think it would be noticable. The truth of the matter is that for that much money it should be better image quality than a lens half the price and that just is not the case. I have heard there are good samples out there, but I'm still looking. Another point to make is that if the edges are soft on a 1.6 crop camera, then they would surely be worse on a 1.3 or full frame.

wibbly

11th of June 2005 (Sat), 11:42

but anytime you have the subjects all on the same plane and filling the frame, I think it would be noticable. The truth of the matter is that for that much money it should be better image quality than a lens half the price and that just is not the case.

Do you mean the above from your experience of this lens, or from the sample shots I took?

J

cactusclay

11th of June 2005 (Sat), 11:45

from my own experience.

wibbly

11th of June 2005 (Sat), 11:52

The truth of the matter is that for that much money it should be better image quality than a lens half the price and that just is not the case.

What lens in Canon's line up meets that criteria, in your option (half the price, equal quality, approx same focal length range and speed)?

Or do you have to/can you go non-Canon to get that?

J

cactusclay

11th of June 2005 (Sat), 11:55

17-40L

wibbly

11th of June 2005 (Sat), 12:11

Hmm. an f4 lens, right? Well, it certainly does better here

http://www.fredmiranda.com/reviews/showcat.php?cat=27

J

Tom W

11th of June 2005 (Sat), 13:00

My copy of the 24-70 is significantly sharper than my 17-40 lens was at wider apertures. That's not to say that the 17-40 was bad, but that the 24-70 is better. Both enjoy great color, freedom from flare, and good contrast. Both work well wide open, but the 24-70 is incredibly sharp stopped down.

The 17-40 is a great lens though, and I found it very useful on the 10D (1.6 sensor). But the 24-70 is a better match for the 1D2. I found that on the bigger sensor, I didn't need the wide end as much as the longer end. 24 is wide enough for most of my shooting.

Tom W

11th of June 2005 (Sat), 13:13

What do you think of this lot?

http://camera.fotopic.net/c574855.html

Technique was to set centre focus point on the building (should have been near enough infinity in these cases, I think), focus lock (half press shutter in one shot focus mode) and recompose to get the building "target" in a different part of the frame. All reasonably fast shutter speeds, hand held.

All jpg, parameter 2, Av, AWB

You can look at the original, and see the EXIF data.

I've only renamed the files (no cropping, resaving, or anything else).

So trying to take a step away from the theory of how good L glass is supposed to be, and pixel peeping, if YOU have one of these lenses, is the lens I have showing typical performance, given the technique, do you think? :confused:

I don't see any real problems with your images. In fact, I'd say that they are within the range of typical, given what I see on the 100% images. Any normal-large print size will be very good, especially after modest post-processing. Yes, the f/2.8 images aren't as good as the f/8. The depth-of-field isn't real deep so the parts of the image that are closer or farther (mostly closer if you're close to infinity) than the focus point are not going to be in focus. Plus, f/8 is almost always going to yield better results than wide-open apertures on any lens.

Form follows function - when you're using f/2.8, it's because you have low light or you're trying to isolate a subject from the background with a shallow depth-of-field. When you're shooting architecture, you're more apt to use f/5.6-f/11 or even f/16. Close-ups (almost macro) will require stopping down to keep the DOF great enough. Test your lenses with the actual purpose in mind, meaning you probably don't need to test f/2.8 in the type of images that you're shooting. Patrick's "Snickers Bar" images were a more typical situation for the shallow depth-of-field.

cactusclay

11th of June 2005 (Sat), 15:57

Hmm. an f4 lens, right? Well, it certainly does better here

http://www.fredmiranda.com/reviews/showcat.php?cat=27

J
Those are user ratings on FM. Like Tom was saying the 24-70 is suppose to be a sharper lens than the 17-40, so if it is not, then something is not right. Anyway if any lens is soft in a corner or on the side, where it's the 17-40, 24-70 or even the 18-55, something is not right, specially since there are copies that are sharp all across the frame. Why should only some share in a sharp lens while others have to settle for QC inconsistancies?

wibbly

11th of June 2005 (Sat), 18:07

specially since there are copies that are sharp all across the frame.

I would like to see some samples from people that have these lenses. The sort of shots that would actually reveal softness outside the centre of the frame. At f2.8, I like many are often taking deliberately shallow DOF, and looking for creamy bokeh in all but the centre of the frame - not sharpness! But 'often' is not 'always' (say low light situations forcing wide open shots), hence my desire to see what others get in such circumstances.

I get advice on taking "test" shots of newsprint, or telephone books, but as I've nothing to compare such shots to, there's no point in taking them - I've no idea what I should be expecting. No lens is perfect. Has anyone ever actually taken and posted these sort of shots with a 24-70L they consider to be a good copy?

J

wibbly

11th of June 2005 (Sat), 18:09

I don't see any real problems with your images. In fact, I'd say that they are within the range of typical, given what I see on the 100% images.

Ah. Good to hear. This is the kind of response (-ve or +ve) I'm looking for from those with more experience of this lens than me. Thanks.

J

Tom W

11th of June 2005 (Sat), 18:33

I'll throw a few links up. Images taken with the 24-70, either with my 1D Mk II or the previous 10D that I sold last fall.

http://www.pbase.com/photosbytom/image/41034367

http://www.pbase.com/photosbytom/image/43129777

http://www.pbase.com/photosbytom/image/44146562

http://www.pbase.com/photosbytom/image/36845565

http://www.pbase.com/photosbytom/image/44146443

http://www.pbase.com/photosbytom/image/41036844

Hopefully, that will give you a little idea of how the lens reacts.

Tom W

11th of June 2005 (Sat), 19:18

One more (very big) image, prior to processing:

http://www.pbase.com/photosbytom/image/44661217/original

this is the full-sized version of one of the images listed above.