Testing Four Often-Recommmended EF Value Lenses

[zachary]

EF-S 18-55mm f3.5-5.6

EF 28-105mm f3.5-4.5

EF 50mm f1.8

EF 70-200L f4

 

 

At least one of these lenses are almost always mentioned when someone

asks about buying a first or second lens for their DSLR camera. They

are versatile, inexpensive and relatively inexpensive. Depending on

your needs, chances are one that of these lenses will help you. So

now that I finally have them all I would like to help others by

testing thier true optical performance and posting the results on

photo.net. You ask, "what is wrong with all the other lens tests

around?" Nothing. The only thing is that they are all tested by

different people using different cameras and people all have

different techniques. By testing them all myself, you will see a

level playing field all around. Obviously zoom factors are not the

same on all of the lenses, but there is quite a bit of overlapping. I

will also pot results of focal lengths that the other lenses dont

match (such as the 18mm of the 18-55 and the 150 and 200 of the 70-

200). This isnt a lense comparison test ("which one is better?"), its

merely their optical performance.

If I can help one person by this then the whole test is worthwhile.

If anyone has any suggestions for the test please let me know.

 

I will be using a Rebel XT, (ISO 100), a Bogen 3021, and the test

will be indoors so I can control environmental factors such as light,

fog, and wind. This will not be shots of FTM charts or technical

charts, it will be a real life type test so you can actually see how

a real life subject will be affected.

 

Results should be posted in the next few days.

 

 

Zach

[jimstrutz]

Don't tell me about the labor pains, Zach. Just show me the baby. :)

[caesar]

I think the EF-S 18-55 that people talk about is the EF-S 18-55 USM, not the kit lens, which is non-USM...

[bobatkins]

Makes no difference. Same optics.

[mark_chappell]

<I> By testing them all myself, you will see a level playing field all around. </i><P>

 

Except that you're not accounting for the possibility of sample variation, of course. If you

want to do this test in a really meaningful way, you'll need to get six or eight copies of each

lens and test them all, using appropriate descriptive statistics. That should give us a

reasonable indication of how much intersample

variation to expect. For zoom lenses you should test multiple copies of both new and well-

used units, to make sure that all those moving parts (lens elements moving back and forth,

etc.) don't lose fine tolerances with wear and tear.

[zachary]

Oh yeah, a quick question, should I use Paramater 1 or Parameter 2? 2 is more neutral, but differences will be more clearly seen with 2, which I think is more like the way people will be using the lenses.

 

Yes,I understand that I cant count for lens variations (I dont have 8 lenses of each :), but the test will be assuming I have an average one, which I would say accurately describes 80-90 percent of lenses that leave Canon's factory.

 

 

Zach

[taner]

Unless you focus at infinity, focus accuracy of each lens will always be potential weak links in your test; how about 'real life targets' at infinity (outdoors obviously)?

[bobatkins]

How do you know you're properly focused at infinity? How is it easier and more accurate to focus at infinity than at 20ft?

 

Using manual focus and just hitting the long focus limit stop isn't an accurate method of focusing.

[chris76]

I can save you some time to tell you that performance will be as follow

1. 50mm - Good from F2.8 very good

2. 70 - 200L - Very good

3. whats left - Okay to good - Stopped down

 

Dont own all the lenses but have used them at one stage or have done some research on them.

 

Dont let this stop you as it will be nice to see if we agree in our findings.

 

Good luck

 

Chris

[mark_chappell]

<I> which I would say accurately describes 80-90 percent of lenses that leave Canon's

factory.</i><P>

 

And you base this assumption on.... what? This is sort of like flipping a coin twice, getting

heads both times, and concluding that coins always come up heads. Without testing

considerably more than one lens, you have <B><I>no basis at all</i></b> for

concluding that a particular copy is average, or if 80-90 percent of produced lenses are

'accurately described' by a given sample. This is fundamental statistical theory.<P>

 

It may be that QC is excellent and that most lenses are very similar to each other, but

without a reasonable sample size there's no way of knowing that in any meaningful way.

The only thing you can conclude from testing a single copy is that at least one lens is that

good (or bad).

[sheldonnalos]

Since you're already up to the test, I'll add my own lens tests as an additional data point for 3 out of the 4 lenses involved. (Sorry, no 18-55mm since I'm using a 10D). Hopefully having all these extra tests in one thread will help with showing sample variation as well.

 

Regarding methodology on my tests, I've found that focusing at something closer to infinity (30-50 ft) helps take slight focus errors out of the equation. I've used my backyard on most tests. I use Parameter 0 on the 10D (sharpening, contrast, saturation and tone all set to zero), mirror lockup, tripod, self timer. You might also give my formatting a try, since it makes visual comparison for others much easier.

 

Here goes a bunch of posts, hopefully I don't clog the thread too bad!

 

Sheldon<div></div>

[sheldonnalos]

and another...<div></div>

[sheldonnalos]

whoops, another try at that one...<div></div>

[sheldonnalos]

and yet another...<div></div>

[sheldonnalos]

And more still<div></div>

[sheldonnalos]

And on to another lens...<div></div>

[sheldonnalos]

And the edge crop...<div></div>

[henrik.ploug]

"If you want to do this test in a really meaningful way, you'll need to get six or eight copies of each lens"

 

Then we would never see any tests.

[sheldonnalos]

On the wide end, in the center...<div></div>

[sheldonnalos]

On the wide end, at the edge...<div></div>

[sheldonnalos]

At the long end, in the center...<div></div>

[sheldonnalos]

At the long end, edge crops, pardon the focus errors on this one...<div></div>

[sheldonnalos]

And one more of the 70-200mm... center crops...<div></div>

[sheldonnalos]

And the last one... edge crop....<div></div>

[sheldonnalos]

So, now that I've completely wasted the bandwidth of the forum, here's my rough thoughts.

 

The 50mm f/1.8 represents an excellent value for the money, though I did find one of the two copies I've tested to be a little on the soft side. Still, I hold nothing against this lens.

 

The 28-105mm is relatively good, but nothing stunning. For a mid range zoom it's not a bad performer at all. Still, a 50mm prime does outperform it, as does L series glass (which of course you would expect).

 

The Canon 70-200mm is also quite amazing, a great value even at $550. I was quite impressed that it held its own against the bitingly sharp 85mm f/1.8. You really can't ask for much better glass than this. Sorry I didn't have any tests to show the 70-200mm f/4 L at the long end. My experience has been that it remains sharp throughout the range.

 

Anyhow, all this really should be hosted on a web server somewhere, but I don't have my own site and don't really have the inclination to do the work (beyond testing the lenses).

 

I'm also excited that I'm going to probably be getting even more lenses to add to my testing database here in a month or two. I should be able to test a 16-35mm f/2.8 L against my 17-40mm, a 70-200mm f/2.8 IS against my 70-200mm f/4, and a 28mm f/1.8, 50mm f/1.4 (another copy) and a 85mm f/1.8 (again, another copy). A friend is going to loan me his full kit as a backup on a wedding I'll be shooting.

 

Anyhow, hope this is of some use to you all!

 

Sheldon