Another 70-300mm DO test

[richard thompson www.fotoz]

Quite in depth review...<br><br>

 

<a

href="http://www.wlcastleman.com/equip/reviews/70_300/index.htm"> ...

Here </a>

[davidhay]

A detailed and balanced review that comes to similar conclusions as I did. I asked myself, did I have a poor copy of the 70-300 DO and a good 75-300. This new review suggests mine were typical.

[jbq]

Reading those tests, am I the only one unimpressed by the 70-200/2.8L IS?

[pbizarro]

I too am surprised to see that this bunch of tests seems to label the 70-200 IS as a lower quality lens than expected. I can say, from experience, that this is not the case.

 

Also, from my experience, under normal shooting conditions, the DO zoom fares extremely well. The flare example comes from a studio shot, so I am not sure how representative it is of "normal" shooting conditions. Finally, I have some shots at 300mm wide open that suggest a better performance than mediocre (see here: http://www.photo.net/comments/image-attachment?comment_id=488821&return_url=%2fequipment%2fcanon%2f70%2d300do%2f%3f).

[yakim_peled1]

My first impressions have not changed. Much too expensive for what it gives.

 

Happy shooting ,

Yakim.

[ron c sunshine coast,qld,a]

Thanks Richard.It gives abit more of a picture of lens characteristics but i don't think i'll bother with those resolution charts.Somethings very wrong if the 70-200 IS get's flogged by a 75-300....

<BR>Why no comparison examples wide open?

<P>Can someone PLEASE do comparisons of real world subjects (not charts or maps)at constant magnification (so we can see how the lenses get better or worse as they are zoomed .come on! this is obvious!!) and at different apertures (especially wide open !)

[william_castleman4]

There seem to be some misunderstandings about the review I posted.

 

1. The 75-300mm III doesn?t outperform the 70-200mm f/2.8 IS. It only has slightly better monochromatic resolution at 135 and 200mm . Once you consider the chromatic aberration and lower contrast of the 75-300, performance is noticeably lower.

 

2. The 70-200mm f/2.8 IS isn?t as sharp as the non-IS lens that built the reputation of the lens. This has been noted by many. I have tested two copies of the new IS version. It just isn?t terribly sharp (but fine for digital cameras).

See here: http://www.wlcastleman.com/equip/reviews/70_200/index.htm

 

3. I don?t think I had a bad copy of the DO lens. If anyone wants to lend me their good copy, I?ll be happy to test it.

 

4. I did test the lens by taking photographs of real birds. Sorry, at 300mm it is a mediocre performer compared to the EF 100-400mm f/4.5-5.6 L or the EF 300mm f/4 L if you print images on Super A3 paper (12 x 18 inches).

[jbq]

Thanks William!

 

#1 would mean that the 75-300 is a worthy contender if the main subject is dead in the center of the frame at f/8 or f/11. Am I wrong here?

 

#2 the results of the dual of the 70-200s are, well, striking.

[william_castleman4]

<P><I>"#1 would mean that the 75-300 is a worthy contender if the main subject is dead in the center of the frame at f/8 or f/11. Am I wrong here? "</I><BR><BR> The 75-300 offers very good lens performance for its cost. At 300mm and under most lighting conditions, the 70-300 DO does offer a small improvement in function at its much higher price. However, compared to the prime lens like the 300mm f/4 L, the performance of both the 75-300 and the 70-300 are disappointing to me. The only Canon zoom lens I have shot at 300mm in the f/5.6 or f/8 range with performance getting close to a prime lens is the 100-400 L. Others with more experience might comment on their favorite zoom that offers performance comparable to a 300 prime.</P>

<P><I>"#2 the results of the dual of the 70-200s are, well, striking. "</I> <BR><BR> When I first received my new 70-200mm f/2.8 L IS and compared it against my old non-IS lens, I was shocked by its lower resolution with Tech Pan. I assumed the IS lens was defective and promptly sent it back to B&H. A few days later, I received another brand new IS lens. Resolution curves with the second IS lens sample could be superimposed on those of the first. When I tested the IS and Non-IS lens with film and an EOS-1Ds I could detect resolution differences on film images, but not with the EOS-1Ds. </P>

[bobatkins]

These are interesting observations.

