A survey of consumer telezooms, sort-of

[rdenney]

What can I say, all this reading about lenses has given me a bad case of Lens Lust, despite that if I buy one more expensive lens I'll be knocked upside the head with it.

So, I had to content myself with a cheapie.

I spent some time looking at MTF charts and understanding them. I think I can now look at them and tell how good a lens will be in terms of contrast, resolution, and even background blur (up to a point).

Armed with that knowledge, I went to Photodo to reviewe the general performance of 75-ish to 300-ish 4x telezooms. Back in the 70's, I owned a Vivitar 75-260 zoom in a Canon FD mount that performed quite well, and I still have it as well as a Canon FD 100-300 f/5.6. These lenses had reasonable contrast, considering they are long zooms and not primes, and they were sharp for prints up to 11x14.

That FD 100-300 showed a combined MTF score of 2.9, so I figured that was a useful floor for consideration.

With the sole exception of the Zeiss Vario-Sonnar for the Contax, the best of the lot in this general range was, of course, the 100-400 L, despite that it's a bit long for my purposes. But that lens is heavy and well-built, and would not feel good getting knocked into the side of my head, hard though my skull be. So I looked a bit downmarket.

I found an astonishing result. The lowly $200 Canon 75-300 is just about the best-rated 4x telezoom on the list, with the exception of the longer L and the Zeiss lens.

Photodo rated the Canon lens at 3.1. Nikon's best offering was negligibly better at 3.2. The old Canon EF 100-300 5.6 was a bit better at 3.3, and Tokina had a 100-300 at 3.1. But all the 4x zooms that started at 70-80 (where my mid-range zoom stops) were worse. My Cosina 100-400 was the worst of the lot at .9, though that is no real surprise to me (though I've still gotten good images from it).

Yes, I know that the front element extends when it focuses, and that it rotates. Yes, I know that it doesn't have ring USM and it's slower. Yes, I know it's plastic piled on plastic. But it has a metal mount and I think the build quality is similar to my 20-35, which is just fine for my purposes. And it's light. An all those aspects of cheapness make the price drop down into the skull-saving range, and I don't need it for sports.

So, tonight I bought one (camera shops that stay open late should be banned).

Then, I went to the Canon site and studied the MTF curves that they publish there. The good news is that you get to draw a line on the curve at 13.4mm--the half diagonal of the 10D sensor. Lots of the bad edge performance just disappears when you do that. The only time the MTF fell below a threshold of 60% was when considering resolution wide open, when racked all the way out to 300. At f/8, MTF values at 300 stayed above 63% for resolution and 83% for contrast (nearly as good as a racked-out 100-400L for contrast), with virtually no loss of performance at the edge. At f/8, 75mm, it was above 70% for resolution and 90% for contrast, which is actually pretty good. At the edge of the 35mm frame, it goes downhill from there, though.

So, when we look at lens performance for the 10D, let's remember what we already know: We are only using the center of the lens. That opens the door to lots of possibilities for those of us who are already in trouble with our budgets because of that $1500 digital camera body.

Of course, MTF is a few pixels on my computer screen. We'll see how it does making real images. Tomorrow's post from me might be decidedly less upbeat.

Rick "who wonders why Canon doesn't include that dadgum lens shade" Denney

[Stoneh]

thanks for that, very usefull.... awaiting the results eagerly

[MarkH]

This is the 75-300 III USM?

These pics were taken with mine:
http://arrdee.net/%7Eflyinass/markpics/Web/Gulls.htm
http://arrdee.net/%7Eflyinass/markpics/Web/Gulls2.htm

I've added a lens hood since, not sure how much difference it makes.

I might replace it with an expensive L lens next year, but for now I have a huge credit card bill to work on.

For now it gives me the zoom I want and really is good value at the price.

[Jeppe]

Yes.. the Ef 75-300 will give you resonable results for the price, but compared to L-glass its not really compareable.
Rdenny: Well yes you just use center of glass, but still there is a VERY large differential between the glass in canon budget and canon L.

You will most defently see the differential between L-glass and budget-glass. And its just that diffrential that matters for some photographers, especially pros.

