Anybody feel the same way?

ljfinger said:

RedFox88 said:

Tests on the 8MP XT will show for instance that the 50 1.2
resolves 25% more than the 50 f1.4 no matter how what value the 50
f1.2 is said to resolve. It's a relative thing.

What does it matter if you look at a test on a 8MP camera and lens X
gets a score of 100 vs. a test on a 15MP camera and same lens X gets
a score of 200? In relatives terms the score is the same.

Click to expand...

Unfortunately, that is not true, and that is the problem. If it were
true, then there would be no real problem with these tests.

The problem is, the math doesn't work that way. The way it works is,
if the sensor is low res, the difference between a sharp lens and a
soft lens is extremely tiny in the combined system result - so tiny
that the difference might even fall into the error band, making the
test useless.

Click to expand...

Review their tests. The results aren't "tiny". I respectfully disagree.

This is one of the myths I was talking about. Lenses don't "outresolve" sensors. Resolution is a combination of sensor and lens (and RAW conversion). A lens may be markedly inferior to the resolution possible by the sensor or vice versa but this is NOT outresolving. Using the term implies that somehow improving the sensor even further and using an inferior lens will produce NO improvement in resolution, it will! The improvement will not be as great as that with a superior lens but it will still be there.

Using a sensor that has superior resolution to the lens also means that less sharpening needs to be used, a weaker AA filter can be substituted and jaggies and other digital artefacts will be less noticeable.

Flying Fish said:

Chris59: Thanks. I think I'm getting there. I see how a good lens
can produce a very good image if you get everything right, and so if
you don't get everything right, the difference will be very
noticeable. But would the image with a good lens and bad technique
be worse than the same image with a poorer lens and the same bad
technique? In other words, is it the comparison with what the lens
is capable of that makes the difference?

FF

Click to expand...

No. Poor technique with a good lens will give a better image than the same poor technique with a poor lens BUT... and it is a BIG but... poor technique will absolutely SWAMP any differences in the ability of the lenses. Even moderately good technique with an ordinary lens will give an image that looks better than a superb lens and slightly sloppier technique.

Try this experiment. Use a 50mm prime lens and say the 18-55mm kit lens using the best techniques and shooting the same subject for both. Use ISO 100, a tripod, the sharpest aperture, a remote release, mirror lock up and focus precisely. Now, compare the photos and you will see that the prime lens produces a sharper more detailed image with better contrast and probably better colour.

Next, take the prime lens and shoot the same scene but take away the tripod and shoot hand held at ISO 400 with the aperture a stop smaller than optimum using auto focus. Now compare the image at 100% to the kit lens as taken above and you will see what I mean.

I think it was silly using the 350D in the first place, it has the flakiest AF and isn't very demanding of lenses due to the lack of rez - the 5D or better still (for AF reasons) an old 1DS mk1 should have been used as well as a 40D for the job so FF can be tested and the lens given a faitr trial for rez and AF ------ that is if you're doing it seriously which PZ seem to want to do.
At least they can test EFS lenses properly now they have a 50D

--
Please ignore the Typos, I'm the world's worst Typist

You cannot compare the LW/PH numbers directly and say that "1315" on the 350D makes this a poor lens, but scoring "1736" with the 50D makes it better. That's not true. In both cases, the score is just barely into the "GOOD" range. Just as was the case when comparing how lenses were scored on the 350D v. D200, you have to compare the normalized result rather than the actual LW/PH value.



v.



People made the same mistake in declaring the Nikon 18-200 VR to be sooooo wonderful and the Sigma 18-200 OS not so. They were comparing the LW/PH values directly, but that was skewed by the difference between the 350D and D200. Normalized via Photozone's rating scale, they have virtually identical overall performance.

RedFox88 said:

they shouldn't start testing/retesting lenses with the 50D instead of
the XT and here's why.

Ok, you look at the same lens they test of the XT and then the 50D
and suddenly the lens resolves better you might think, right? No.
It is true as their site says, that you cannot compare test between
the 2 camera models used.