 

I'd fully agree with the results concerning the 75-300 and 300/4. The 300/4 is a superb lens, the 75-300 (at 300) an OK lens. I'd differ in regard to the 100-400 though, which in my tests (on film) was clearly inferior to the 300/4 and not really significantly better than the 75-300 over most of the image (though it was better at the edges).

 

I've never seen a zoom that could come close to the 300/4L - on fine grain film.

 

The digital/film thing is also interesting. I'm pretty sure that digital is reducing the differences that people see between lenses. Lenses that look quite different when tested on high resolution film can look very similar when used on a digital camera - and remember that the camera with the highest intrinsic resolution is the 10D/300D, not the 1Ds. The 10D/300D has the finest pixel spacing of any Canon DSLR and so the highest raw "lp/mm". This doesn't mask all differences of course. Good lenses are still better - and test better - than bad lenses, but excellent lenses may not perform any better than good lenses with current DSLRs.

 

This may account for some very different opinions being expressed about lenses. If you're shooting with a 1Ds, lens tests done in Tech Pan may be fine - but they don't matter! What matters to you is performance on a 1Ds. Paying double for a lens that's better when using Tech Pan may make no economic sense at all. Small differences between the 75-300 and 70-300DO which show up on Tech Pan may be totally invisible when the lenses are used on a 10D.

 

People might ask why do digital images look so good if they're not resolving differences that you'd see on film, and that's a valid question. I think the answer lies in the fact that digital actually does better in the range from 0 to around 60lp/mm than film does, but film does better from 60 lp/mm out to whatever the resolution limit is (say 100 lp/mm). Higher MTF in the 0-60lp/mm range gives rise to better looking (sharper looking) prints at all normal sizes. If you made 4ft x 6ft prints it's quite possible that prints from film would look better than prints from digital, but if you're making 8x10 or 11x14 prints - or even 20x30 prints - the higher digital MTF at lower spatial frequences makes them look better. It's known form SQF studies that "quality" estimates strongly correlate with spatial frequences between 0.5 and 2 lp/mm in the PRINT, so for a 20x30 print, what really counts is the MTF between about 16 and 66 cycles/mm, with the MTF around 33 lp/mm being most important (if you're shooting with a 10D - for a full frame DSLR like the 1Ds the numbers become 10-42 lp/mm, peaking around 21 lp/mm)

 

What you want for digital is a lens with the highest possible MTF in the 0-65 cycles/mm range. with emphasis on maybe the 5-20 lp/mm range. What happens above 65lp/mm is really pretty unimportant - and "resolution" only measures the spatial frequency at which the MTF gets close to zero, typically the MTF at spatial frequences well above the cutoff limit of digital sensors.

 

Perhaps what we should be measuring from the point of view of digital perfromance is the contrast (MTF) at, say, 50 lp/mm rather than the resolution cutoff spatial frequency. Not that the two aren't somewhat correlated, but the relationship is different from lens to lens and aperture to aperture. The classic case where resolution doesn't give a good estimate of image quality is the mirror lens, where while resolution may be high, MTF at lower spatial frequencies is low due to the obstructed design, and so the images lack contrast and don't look "sharp".

[william_castleman4]

<P>Bob Atkins makes at least several important points worth reemphasizing. I also will offer additional perspective on the EF 100-400mm L.</P>

<P><I><B>Resolution, MTF, and the 65 lp/mm barrier for current dslr's</B></I>: Resolution tests with Tech Pan (soon to be discontinued by Kodak) provide valuable information about optical function in a lens designed for a film camera. They provide useful information if you are still shooting with film. I started testing new and used lenses I bought for film photography to assess their function and to check to make sure they weren't damaged or defective. These tests have helped me to detect damaged/defective lenses and substandard performers. The tests helped me identify the best performance f/stop for the lens with film. The test data also serves as a benchmark should I drop one of my lenses and need to assess if it was damaged. <BR>The value of increasing lens resolving power hits a plateau in current Canon digital slr cameras around 60 lp / mm. You are unable to see a difference in resolution by lenses functioning at 80 lp/mm versus 60 lp/mm (so long as contrast and chromatic aberrations don't significantly differ). You can see the difference between 80 and 60 lp/mm performance in fine grain film. Bob Atkin's points about MTF being especially important to what we perceive as high image quality in the 16-66 lp / mm range and more narrowly for the EOS-1Ds in the 10-42 lp/mm is helpful to understanding why Canon plots their MTF graphs for lines arranged at 10 and 30 lp/mm and why Photodo.com gives MTF results for 10, 20 and 40 lp/mm.</P>