0.5 pts in the MTF-chart is a visible change. So an lens with MTF 4.1 will show a clearly viseble change in quality over an with 3.1

[rdenney]

Jeppe wrote:

You will most defently see the differential between L-glass and budget-glass. And its just that diffrential that matters for some photographers, especially pros.

0.5 pts in the MTF-chart is a visible change. So an lens with MTF 4.1 will show a clearly viseble change in quality over an with 3.1

Of course. I never doubted that the L lens was better, and in fact I identified it as such. I just can't afford it.

But Photodo only rates the 100-400L as a 3.6, which is your .5 "visible" change--just. The L was made for making slides with no post-processing opportunities, and I can correct for flat images from this lens to some extent. What surprised me was that nobody makes a really first-class 4x zoom starting at 70 or 80 mm. A couple were as good, but most were worse. Reviled as it is, there is nothing better for less than four or five times the price.

Here is an image I made with the cruddiest of all zoom lenses:

http://www.rickdenney.com/images/pinnacle_in_last_light_lores.jpg

This Cosina 100-400 lens rated a 0.9 on Photodo, with its worst performance at the long end that I used for this image, making it the worst lens in the Photodo database. The image still seems useful to me, though it would not support much enlargement. I used a tripod and a small aperture, to be sure.

3.6 for the L may be better than 3.1 for the 75-300, but 3.1 is a vast improvement over 0.9.

My needs are few, heh, heh.

Rick "covering a lot of gap between what he has and an L lens for a coupla hundred bucks" Denney

[Jeppe]

I thought it was the other way around accually. That the Cosina/Soligor/whatever was at its best @ 400.

Also, you cant compare MTF-values just like that. You can only compare if they are @ same aperture and same focal length/zoomfactor.

Dont get me wrong here.. Im not complaining, but i have tried an 75-300 and it doesn't deliver for me.

But im sure that it will for some people (nothing wrong with that)

[rdenney]

Jeppe wrote:
I thought it was the other way around accually. That the Cosina/Soligor/whatever was at its best @ 400.

Also, you cant compare MTF-values just like that. You can only compare if they are @ same aperture and same focal length/zoomfactor.

Dont get me wrong here.. Im not complaining, but i have tried an 75-300 and it doesn't deliver for me.

But im sure that it will for some people (nothing wrong with that)

Well, you may be right, and I may hate it when I actually make some images with it. And my wife may determine that even it wasn't cheap enough, and its poor build quality will get tested against the side of my head.

On the Cosina, though, it's worse racked out. MTF is in the dreadful 40's at 400, and the almost mediocre low sixties by 100. And the weighted MTF for resolution (at 40 lpm) is only 16%! I suppose that means they looked but couldn't find those parallel lines.

Used with care, it provides some results. But it's no action lens. I suspect the same will be true of the Canon, but we'll see.

I read a really good review of the similar 90-300 a little while ago. The reviewer (whose name escapes me--and IE crashed before I could record the link) drew a graph of the sweet spot on this lens. He said 5.6 or better at minimum focal length, 8 or better at 200, and 11 or better at 300. But he also had another curve that showed the lens getting worse as you focused to infinity at the longest focal length, and so for infinity focus or near it, he suggested f/16 or smaller at 400. I'd bet that's a lot smaller apertures than most people use with it. We'll see how much of a PITA that turns out to be, but I like the guy's approach.

More later. One thing's for sure: If the Canon is no better than the Cosina, it's going back to the store.

Rick "who hasn't had it on the camera yet" Denney

[TerryS]

Rick,

You have clearly put a lot of thought into your lens choice, however there is an error in your analysis. You write: "So, when we look at lens performance for the 10D, let's remember what we already know: We are only using the center of the lens." I hate to say it, but this is not actually true. Every point on the surface of the objective lens contributes to every point in the image.

When a camera that has a sensor smaller than a 35mm film negative, such as the Canon D30/D60/10D, then it is the image that is being cropped and not the light that is incident on the objective. Points near the periphery of the objective will still contribute to the image on the sensor.

