Rainer wrote:
Very interesting observation. Does anyone happen to know how it is physically possible for a lens to increase contrast? I can easily imagine that this is possible at the sensor level, but I'm clueless about how a lens could do that. Any input is very much appreciated!
In order to increase contrast a lens should either selectively amplify highlights or selectively absorb shadows. Non of these is happening in any ordinary lens (not even in a Zeiss ). On the other hand any real lens would lower the contrast. It happens because some light reflects from diffrent lens elements and ends up in a wrong place. It is flare, and I would be really surprised and delighted to learn that it is actually beneficial.
Cogitech, could you demostrate the effect or point to some information about it?
Well, what I read in that thread, so far, seems to indicate that I was wrong. Well, half-wrong, anyway. It seems that the lower contrast lenses *do* tend to reduce the contrast because of "stray light". It is also explained that this is because there seems to be a trade-off between having a lens with high resolution or a lens with high contrast. Also, it seems that lens coatings *can* increase the contrast of a lens, which gives the appearance of "sharpness", but often true resolution is sacrificed and fine tonal gradations can be lost.
Read the whole thread for the full story, but what it comes down to is some would rather have the resolution and finer tonal gradations and others opt for the instant gratification and "pop" of the contrasty lenses. In the age of digital, I know which one I prefer.
Yes, that was a very interesting thread. And Paul, you might still be right: could it be that the sensors and image processors are optimized such that they render "perfect" (whatever that means) images with medium contrast lenses? In that case, high contrast lenses would actually yield too much contrast with those sensors, giving an unnatural look. Please keep in mind that this is just a wild theory, but it would explain Paul's observations.
Well, there certainly was a move back in the film days towards optimizing lenses for contrast (to achieve the "look" of sharpness). Apparently Nikon and Canon started the trend and others (like Leica) followed. This may have produced images with more "pop" right out of the camera than the older lenses (which makes sense for the average film shooter), but those really black blacks and white whites (and reduction in tonal gradations) of the ultra-contrasty lenses seems to me to be inappropriate in the digital realm.
I also want to re-iterate what I said about dynamic range. If medium to lower contrast lenses produce an image circle with less black blacks and less white whites, but also finer gradations in all the mid-tones, wouldn't this give the metering system and sensor the opportunity to expose all parts of the image more evenly and reduce the likelyhood of "black" shadows and "blown" highlights? Maybe this is the answer to the dynamic range "problem". Maybe the "problem" is that the lenses are throwing way more dynamic range than necessary (and not enough fine tonal gradations in the mids) at the sensor.
Again, contrast can be added easily, if necessary, but adding resolution and/or finer tonal gradations is next to impossible.
Edited by cogitech on Feb 09, 2007 at 10:21 AM GMT
Paul, I tend to agree with your observations and came to the same conclusions after reading both that thread and before, Sean Reids observations. Sean Reids site is a pay subscription site so I will not link to what cannot be read for free but he has demonstrated this in an article published over at LL here: http://www.luminous-landscape.com/reviews/lenses/fastlensreview.shtml
Look towards the bottom for "Sunny Day lenses" and "Preserving Detail...." which gives side by side illustrations of this idea.
Sean explains it much more eloquently than I do, but I've surprised myself by noticing this on my own.
What he says about the dynamic range being presented to the sensor and the "milder" image of the lower contrast lenses is exactly what I was trying to get at.
I also agree that there are artistic uses for both types of lenses.
Thanks for the link. Much appreciated.
BTW, hubsand, sorry for inadvertently highjacking the thread. I am sure we are all still very interested to hear more about the original topic.
So-called "high-contrast" lenes are not so much enhancing scene contrast (as is being implied) but rather accurately representing it. The older lenses in Sean Reid's review do not adequately trnasmit actual contrast -- they tend to reduce it. It's just old technology. Single-coated lenses, like many of these older lenses, reduce contrast relative to more modern multi-coated lenses (light transmission is reduced, flare is increased, etc. (essentially adding a slight haze to the image). Now, as Sean Reid was suggesting, you can use these older, contrast-reducing, lenses selectively to help control high contrast scenes that will be difficult for the sensor to adequately capture. He was not suggesting that these lower contrast lenses be used in all circumstances as a matter of course.
Now, I know that very few of you actually believe or trust precision measured MTF charts (one wonders why they bother producing them!). But you will note from these MTF's that no lens fully can represent actual contrast. The top lines in the MTF charts (10 lp/mm, or in some cases, 5 lp/mm) represents the contrast of the lens. In none of these charts, do these line ever exceed 100% -- they are all less than perfect at conveying actual contrast. The so-called "high contrast" lenses, just do it better than lesser lenses, but do not in any way increase or enhance contrast.
