CI Photocommunity

Register a free account now!

If you are registered, you get access to the members only section, can participate in the buy & sell second hand forum and last but not least you can reserve your preferred username before someone else takes it.

Does MTF describe a lenses accuratly

benwang

Well-Known Member
I went to photodo.com and they have tested a lot of lenses for their MTF and rate these lenses accordingly. However, in some case, although according to the MTF, some of the lenses are equivalent, but there is a different between them, the stand out lenses are usually the ziesis or leica. So, I want to ask everybody here, if the MTF can not describe the lenses properties 100%, what are the different between these lenses? What are the properties that is not objectively covered and how we can describe them or measure them accordingly?
 

rbudding

Well-Known Member
I suggest that you look at some threads on this site about Bokeh, which is the characteristic of the out-of-focus areas(or try
Please, Log in or Register to view URLs content!
). Zeiss glass generally has beautiful Bokeh. But I don't know of a way to measure this characteristic. BUt I sure do know it when I see it!
 
W

wojo

Robert is right on tack about the bokeh in Zeiss glass, at least those I've used. Nice, smooth transitions from sharp to out of focus, no geometric or chromatic distortions in the OOF area.
If you shoot color slides, there's a definite warmer cast to the images, not very important with scans, you can color correct to your hearts content there, but significant with projected images.
Another area that doesn't seem to be measured, or reported in photodo's tests is flare resistance. The Zeiss coatings seem to suppress flare very well, though I can't say whether Nikon or Canon are any different, not having done any side by side tests.
Perhaps It's related to flare, but ghosting seems to be less a problem with the CZ glass.
The sharpest lens isn't necessarily the best for your purpose. Doing portraits, are you sure you want to have every pore and tiny wrinkle rendered with maximum contrast and clarity? While high contrast is a plus in some situations, it can be less than useful in others. I don't know that contrast can be interpreted from the MTF curves. Perhaps if there were a graphical rendition of the output when plotted against a spectral sweep, there may be peaks or valleys which would explain the higher or lower contrast in a lens system. I haven't seen any such, nor do I know if it's possible, but thats really a bit off topic...
JW
 

bollefen

New Member
re bokeh

at the risk of creating conflict i'll use nikon lenses which in the past wer over corrected (imho) for spherical abberation, this gave exceptional lpm at high contrast at the expense of bokeh and out of focus images where a windowpane dividers in the background would actually image a double image of the dividing strip compared to a cz or canon or leitz which would render a blurred divider. these images were available on the web years ago but i don't recall where. i have no experience with current nikon lenses and bokeh is a metter of personal taste.

re color/warmth again this is a matter of lens/coating design and then personal taste.

i suspect these nuances are what really drove folks to be a nikon/canon/leica/zeiss fanatic over the years.
 

benwang

Well-Known Member
I agree with you Robert that the Bokeh is great for the ziess and Leica, and IMO, this may beccause these lenses can capture more colour comparitively. in terms of MTF, although it offer similar, however the image appear sharper, this is because there are more colour. Any opinion?
 

benwang

Well-Known Member
Dale, MTF is a method to measure how sharp the camera lens is. What they do is they will take pictures with the lens testing against some very fine arranged lines (cotton thread?). They will take these picture with different distance and F-stop. So, the sharper the lens, the more accountable of these lines and hences you can objectively compare the lens side by side. And as long as you have follow the standard operating procedure, the test is objective!

"bokeh" is the first time that I heard this word, but going to the link Robert has provided, my understanding is basically the out-of-forcus area.

What I was thinking is that if it was the colour reproduction that Zeiss or Leica is differ from the other brand (canon, nikon e.t.c), isn't it that we could use the MTF as an ex&le, and construct an experiment similar to MTF and compare these lens side by side?

If we could establish a print with "true" colour, i.e.all the colour available, picture it with the comparing lens, and use the computer to count the colour and compare, will it work? is that objective?

Ben
 
M

mike_nunan

Ben,

The test you describe is closer to an outright resolution test than an MTF test. The point of MTF testing is that it measures the contrast of a lens at a number of different spatial frequencies (spacings of the test pattern, IOW). If you look at the figure for 10 line-pairs-per-mm then you get a good idea of the contrast of the lens, while at 40lpmm a high figure indicates good resolving power for fine details. The old-style resolution test sets out to establish the finest spacing that the lens can pull any meaningful information out of, which is quite a different thing and isn't really as useful for assessing the real-world perforamance.

Your idea for comparing the colour rendition of different lenses is likely to run into trouble because there is no imaging system (based either on film or direct digital capture) that can faithfully portray the full range or gamut of colours that we perceive. Even for the colours that can be reproduced, there is always a bit of colour shift introduced by shortcomings of sensors and output devices. Colour profiling is supposed to compensate for these factors, but it's not perfect. Don't let that stop you from thinking about better ways of solving the problem though, you'll be rich and famous if you do, and a lot of photographers and reprographic people will warmly thank you! If you want to read up on the intricacies of colour theory, a good place to start is
Please, Log in or Register to view URLs content!
, which is the website of the International Color Consortium who define the worldwide standards for colour processing and reproduction. Be warned though, it's a deep subject.

Regarding your original question, there are many other criteria beside bokeh and basic MTF measurements that can be used to characterise a lens. If you really wanted a full picture of a lens's performance, you'd need to know the following (others please feel free to add anything I've forgotten)...