However, having said all this, it is true that, in general, imperfections in the lens reveal themselves most towards the edges of the image, and these are being cropped away by the D30/D60/10D sensor. So although there is some validity in your argument for using 'consumer' lenses, it is not for the reason given and the benefits of using higher quality glass will be greater than you suggest even with a small sensor.

Best regards,
Terry.

[rdenney]

TerryS wrote:
Rick,

You have clearly put a lot of thought into your lens choice, however there is an error in your analysis. You write: "So, when we look at lens performance for the 10D, let's remember what we already know: We are only using the center of the lens." I hate to say it, but this is not actually true. Every point on the surface of the objective lens contributes to every point in the image.

When a camera that has a sensor smaller than a 35mm film negative, such as the Canon D30/D60/10D, then it is the image that is being cropped and not the light that is incident on the objective. Points near the periphery of the objective will still contribute to the image on the sensor.

However, having said all this, it is true that, in general, imperfections in the lens reveal themselves most towards the edges of the image, and these are being cropped away by the D30/D60/10D sensor. So although there is some validity in your argument for using 'consumer' lenses, it is not for the reason given and the benefits of using higher quality glass will be greater than you suggest even with a small sensor.

Best regards,
Terry.


I don't disagree with what you say, but when the center performance of the lens measures to be better than the edge performance, it is the center of the image where they are doing the measuring.

I have performed extensive testing of lenses for my funky second-world medium-format cameras, measuring center performance separately from edge performance, and the differences are visible right on the film.

If you look at Canon's MTF curves, the horizontal axis is the diagonal distance from the center of the image as it is formed at the focal plane. We can draw a line at 13 or so MM from the center, instead of 21.5. That cuts off a lot of the worst part of the curves for many lenses--even the L lenses when used wide open.

Rick "for whom the point is nearly moot anyway" Denney

[rdenney]

Jeppe wrote:

Dont get me wrong here.. Im not complaining, but i have tried an 75-300 and it doesn't deliver for me.



Well, in the end, it didn't for me, either.

I found that the lens worked acceptably at 300mm. But it only did so when I was at f/11 (when at infinity) or f/8 (when closer). In practice, though, I could not achieve these settings. The lens is too light, and provided no inertia and no place to hold it steadily. Thus, to use it at 300mm, I needed a big tripod, or a shutter speed of at least 1/500. The monopod was not good enough, surprisingly--the lens just wiggled back and forth. Bracing it against the nearest tree, which is a favorite trick of mine, doesn't work because the whole front half of the lens rotates and there's no stable surface to brace against. I can see why Canon chose this lens for it's consumer IS.

Had the ergonomics of the lens been acceptable, I might have kept it. But I took it back to the camera store Saturday morning, and I brought my wife with me. I figured if I was going to get smacked, I might as well protect myself by being in a public place!

Sigh. I bought the 80-200 f/4 L.

You guys were right. I'm sitting here looking at a f/5.6 image from the L compared with the f/11 image from the 75-300, and the L lens makes a sharper, snappier image. Of course, that was expected. But what really sold me on the L was the bokeh. The 75-300 isn't actually all that bad in this department, but I think part of the reason for that is that there is a general softening of all details. I found the L lens had more pleasing bokeh, with out-of-focus highlights shown as a soft-edge disk. It was better than the 50mm macro (which wasn't bad at all), and far better than the cheapie 35-80. I'll be posting some images of that comparison in the next few days, hopefully.

The nearly instantaneous, utterly silent focusing was what sold my wife, plus the pleasingly smooth bokeh. (Though she doesn't know what to call it, she's highly sensitive to it, I've found. She has more wisdom and less technology.)

In the end, the lenses that make me want to make images are the ones I don't regret buying, no matter what the cost, and in my wandering through the MTF curves and other technical issues I had forgotten that. Oh, well, it did me no harm to learn it yet again. After taking my few test shots, I found that I could not stop with the L lens, and spent another hour wandering around the garden capturing the week's crop of flowers.

Rick "with a smoking credit card" Denney

[CyberDyneSystems]

What an excellent read this thread was!

Thank you Rick!

A lot to learn here.

CyberDyne "bummed that more money allways seems to equal better" Systems