Modulation Transfer Function...What does it mean

2

Comments

  • AdeAde Posts: 1,071Member
    You're fixated on one kind of MTF charting.

    What's more relevant to the original discussion (resolution) is actually a different kind of MTF charts (MTF50 charts) -- such as ones produced by Imatest. They're not about sharpness "from center to corner", and they look nothing like the charts in the examples above.
  • TaoTeJaredTaoTeJared Posts: 1,306Member
    edited June 2013
    Ade - I get you want to talk about something else, but the topic is:
    "Modulation Transfer Function...What does it mean"
    Ade you are trying to send the discussion into another realm and off topic - one that only includes Imatest who does go beyond the standard MTF and dives into sub categories. Those are not "MTF" charts but they do fall under various sub categories of measuring specific things that make up the MTF. Using the car analogy, (I can't believe I'm using it) MTF is like a measurement of Horse power and torque. Imatest's "other" charts show the other things as an equivalent of gear ratios, bore size, air mixture, etc. on an engine.

    I like Imatest and I think their results are better, but that yet again should be it's own thread as Nikon, Sigma, Tamron, Canon, Zeiss etc. do not use them for tests. The only time people will run into Imatest's charts is if a lens just happens to be tested by someone with that software. I wish more were, but that is just not the case thus it needs to be separate of the general MTF.

    I'm taking in good faith of the exact title; "What does it mean"

    Understanding MTFs is nasty, made for engineers, difficult, (and in my humble opinion - mostly useless.) At best for photographers they are merely for "generalizing" performance of same brand lenses at similar focal lengths.

    A couple of Key points:
    1. You can not compare MTF from different manufactures. For example you can not take a 50mm Nikkor lens MFT (from Nikon) and compare it to the MTF chart that Sigma releases of their 50mm. It is an apples to oranges comparison.

    2. MTFs are not resolution tests - they only measure sharpness which is defined as the ability of recreating absolute contrast between white and black on a target.

    3. If comparing, it needs to be at a similar focal length. I.e. 24-70 vs 24-120 at the 24mm end.

    I have had this video bookmarked forever - even though the sound is out of sync, he does a good job of simplifying it.

    Post edited by TaoTeJared on
    D800, D300, D50(ir converted), FujiX100, Canon G11, Olympus TG2. Nikon lenses - 24mm 2.8, 35mm 1.8, (5 in all)50mm, 60mm, 85mm 1.8, 105vr, 105 f2.5, 180mm 2.8, 70-200vr1, 24-120vr f4. Tokina 12-24mm, 16-28mm, 28-70mm (angenieux design), 300mm f2.8. Sigma 15mm fisheye. Voigtlander R2 (olive) & R2a, Voigt 35mm 2.5, Zeiss 50mm f/2, Leica 90mm f/4. I know I missed something...
  • TaoTeJaredTaoTeJared Posts: 1,306Member
    edited June 2013
    I'll take a stab at this. This is my simple down and dirty way of looking at these. (I slapped a couple of example images together)

    image

    1. The top line represents stark contrast in large detail. (example: a large black box in front of a white wall.) Top line should be above 80 and remain close-ish to the same height till the top/bottom edge (12mm).
    Dashed line represents "fuzziness" of the edge. Two lines should be close to each other.
    2. This is the Mid tone line. Not always included. Same as above, but for most details. Example: iris vs white of eye.
    Dashed line represents "fuzziness" of the edge. Two lines should be close to each other in the center but begin to spread apart.
    3. Finest detail resolved. Example: distinguishing separate hairs in the eyebrow. The higher the number measured (lp/mm) for the finer detail. Nikon's standard is 30lp/mm.
    Dashed line represents "fuzziness" of the edge. The closer the sharper but expect to be more distance apart.

    I always think of the word "Sagittal" as the basic structures (a face) and the "Tangential" as the CAs (wider deviation from a solid line) or crispness of that line. Some refer to it as micro contrast.

    I'm sure some will want to nit pick as usual and redefine terms and move the direction off of what MTF is, but this is the basics of it that is USABLE.

    Much of the reason there is so few voices on this thread as others, is when some nit-pick details of someone's post and don't actually add anything that is usable to the general photographer. Engineering descriptions are fine and dandy for engineers, but this is a photography site and there is an overabundance of engineering descriptions out there that no one can understand and few care to read. I don't think anyone came here for that.