They could use an old 3MP dSLR for their tests and it would still
show which lenses are better than others. You use a standard test
and you ARE testing the quality of the lens based on the parts of the
test. Tests on the 8MP XT will show for instance that the 50 1.2
resolves 25% more than the 50 f1.4 no matter how what value the 50
f1.2 is said to resolve. It's a relative thing.

What does it matter if you look at a test on a 8MP camera and lens X
gets a score of 100 vs. a test on a 15MP camera and same lens X gets
a score of 200? In relatives terms the score is the same.

Click to expand...

RedFox88 said:

Review their tests. The results aren't "tiny". I respectfully
disagree.

Click to expand...

Their results are closely bunched, and they have a history of obviously wrong results that indicate they have a very wide error margin. For example, they used to claim the 18-55 was sharper wide open than the 300/2.8L IS.

--
Lee Jay
(see profile for equipment)

Many thanks.

FF

ljfinger said:

Luke Kukovica said:

... what is a real lens test ? ¿

Click to expand...

A test that would determine the MTF of the lens, not of the lens-AA
filter-sensor-demosaicing software combination. There are multiple
ways of doing these.

Click to expand...

Tell us about these multiple ways to test a lens without a body.

ljfinger said:

RedFox88 said:

Review their tests. The results aren't "tiny". I respectfully
disagree.

Click to expand...

Their results are closely bunched, and they have a history of
obviously wrong results that indicate they have a very wide error
margin. For example, they used to claim the 18-55 was sharper wide
open than the 300/2.8L IS.

Click to expand...

All sites make errors and it sounds like they corrected it since you say "used to". They don't even have a 300 f2.8 listed on that site anyways.

If you want to call their results "closely bunched", go ahead. But you can still see that a lens resolves 25 or 40% more than a different lens.

RedFox88 said:

All sites make errors and it sounds like they corrected it since you
say "used to". They don't even have a 300 f2.8 listed on that site
anyways.

Click to expand...

They removed it because so many people like me were making fun of them about it. I now notice they've removed their teleconverter tests, probably since they were self-contradictory.

RedFox88 said:

If you want to call their results "closely bunched", go ahead. But
you can still see that a lens resolves 25 or 40% more than a
different lens.

Click to expand...

No, you can't. You can tell one system does, plus or minus some large error bounds. You can't tell how much different lenses are, especially sharp lenses.

Let's take the Rebel XT tests as an example. According to DPReview, the camera is capable of around 67lp/mm, which is about right considering the theoretical maximum is 76.8lp/mm. Let's say we test two lenses, one capable of 100lp/mm and one capable of 150lp/mm. The results from PZ, if they do it right, would record 55.7lp/mm and 61.2lp/mm for those two lenses on that camera. That's a pretty small gap of just 10% when the real gap is 50% - small enough that it probably fits within their error bounds.

--
Lee Jay
(see profile for equipment)

Having both the xsi and a 50D it is amazing how good the 50D is and yes you can get away with lesser lenses,,, but do you want to.... to shoot indoors you still need F4 or better, for sports you need F4 or better,

The kit or non usm lenses are slow so you miss shots, thay are dark and suck in doors.....

Based just on resolution the 18-55is is the only lens you would ever need in this range... I used mine only once or twice and it does take good pics but you'll hate useing it....

So chep lense are just chep even on the 50D, But a good usm lens might be good enough for most... I have the 50D with the 18-200 and it takes really good pics I was amazed that I liked the lens that much.. I put it on my xsi the next highest resolution canon and I'm still not sure if I like that combination, the xsi is much better with L glass..... and the 50D is amazing with L glass...

So yes the 50D will make up for bad glass to a degree...........

"nt" means no text.

Joe Kurkjian

Galleries: http://www.pbase.com/jkurkjia



SEARCHING FOR A BETTER SELF PORTRAIT