<P><I><B>Back to the EF 70-300 DO versus EF 75-300 lens distinction</B></I>: Even though the 75-300 lens had slightly higher monochromatic resolution at f/8 and 300mm than the 70-300 DO (50 versus 40 lp/mm, respectively), the 70-300 has better MTF (contrast) performance at f/8 at both 10 and 30 lp/mm ( <A HREF="http://consumer.usa.canon.com/ir/controller?act=ProductCatIndexAct&fcategoryid=150" TARGET="_blank">MTF data is at the Canon lens site here</A>). The 70-300 DO also does not have chromatic aberration, whereas chromatic aberration is marked in the 75-300 at 300mm. This explains why images produced by the 70-300 DO at 300mm look sharper than those produced by the 75-300 even though the 75-300 has slightly higher resolution performance.</P>

<P><I><B>EF 100-400mm f/4.5-5.6 L IS lens versus the EF 300mm f/4 L lens and current dslr's</B></I>: My resolution data with film are consistent with Bob Atkin's perspective that the EF 300mm f/4 L IS is optically superior to the EF 100-400mm f/4.5-5.6 L IS lens at 300mm when using fine grain film. However, both lenses have resolution above 60 lp/mm. Data at <A HREF="http://www.photodo.com/nav/prodindex.html" TARGET="_blank">Photodo.com</A> shows that these two lenses have almost identical MTF scores when shot wide open or at f/8 at 300mm. In my tests with the two lenses on an EOS-1Ds, I have been unable to demonstrate differences in the performance of the two (with the exception of being able to shoot at f/4 with the EF 300.) Both of these lenses perform better than the EF 70-300 DO at 300mm which only achieves resolution of about 40 lp/mm on film and has<A HREF="http://consumer.usa.canon.com/ir/controller?act=ProductCatIndexAct&fcategoryid=150" TARGET="_blank"> lower MTF performance</A> than the EF 300 f/4. </P>

[bobatkins]

My only extra comment here is that the chromatic aberration of the 75-300 is, of course, mainly an edge effect. In the center of the image chromatic aberration (or at least lateral chromatic aberration) is not present. For those shooting a 300mm lens (typical uses are wildlife and sports) the need is often for center sharpness, with the edges being relatively unimportant. Users should bear that in mind when looking at data averaged across the frame.

 

Also lateral chromatic aberration is also known as chromatic magnificaion aberration, where by the red image, green image and blue image aren't quite the same size. This is relatively easy to correct digitally, though the correction isn't perfect.

 

It's an interesting question as to whether digital will ever reach the point where it can "test" lenses. Smaller pixels mean more noise, so where will the break even point end up? The noise is intrinsic (it's a quantum effect). The pixel density on the 10D is the same as that you'd see on a 16MP full frame sensor. If, as some are predicting, a 10D replacement comes out with an 8MP 1.6x sensor, that would equate to a 21.5MP full frame sensor in terms of raw resolving power. That would give a "resolving power" of maybe 70-75 lp/m. Closer to fil, but still not quite there.

[jbq]

To answer Bob's comment, I have the feeling that we'll eventually get full-frame sensors with a pixel density comparable to what's currently available on the 8MP digicams. That'll be 120MP. Silicon process improvements will allow for high fill factors even with such small pixels, so that it'll be possible to downsample those images and end up with the same noise as we'd get with smaller sensors.

 

At such a resolution, clearly the sensor will outresolve almost any lens, and won't need any anti-aliasing filter.

[william_castleman4]

<P>I don't understand Bob Atkin's comment about chromatic aberration being an edge effect. <A HREF="http://www.wlcastleman.com/equip/reviews/70_300/ca/808S0932_75-300_300_f56_acrop.jpg" TARGET="_blank">This is the link to the chromatic aberration example in my review article for the EF 75-300mm f/4-5.6 III USM lens.</A> Chromatic aberration is easily noticeable in the dead center portion of the field, as well as the edge areas.</P>

[ron c sunshine coast,qld,a]

Ooops,sorry if it looks like i'm bagging anyone's reviews.

Just abit grumpy i guess.

<BR>Very interesting comments Bob, re the MTF of digi v film.