    I challenge everyone who posts to make it about transferring what they see on a MTF chart into what it does for the image.
    Post edited by TaoTeJared on
    D800, D300, D50(ir converted), FujiX100, Canon G11, Olympus TG2. Nikon lenses - 24mm 2.8, 35mm 1.8, (5 in all)50mm, 60mm, 85mm 1.8, 105vr, 105 f2.5, 180mm 2.8, 70-200vr1, 24-120vr f4. Tokina 12-24mm, 16-28mm, 28-70mm (angenieux design), 300mm f2.8. Sigma 15mm fisheye. Voigtlander R2 (olive) & R2a, Voigt 35mm 2.5, Zeiss 50mm f/2, Leica 90mm f/4. I know I missed something...
  • FlowtographyBerlinFlowtographyBerlin Posts: 477Member
    They were painful to read. Analogies are good if they get to the point quickly, and it doesn't require a bunch of pre-existing knowledge of another topic just to understand the original topic. ;)
    Where exactly did you have the feeling you were lacking pre-existing knowledge on the topic of the speed of a car?
  • SquamishPhotoSquamishPhoto Posts: 608Member
    What a boring thread.
    Mike
    D3 • D750 • 14-24mm f2.8 • 35mm f1.4A • PC-E 45mm f2.8 • 50mm f1.8G • AF-D 85mm f1.4 • ZF.2 100mm f2 • 200mm f2 VR2
  • AdeAde Posts: 1,071Member
    edited June 2013
    Everyday, we use things without knowing how they really work. And that's by design. To drive a car -- using TTJ's favorite analogy -- we don't have to know the details about internal combustion engines; just put gas in the tank and off we go.

    But people are curious beings. So some of us would like to understand more about the things we use and see and touch everyday.

    There are no limits to knowledge. Going back to MTF, most photographers wont even recognize the abbreviation. Others may have learned to "decipher" a basic MTF chart to quantify whether a particular lens is "good" or "bad", but without really understanding what MTF is all about.

    I had hoped that this thread might help photographers understand MTF, not how to read some chart. After all, there are many different kinds of MTF charts. And without that deeper knowledge, we get the kind of misleading conclusions that gets repeated so many times in forums that many begin to think they are true. There are statements in this very thread, for example, which are just utter nonsense. Only with an open mind, can we learn new things.
    Post edited by Ade on
  • IronheartIronheart Posts: 3,017Moderator
    I already understood how to read a two axis graph, with contrast on the Y and distance from center on the X. I also understood that measurements are taken at different frequencies (10 and 30 lp/mm for Nikon) and at right angles to each other (sagittal vs. meridional).
    What I've learned from this thread and the related materials is why. The low frequency is used to measure contrast and the higher frequency used for resolution (a.k.a. micro-contrast or sharpness). Now before someone jumps on me I understand that absolute resolution measurement is not possible, (see my first post on this thread), but relative is. Thanks.
  • JJ_SOJJ_SO Posts: 1,158Member
    I don't think I understand MTF curves fully, but I know a bit more about what they are not.

    Thanks Ade from me, too.
  • IronheartIronheart Posts: 3,017Moderator
    And TTJ, and FlowtographyB, and the rest for keeping it civil, straightforward, and on-topic :-)
  • TaoTeJaredTaoTeJared Posts: 1,306Member
    Off our NR front page (quite a while ago) the Nikkor 50mm MTFs.

    image

    I have been using Leica MTF (mainly because they were clearer) but it's good to see a comparison of the Nikon 50's.

    If you look at the old "D" lenses, (right side) the thing that pops out is the dashed lines being further away from the solid. This would mean less edge, I'll use the term crispness. They are sharp lenses but are known for CAs. This is one of the items that spreads those two lines apart. You really can see it on the 50mm 1.4 D where the dashed line is wavy from the get go on the lower, higher detail line. I own the lens, and yes, CAs are noticeable.

    On the 50 G lenses (left side) you can see a noticeable difference - and for good reason, they are designed for a different goal. The first thing that comes at you is that the 1.8G is sharper than the 1.4 throughout the frame. Something that everyone has written about is how the 1.8g is sharper. (Remember at 12mm (bottom axis) we are at the top and bottom of the image.) Now the dashed line on the 1.8 is close to the solid line out to about 18mm (left and right side of image) which indicates a sharper/crisper image across the frame.