One interesting point which sort of ties this whole lens contrast question back into the original thread is that the same manufacturer, Cosina, who make many of the Zeiss ZF lenses(a high contrast lens line) also make the lower contrast, highly regarded Voigtlander branded rangefinder lenses, many of which compare very favorably to Leica M lenses. Furthermore, one of the lenses they also make, the 40mm 1.4 Nokton is offered in both a Single Coated limited edition version(and they state this may be preferable for B&W) and as a multi coated version. I don't think its too far of a strethch to conclude that a lens designer today has very strict control of the look and contrast of a given lens line throgh all the tools and options available to them including coatings. The way a lens should render a given scene is obviously a very subjective thing.
I agree (about subjectivity). That being said, should anyone really be using macro-contrast as a determining factor of what makes a lens "good" or better than another lens? Should we not simply be commenting on the "type" of contrast that a lens offers, rather than passing judgement and/or dismissing medium and low contrast lenses as being somehow inferior. The mainstream certainly tends to be critical of lenses with less-than-shocking contrast and the majority of lenses are made to be as contrasty as possible. In actual fact, these lenses are just different and they have their pros and cons.
I, for one, have recently experienced a paradigm shift with regards to all this. No longer will I see remarks such as "lacks contrast" as a fault, but rather simply one of the qualities of the lens. This is not to say that I'll be seeking out only low contrast lenses. I love that 3D Zeiss T* look too, but for different images it would be nice to have various "contrast profiles" to choose from.
I am also curious about the relationship between contrast and MTF graphs. I had always assumed that MTF was a measure of resolution only, but I now understand that contrast is integral to the calculations. It has been stated (in the various threads and articles) that it is possible to have high resolution but low contrast. Where would this be reflected in an MTF graph? If it is not visible on an MTF graph, then MTF graph data is always inherently related to the lens contrast, and MTF graphs are of much more limited use than I had believed before.
In other words, despite the resolving power, lower contrast lenses would appear to have lower MTF values than higher contrast lenses. Is this correct? If so, I'll be taking MTF graphs with a bag of salt from now on.
cogitech wrote:
I am also curious about the relationship between contrast and MTF graphs. I had always assumed that MTF was a measure of resolution only, but I now understand that contrast is integral to the calculations. It has been stated (in the various threads and articles) that it is possible to have high resolution but low contrast. Where would this be reflected in an MTF graph? If it is not visible on an MTF graph, then MTF graph data is always inherently related to the lens contrast, and MTF graphs are of much more limited use than I had believed before.
In other words, despite the resolving power, lower contrast lenses would appear to have lower MTF values than higher contrast lenses. Is this correct? If so, I'll be taking MTF graphs with a bag of salt from now on....Show more →
Very good point and question. I'm curious as well about this.
cogitech wrote:
I am also curious about the relationship between contrast and MTF graphs. I had always assumed that MTF was a measure of resolution only, but I now understand that contrast is integral to the calculations. It has been stated (in the various threads and articles) that it is possible to have high resolution but low contrast. Where would this be reflected in an MTF graph? If it is not visible on an MTF graph, then MTF graph data is always inherently related to the lens contrast, and MTF graphs are of much more limited use than I had believed before.
In other words, despite the resolving power, lower contrast lenses would appear to have lower MTF values than higher contrast lenses. Is this correct? If so, I'll be taking MTF graphs with a bag of salt from now on....Show more →
The MTF graph actually shows the contrast of a lens at a given resolution. This link might be helpful. Canon provides MTF graphs for resolutions of 10 and 30 LP / mm, Zeiss for 10, 20 and 40 LP / mm.
To see what the actual resolution of a lens is, you would need MTF curves that show contrast vs. resolution (and we'd have to abandon the dependence on the distance from the center of the frame, or make a 3D-graph). You would find the resolution where the contrast drops to zero, or a very small number.
I'm going to attempt my 135mm test (which I messed up last time) again this weekend and while I'm at it, I'll try to get some shots that illustrate what we've been on about here.
I plan to use the Zuiko 21/3.5 (lower contrast) and Zeiss 28/2.8 (higher contrast). Any input from you guys would be welcome.
I guess I aught to give hubsand his thread back and start a new one if/when I try this.
Paul:
IMHO, if you want to make your case you should
1. In both cases select your exposer for the highligts
2. Took pictures in RAW
3. Using your best conversion and PS technique demonstrate that the low contrast lens delivers better detail in the shadows.
Thanks Alex. I'll take your advice and I'll also do some tests exposing for the shadows, to see if I get less (or more?) blown highlights with the lower contrast lens.
Here's hoping that Toronto's weather will be favourable tomorrow.
). On the other hand any real lens would lower the contrast. It happens because some light reflects from diffrent lens elements and ends up in a wrong place. It is flare, and I would be really surprised and delighted to learn that it is actually beneficial. 