<br>It does indeed seem like digital gets the very most out of it's 'limited abilities'

[bobatkins]

William Castleman wrote :"<em>Chromatic aberration is easily noticeable in the dead center portion of the field, as well as the edge areas</em>.

"

<p>

 

Transverse (lateral) chromatic aberration (TCA) - the kind normally seen in lenses - is BY DEFINITION an off axis aberration, which means that it does not exist in the center of the image and it increases in magnitude the further off the optical axis you go.

<p>

It's caused by small differences in magnification for different colors, and it's why it appears as, say, yellow fringing on the inside of a line (side closest to center) and blue fringing on the other. In this case the magnification for yellow light is slightly less than the magnification for blue light. This all stems from the fact that the refractive index of glass depends on the color (wavelength) of light.

<p>

Your test shot shows some transversechromatic aberration in the pattern part of which you can just see in the very bottom right corner. It's also visible at the top right and top left

<p>

The center of your image shows nothing I'd charavterize as chromatic aberration. It does show a contrast drop and a blue "wash" which I'd attribute to flare either in the original image or due to the scanner used to digitize the slide, or both. It's not possible to determin the exact cause of the contrast drop without looking at images at various apertures and at various focus positions around best focus. Different aberrations have different in and out of focus characteristics. However it's <em>not</em> due to what most people would recognize as chromatic aberration.

<p>

There <em>is</em> a second type of chromatic aberration, longitudinal (axial) chromatic aberrration (LCA). This can be present in the the center of an image and is due to change in focal length as a function of wavelength. This is fairly easily corrected and is rarely seen in modern lenses. It's characterised by a uniform, diffuse colored fringe around objects, e.g a black box surrounded equally on all 4 sides by a uniform red "fuzz". I see no evidence of that on your test shot, nor would I expect to since as I said above, LCA isn't too hard to correct for.<p>

You can distinguish between LCA and TCA not only from their visual appearance, but because LCA is improved by stopping down, whereas TCA is not. Of course stopping down also reduces spherical aberration and

spherochromatism (change in spherical aberration with wavelength) as well as any tendancy to flare, so you can't attribute sharpening of the image or increase in contrast when you stop down to LCA.

<p>

The best way to figure out exactly what's going on is to use a microscope to look at the aerial image of a bright point source which the lens produces and see how the nature of that image changes as you move though the focal plane. Of course none of this is very important to photographers! Simple "resolution" type tests are all a photographer really needs to know, though measurement of MTF in the 20-60lp/mm range would actually correlate better with image quality that resolution numnbers do.

[william_castleman4]

<P>Thanks to Bob Atkins for explaining lateral and axial chromatic aberration and the possible effects of flare. The central "blue wash" is less severe in ambient light exposures than flash exposures and disappears as the lens is stopped down. Peripheral (lateral) aberration persists at the edges even when the lens is stopped down. It would appear from Bob's explanation that much of the strange color effects mimicking chromatic aberration may be flare.</P>

[bobatkins]

I'd be reluctant to characterize the contrast reduction (blue veil) in the center of the image as more than that, i.e. without more testing it's pretty difficult to identify the origin as flare, some form of sperical or sphero-chromatic aberration etc. or some combination of several different possible sources.

 

If I were describing a lens based on the test image cited above, I'd say that it shows lateral (transverse) chromatic abberation at the edges of the frame and suffers from some loss of contrast of a high contrast taget in the center of the frame due to a blue veiling.

 

I'd also make sure that what I saw on the scan was exactly what I saw through a microscope, since scanning certainly introduces additional defects.

[william_castleman4]

The images are not film scans. They are EOS-1Ds digital files. The changes appear to be more than loss of contrast and blue veiling since the changes are oriented in a uniform manner away from the center of the image and particularly noticeable in black and red patterns with the lens wide open. The other thing that is peculiar is that the changes were seen at 135, 200 and 300mm settings, but not at 75mm. The blue transition got progressively worse with extension of the lens towards 300mm.

[william_castleman4]

<P>I have updated <A HREF="http://www.wlcastleman.com/equip/reviews/70_300/index.htm" TARGET="">the review</A> with addition of 50% MTF data, examples of bokeh of the lens, and lens sharpness examples. Other data that is now of low relevance has been removed.</P>

[william_castleman4]

n/t