    Now the 1.4g (designed for bokeh) the solid line drops quickly from the center (about the center quarter of the frame is very sharp and crisp) and the separation of the dashed line grows quickly. The top line is the larger structure detail - that falls off and the dashed line separates quickly = Bokehlicious. Now the bottom (blue) dashed line is close to the solid line - that would indicate CAs are held better and micro contrast is crisper. (Micro contrast is that "pop" or 3d look people are always talking about.)

    D800, D300, D50(ir converted), FujiX100, Canon G11, Olympus TG2. Nikon lenses - 24mm 2.8, 35mm 1.8, (5 in all)50mm, 60mm, 85mm 1.8, 105vr, 105 f2.5, 180mm 2.8, 70-200vr1, 24-120vr f4. Tokina 12-24mm, 16-28mm, 28-70mm (angenieux design), 300mm f2.8. Sigma 15mm fisheye. Voigtlander R2 (olive) & R2a, Voigt 35mm 2.5, Zeiss 50mm f/2, Leica 90mm f/4. I know I missed something...
  • AdeAde Posts: 1,071Member
    edited June 2013
    Again, at this point I would strongly recommend everyone to focus on the fundamentals of what is MTF -- as we've been discussing -- and not get distracted by the mechanics of chart reading.

    On Internet forums, reading MTFs is like reading tea leaves. We "see" things which are just not there. If enough "experts" keep saying the same things, we begin to believe them, even if there is no scientific basis to what they are saying.

    In particular, the MTF charts above say nothing about chromatic aberration.

    How could it? Chromatic aberration is caused by different wavelengths of light falling in different positions. But in the MTF testing we've been discussing so far, we never tested different wavelengths of light. We only measured contrast at different distances from center, f/stops, and resolutions.

    There is a different type of MTF chart which plots MTF vs. wavelength. As far as I'm aware, Nikon / Canon have never published those kinds of MTF charts.
    Post edited by Ade on
  • spraynprayspraynpray Posts: 6,545Moderator
    I took it that TTJ was implying that when MTF curves have the wavy line, the lens that they are representing has bad CA. I didn't read him saying that the lines were the result of a test to measure CA. :-?
    Always learning.
  • MsmotoMsmoto Posts: 5,398Moderator
    This thread has a lot of information. But, one issue is the semantics involved in any discussion which is at best confusing to most and at worst, just a bunch of strange wavy lines.

    The issue of CA is one I actually had not given much thought to but is relevant and if the CA is demonstrated by the divergence of the sagittal vs. tangential lines...how interesting. I noticed on two lenses from the 1960's which I had modified so as to be able to try these on my D4, that while they were sharp, the CA was rather horrid.

    Anyway, thanks for all the contributions on this thread......maybe for me TMI...LOL
    Msmoto, mod
  • AdeAde Posts: 1,071Member
    edited June 2013
    Wavy lines, straight lines, lines close together, lines separating, etc. -- none of them directly imply CA. A lens with any of those characteristics might have CA or might not have CA; we can't really tell from the usual MTF charts.

    To determine CA, we have to plot the behavior of each wavelength (red, green, blue) separately. CA is present if the plot of the the three color curves diverges. CA can be defined as the sum of the distances between the curves (the area/integral between the curves).
    Post edited by Ade on
  • SquamishPhotoSquamishPhoto Posts: 608Member
    Will understanding this nonsense help me improve my photography in any way?...



    ...didn't think so.
    Mike
    D3 • D750 • 14-24mm f2.8 • 35mm f1.4A • PC-E 45mm f2.8 • 50mm f1.8G • AF-D 85mm f1.4 • ZF.2 100mm f2 • 200mm f2 VR2
  • spraynprayspraynpray Posts: 6,545Moderator
    +1 Msmoto and SquamishPhoto =))
    Always learning.
  • AdeAde Posts: 1,071Member
    Understanding MTF is a tool, much in the same way a cable release is a tool.

    Can MTF (or a cable release) make you take more interesting images, with better subjects, composition, lighting, etc? No.

    Can MTF (or a cable release) help you take the sharpest possible image? Yes.

    Readers of this site often give much weight to recommendations from people like Thom Hogan and Roger Cicala. But how do they reach their conclusions? In part, by running their own MTF tests to add objectivity to their subjective experiences. So as a community, we all benefit from MTF testing.

    There are, of course, some types of photography where knowing MTF or its equivalents becomes critical. Microscopy, medical / scientific imaging, etc.