* MTF at all f-stops and across various focusing distances from infinity to closest
* Contour map of the plane of focus at those distances and for each f-stop
* Measurement of flare characteristics of the lens under a broad selection of lighting conditions
* Light transmission properties of the lens
* Measurement of light fall-off effect at all f-stops
* Measurement of field distortion at all f-stops

In practice it's not easy to achieve this. The multiple MTF measurements are too costly to gather and produce an inordinate amount of raw data. There is no agreed way of quantifying the flare characteristics of a lens, and it's hard to see how this could be tackled well. Bokeh and colour rendition are hard to quantify objectively, for reasons already discussed. Note that because no lens has a perfectly regular plane of focus (and the true shape is not easy to measure), this can create anomalies in the MTF results. Most testers set focus for the centre of the field, then measure the whole field at that setting. If the plane of focus is strongly curved, then the edges will record a very soft image because they are not in correct focus. Similarly, some lenses experience a shift of focus as the aperture changes, which can lead to poor apparent performance at intermediate apertures. The Leica Noctilux f/1.0 50mm lens is notorious for this, for ex&le.

Ultimately there is no substitute for a real-world test of the lens you are using or thinking of buying. This rules out s&le variation and allows you to make some kind of assessment of some of the factors such as bokeh that cannot be easily measured on a bench. Over the years I have found myself choosing lenses based upon the recommendation of people I respect and thorough research of the body of opinion on the internet and in paper publications. I'd love to think that one day all the guesswork could be done away with, but given the complexity of the problem that doesn't seem likely anytime soon.

HTH

-= mike =
 

afranklin

Well-Known Member
> "bokeh" is the first time that I heard this word, but going to the > link Robert has provided, my understanding is basically the > out-of-forcus area.

Bokeh describes the characteristics of the out of focus areas of the image. What your criteria is for good or bad bokeh is, are really up to you...but most people prefer smooth OOF area transitions, instead of abrupt, almost vibrating.

Please, Log in or Register to view URLs content!


Regards,

Austin
 

benwang

Well-Known Member
Mike

I agree with you that the colour theory is a deep subject. However, IMO that if we just forget about the theory, conduct the experiment, run it and see what happen.
Since no body can count these colour, we have to relie on computer software to do the counting. So as long as we are consistance with equiptment and software choices, then the test can be conduct fairily and objectively. and if there is something different can be pick up, hopefully it can pick up!

Ben
 
M

mike_nunan

Hi Ben,

Thinking about it, if your goal is to find out the colour balance of the lens then there are some practical options. Starting with a white incandescent light source of a known spectrum, you could measure the spectral distribution of the resulting image. That would give you a chart showing the amount of absorbtion that occurs across the visible frequency range, which in turn could be used to work out the effective white point for a given type of film or sensor. The equipment necessary to measure the spectral distribution isn't commonplace however.

To make things easier, you could assume a fairly smooth and compatible response for the different film/sensor types, allowing you to work out a single white point that would hold reasonably well for most conditions. One way to measure this would be to use a DSLR to take a shot of a calibrated white card under controlled lighting then look at the numbers in the RAW file. This would certainly tell you whether the lens was a bit warm/cool/yellow etc.

-= mike =-
 

rbudding

Well-Known Member
Good discussion. But the reason I went with Zeiss lenses was the quality of photos they produce. But it's difficult to find any one measure that captures all of a lens's attributes. So, if you're not sure what lens system to buy, you should spend time looking at photos. And the gallery section of this site has some fabulous ex&les of what CZ lenses can produce.
 

benwang

Well-Known Member
Mike, firstly, must thank you for pointing at the right direction in reading some of the artical about MTF, that something good to know. I know more about it (can't say that I understand it, guess with time will understand more).

I do agree with you that firstly produceing the standard white light will be technically difficult{impossible). Measuring these white light, and produce some readable table will be nightmare! I don't know if this is true or not, but the more we know, the more lost it seems to be (what the acient chinese use to say).

So what I am suggesting is more simple (don't know if it will work though), what we need to do is to get the computer and print a "standard" color referece (with all 64 bit or 98 bit) colour. Use the lens to take photo of the print (with a standard film), scan the film with a standard film scaner, and read it with a computer and see how many different colour captured by the lens. Repeat for a few times and get the mean. Collect data from different brand of lenses and compare. Will it works? All of these hardware are available, it is the colour counting software I am worry about (it may take forever to read with 64 bit colour!)

Using a DSLR may be another option, however, mounting all the lenses will be technically very chanlenging, and may cost you a lot of money.

Robert, there is no doult in the past Zeiss or leica produce the best lenses! However, and IMHO it true, according to the MTF, the other brand are catching up. looking at the MTF of Canon, the mean of their score in MTF chart are very close compare to CZ or Leica (sometime better). So if CZ/Leica's lenses do offer something extra, I think it is important that we could describe it, and let other to know! By using subjective word such as great, stunning, fabulous to describe it is simpily not enough! or otherwise, in the very near future (comming very fast! things do change fast these day!), more people will just use MTF to judge the lenses quilities and we will have no more CZ/Leica!
 

dirk

CI-Founder
Please bear in mind that Zeiss is the only brand that is measuring the MTF in during the producting of the lenses. All others (Leica, Canon etc.) are publishing MTF data from s&les/ prototypes, which are not mandatory the same if if you start to produce them in large quantities.

So what you will experience is, that if you buy a Zeiss lens in Tokyo and the same one in NY, you will have two lenses with the same MTF's (within specific, very tight tolerances).

If you do the same for any other brand, you will see bigger differences. This is why tests in photo journals are meaningless, if the lenses are perselected and handchoosen by the manufacturer and then sent to the test team.

See for more details the Zeiss articles in our manuals section I have mentioned before.
 

rbudding

Well-Known Member
Ben,

There are some things that are difficult to capture in a measurement. MTF is only one measure of a lens, and probably not the most meaningful, unless your only concern is resolving power. But, who knows. Maybe some clever person will figure out how to measure Bokeh.
 
Top