Grey scales and MTF's

[IWC Doppel]

Advertisement (gone after registration)

A thought struck me about lens characterisics and preformance....

 

MTF's 'modular transfer function', effectively contrast for varring graduations of black and white lines (5,10,20,40 per mm). This will be I assume at ideal lighting conditions and we are really looking at the resolving power of the lens (accepting that various abberations are not covered including Chroma)

 

The method is also quite binary in it's approach looking at black/white graduations, it reminds me of ANSI contrast measurements for projectors in some ways.

 

Anyway, I wondered if anyone has every measured lenses for grey scale and graduation from black to white. I bet the 75mm Summilux spends a lot more 'time' in the blacks, I bet the response would not be linear. I wonder if these characteristics would highlight some tonal differences for certain lenses more clearly.

 

Anyway thoughts appreciated

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[adan]

MTF's 'modular transfer function', effectively contrast for varring graduations of black and white lines (5,10,20,40 per mm). This will be I assume at ideal lighting conditions and we are really looking at the resolving power of the lens (accepting that various abberations are not covered including Chroma)

 

Well, that is the theory. But in practice I find MTF charts do NOT tell me much useful about resolution. Or more accurately - practical results often reveal that some lenses with lower MTF readings actually capture more fine detail of the type we think of when mentioning "resolution."

 

I find quite often that older lenses that do not have modern (post-1990) multi-coating resolve finer detail than their MTF charts would suggest - and sometimes more than their modern mutli-coated successors. E.G. 50 Summicron "rigid" or DR from the 1960s; 35 Summicron v.4; 90 Tele-Elmarit.

 

As a rough example, in the attached image, the lower-left reproduction of the "original" would chart higher for MTF than the reproduction top center (higher resolution, but lower contrast). And indeed, if you back up 10 feet (3m) and view your screen from there, the "high MTF" image holds up better.

 

Which is why the industry has adopted MTF charting over separate contrast and lppmm resolution charts as were used pre-1970. For the average consumer looking at 4x6 prints, a high-contrast lens that captures 40lppmm "looks better" than a lens that actually captures 80 lppmm, but at a lower contrast.

 

Since I'm not an "average consumer," I look for something else. As shown on the right, I'll take a low-contrast, high res image because low contrast (top row) can be corrected more readily that low resolution (bottom row).

_______

 

But anyway, on to your core question. You're essentially asking has anyone measured the dynamic range of different lenses, independently of other factors (different sensors, different films, different cameras).

 

Not that I'm aware of, although there is lots of anecdotal experience (including my own) that older, lower-contrast lenses can hold shadows/highlights over a longer range than contrasty modern lenses.

 

I.E. no-one has sat down with a Kodak/Tiffen 19-step gray scale and compared how many steps get recorded by a 75 Summilux vs. a 75 Summicron ASPH, or how they "chart out" on a curve.

[colonel]

In a nutshell, you are saying that the contrast between levels of grey for a particular lens might be transmitted better then between the unicolours black and white ?

 

The only way of showing this would be to have multiple MTF charts produced by a 2 dimensional a grid of grey levels against each other. With, say, 5 grey levels (including black and white) that would make 25 graphs for each aperture

 

The problem I have with this is that different levels of grey are just subsets of the spectrum from black to white, therefore our original back/white single graph should sum up the overall performance of the lens, albeit that it might be distributed in a non-linear way within the spectrum. Showing that non-linearity requires some presentational idea that does not occur to me at the moment, except perhaps for an area graph for each of the 4 sets of 2 curves, i.e. each set of different LP/mm curves would need its own graph with a graduated colour area graph showing pieces of the spectrum. Quite difficult to read I would imagine.

[IWC Doppel]

Adan, hi

 

Most interesting, informed and insightful (as usual). It would be good if a grey scale was added to the lens testing, I have noticed histograms are a lot different for the same shot at the same settings for the same focal length

 

The 'shape' of my 35 Summilux ASPHERICAL is notably different to my, now sold, 35 Summilux ASPH.

 

It would seem that looking at older glass is more interesting than ever with digital, with the ability to 'pull' the black point and white points to hold a full range and perhaps the more interesting tones, sat in the middle now 'opening up'. Add to that the well known design change in the 60's (seems early Mandler era ?) to higher contrast lenses, compromising a little on resolution but gaining on contrast might be a different compromise today if used properly.

 

I also assume that the dynamic range captured might be more, I think that's what you are saying ? Simply because the less contrasty lens compresses the dynamic range but will therefore allow the sensor to record more ?

 

I keep thinking about buying a nice 50 rigid as I suspect they will go up in price over coming years.

[pgk]

Simply because the less contrasty lens compresses the dynamic range but will therefore allow the sensor to record more ?

I think that you have to determine what reduces the contrast and it is almost certainly 'flare' - unwanted light bouncing around inside the lens - which is reducing the lens' ability to record tonal detail, potentially in both highlights and shadows. The idea that a lens can somehow reduce a scene's contrast and yet maintain good tonality throughout a greater contrast ration than a higher contrast lens (one not subject to such flare) does not strike me as a feasible one I'm afraid.

 

I'm not a lens designer so I don't understand the nuances of lens design and the effects of calculations and precision lens build quality, however I think that high quality modern lenses resolve just as great detail (at higher contrast) as did old high quality lenses. Lower quality lenses may be of higher contrast than old lenses but their ability to resolve detail is also dependent on their construction and lower quality construction allied with QC issues may well make their overall contrast high but limit their their resolution.

 

Personally I find the current Leica lenses to be high contrast and of extremely high resolution. Their older counterparts show exactly what one might expect, somewhat lower contrast but high resolution both of which are indicative of the highest quality designs then possible, allied together with high QC in construction and manufacture. But I doubt very much that they are a magical way of increasing recorded tonality.

[IWC Doppel]

Paul,

 

I see your point and understand your arguments, still curious about the advantages of digital recording and what can be acheived in processing vs what is recorded in the first place, possibly Bit depth has an impact too.

 

The conecept in my head relates to having the same information recorded at lower contrast settings for a normal image and so the tonal range is more centrally focussed, would this allow a wider dynamic range being recorded if the image had a higher dynamic range that the sensor (Mechanical compression in some fashion) ?

 

If using lightroom you move the black point and white point you are effectively 'stretching' across the tonal range it would seem. This might be pleasing to the eye with a mid tone prominance, I am also not sure what is seen by the sensor, what is expored raw/jpg regarding gamma curves etc and if any non linearity might also play a part.

 

The reason for lower contrast is a very interesting one and I am sure you are right as to one or the main cause.

 

Having spent years with homecinema projectors and in room light control I am very familiar with reflections and contrast.

 

Just curious about what is going on with lenses like the 75 Sumilux and how it responds to low level RGB values as it paints a truly wonderful, weighty picture on gloomy days