    In general, knowing MTF helps you better understand the equipment which you use. If nothing else, some of us would rather be in a position of knowledge, rather than in a position of ignorance.
  • TaoTeJaredTaoTeJared Posts: 1,306Member
    This thread has a lot of information. But, one issue is the semantics involved in any discussion which is at best confusing to most and at worst, just a bunch of strange wavy lines.

    The issue of CA is one I actually had not given much thought to but is relevant and if the CA is demonstrated by the divergence of the sagittal vs. tangential lines...how interesting. I noticed on two lenses from the 1960's which I had modified so as to be able to try these on my D4, that while they were sharp, the CA was rather horrid.

    Anyway, thanks for all the contributions on this thread......maybe for me TMI...LOL
    My really old lenses are terrible as well - but with B&W (primary film - kind of) they were not an issue.
    Will understanding this nonsense help me improve my photography in any way?...
    ...didn't think so.
    +1


    D800, D300, D50(ir converted), FujiX100, Canon G11, Olympus TG2. Nikon lenses - 24mm 2.8, 35mm 1.8, (5 in all)50mm, 60mm, 85mm 1.8, 105vr, 105 f2.5, 180mm 2.8, 70-200vr1, 24-120vr f4. Tokina 12-24mm, 16-28mm, 28-70mm (angenieux design), 300mm f2.8. Sigma 15mm fisheye. Voigtlander R2 (olive) & R2a, Voigt 35mm 2.5, Zeiss 50mm f/2, Leica 90mm f/4. I know I missed something...
  • TaoTeJaredTaoTeJared Posts: 1,306Member
    edited June 2013
    There is only ONE MTF chart. Sometimes they have two lines, some even have four that measure the same thing, but it detail positions.
    This is an MTF chart:
    image
    105mm VR macro

    There are many other charts that help support the MTF chart and break measurements down to other items like CAs, lens Aberrations, diffraction, Log Frequency and the like. It is incorrect to say there are different multiple MTF charts. The additional charts are specific to aspects that the MTF chart does not directly show.

    Indicate: Be a sign or symptom of; strongly imply.
    The High detail lines of a MTF chart (solid and dashed) separate that can indicate CAs. I have experienced that, and my experience is also confirmed by Zeiss, Imatest, Dxo and everyone else. There was no absolutism in my statement above.


    @Ade - It is inappropriate to state people are ignorant just because they do not want to have the same discussion as you or for you to add incorrect meanings or statements to people's posts.
    Post edited by TaoTeJared on
    D800, D300, D50(ir converted), FujiX100, Canon G11, Olympus TG2. Nikon lenses - 24mm 2.8, 35mm 1.8, (5 in all)50mm, 60mm, 85mm 1.8, 105vr, 105 f2.5, 180mm 2.8, 70-200vr1, 24-120vr f4. Tokina 12-24mm, 16-28mm, 28-70mm (angenieux design), 300mm f2.8. Sigma 15mm fisheye. Voigtlander R2 (olive) & R2a, Voigt 35mm 2.5, Zeiss 50mm f/2, Leica 90mm f/4. I know I missed something...
  • AdeAde Posts: 1,071Member
    Repeatedly stating and insisting on things like "there is only ONE MTF chart" when presented with ample evidence to the contrary is ignorance.

    I can off the top of my head think of five different MTF charts. Yes, the all measure MTF, not "sub-categories" or whatever.

    MTF is used for many purposes, and so various MTF charts are used to show different things.
  • TaoTeJaredTaoTeJared Posts: 1,306Member
    edited June 2013
    So I have a question... I've read that MTF charts aren't comparable across lenses of different focal lengths; for example, you can't compare a wide angle lens MTF to a medium telephoto or super telephoto. I'm not exactly sure why this is. I mean you measure lines and the like, right? ...

    ...Now why can't I just compare this to a say 70-200 MTF and let this help me decide if I'll get sharper/more contrasty results?
    (Great photo by the way! I too like longer lenses for portrait work.)

    It seems that you want a "black or white" answer to this - but as many things in life & photography it just doesn't work that way. There is a thin line when using the word "compare" lenses that can get crossed too fast.

    Let's see if I can straddle that line.

    image image
    ---105mm VR macro-----------------------85mm F1.4 G---------

    You can obviously see in real general terms how each lens react. They both have higher contrast, both are sharp. Now the 85mm 1.4G appears to be less sharp. Many testers have found that the 85mm 1.4 is close if not the sharpest lens Nikon has at the moment. (Exclude any debate about that)

    Now this is where things get messy:
    Apertures: Note the 105vr does not state what aperture. The 85mm is measured at 1.4. Inherently as the aperture decreases in size the sharpness and contrast increase (until diffraction).
    Focal lengths: As you noted, wider focal lengths tend to have characteristics that render different looking charts, but so do telephotos - but in the opposite direction.
    Zooms: Zooms just get wonky. It all depends on the tested aperture and tested length. Focus creep/shift can have an effect on results as well.

    There is a whole bunch of other things that can and do have an effect. I don't think anyone suggests to look at it as a "determining factor" but as just a small bit of information to start with. For instance you can look at a MFT from a Leica 50mm 1.4 and a Nikkor 50mm 1.4 and know right away the Leica is much better - but to try to say how much or in what rendering characteristics will show up on an image and to what degree you will see it, is really just a guess. You could look at charts all day but in the end it is just easier and quicker to try the lens out.

    Most I know just look at a MTF as a small factor in any decision about a lens and is usually just the start. I just keep an eye on how close dashed lines are to the solid lines. If Bokeh is the goal, I like to see them wider - to a point (focal length play a much larger role that just that). If razor sharp is the goal, I want them closer.
    Post edited by TaoTeJared on
    D800, D300, D50(ir converted), FujiX100, Canon G11, Olympus TG2. Nikon lenses - 24mm 2.8, 35mm 1.8, (5 in all)50mm, 60mm, 85mm 1.8, 105vr, 105 f2.5, 180mm 2.8, 70-200vr1, 24-120vr f4. Tokina 12-24mm, 16-28mm, 28-70mm (angenieux design), 300mm f2.8. Sigma 15mm fisheye. Voigtlander R2 (olive) & R2a, Voigt 35mm 2.5, Zeiss 50mm f/2, Leica 90mm f/4. I know I missed something...
  • TaoTeJaredTaoTeJared Posts: 1,306Member
    Repeatedly stating and insisting on things like "there is only ONE MTF chart" when presented with ample evidence to the contrary is ignorance.

    I can off the top of my head think of five different MTF charts. Yes, the all measure MTF, not "sub-categories" or whatever.

    MTF is used for many purposes, and so various MTF charts are used to show different things.

    There is only one chart that is the standard across all lens manufactures.
    If you do not understand the basic mathematical idea that you can plot a set of data as a bar chart, pie chart, bubble chart, but the data doesn't change and is showing the exact same thing - well that is your opportunity to learn and search out forums that can help you with that.

    This is getting rather tiresome of you calling people ignorant especially so when you continue to misunderstand various English terms and apply your assumptions and misinterpretations to other's posts. You do this on a continual basis and then you attack individuals continually when you get upset when you misunderstood something or when the conversation is not to your liking.
    D800, D300, D50(ir converted), FujiX100, Canon G11, Olympus TG2. Nikon lenses - 24mm 2.8, 35mm 1.8, (5 in all)50mm, 60mm, 85mm 1.8, 105vr, 105 f2.5, 180mm 2.8, 70-200vr1, 24-120vr f4. Tokina 12-24mm, 16-28mm, 28-70mm (angenieux design), 300mm f2.8. Sigma 15mm fisheye. Voigtlander R2 (olive) & R2a, Voigt 35mm 2.5, Zeiss 50mm f/2, Leica 90mm f/4. I know I missed something...
  • AdeAde Posts: 1,071Member
    @PitchBlack

    You actually got it right from the start. The wider the lens, the more optical problems creep in. Also, telephoto lenses tend to have have smaller max apertures than wide-angle lenses. Hence even a superb wide-angle lens might have trouble matching a mere "decent" telephoto lens.

    So it would be "unfair" to judge wide-angle lenses against the MTF standard set by telephoto lenses, because in such a comparison, all wide-angle lenses tend to appear dismal.

    But if you keep the above caveat in mind, then you can certainly compare across focal lengths.
  • AdeAde Posts: 1,071Member
    @TTJ

    Think what you want. When I have time in the next couple of days I'll post a variety of MTF charts. I think it will be useful for everyone to understand MTF better anyhow, by looking at the various applications of MTF data.

    And TTJ, considering I majored in Math & CS, I think I know my bar graphs vs. pie charts well enough, thank you. I might even know a little more about the math behind MTF calculations than you do.
  • MsmotoMsmoto Posts: 5,398Moderator
    Let's be nice folks.....
    Msmoto, mod
Sign In or Register to comment.