Jump to content

Monochrom histogram divisions: what are they, really?


elgenper

Recommended Posts

Advertisement (gone after registration)

Collected my new Monochrom yesterday: it´s fabulous! :D

 

As many others have commented, the resolution, tonality, and general ´malleability´ of these MM raw files is far, far beyond any B/W conversion from a Bayer sensor camera that I´ve worked with. And the knowldege that now I cannot ´chicken out´ and revert to a colour version of an image that I haven´t got right in B/W is very efficient in enforcing that old-time ´Tri-X discipline´… Also, the similarities in handling to my old friend the M9 is an important benefit, as is the shared set of optics.

 

However, there is one little thing that confuses me a bit, and that is the divisions in the raw histogram (I shoot only DNG:s). I have a decent mathematical background, and I´ve used the zone system for years before the digital era, so the general concept of a histogram and a zone is familiar to me; that´s not the problem.

 

The manual says that the divisions show the eleven zones (0 - X) of the zone system. If that´s so, why aren´t they equally wide (or at least the widths increase on a logarithmic scale that remains the same for all files)? After all, a zone is defined as one exposure step (factor 2) before any compensating development is applied.

 

I´ve tried to think of alternative explanations, but to no avail. They are not made so each division has an equal surface under the histogram. The progression of widths is not monotonous, so it´s not just the gamma conversion. The pattern of widths varies from image to image in a way that I cannot see any pattern from so far. And, obviously they are produced as a result of the camera´s processing: they aren´t there on the first, ´preliminary´ version of the histogram, but only appear simultaneously with the final one a couple of seconds later (at the same time, too, as the clipping warnings).

 

Now, from a practical, photographic point of view, this is really of minor importance. The shape of the histogram plus the clipping warnings tell me everything I really need to know. Only, being a mathematician, I´m just plain curious...:confused:

  • Like 4
Link to post
Share on other sites

Even after using my MM for many months now I'd never thought of what the lines represent, as you say the curve is the important thing along with the clipping points. But I've only got eight unevenly spaced segments (smaller down the shadow end), not eleven?

 

Steve

  • Like 1
Link to post
Share on other sites

Even after using my MM for many months now I'd never thought of what the lines represent, as you say the curve is the important thing along with the clipping points. But I've only got eight unevenly spaced segments (smaller down the shadow end), not eleven?

 

Steve

 

After reading your answer, I find that indeed there are fewer than 11 divisions (10 lines) on most of my files too. Still, the manual (English p 140, German p 56) has an illustration with 10 evenly spaced lines (11 divisions) that confirms with what I, FWIW, would expect from a normally presented zone histogram. One suspects that there is a bug in the FW (mine is 1.002 btw).

 

Agreed, it is of no practical consequence in itself; maybe it will be corrected in a future FW upgrade.

Link to post
Share on other sites

I am not a subscriber but this site talks about a "bug' in the MM histogram. It also clearly illustrates your comments on numbers and width. Perhaps a subscriber could divulge the contents, in precis of course?

 

diglloyd blog - Leica M Monochrom Histogram Delay (Bug?)

 

Yes, the illustration does show exactly how it looks on my camera in a typical case. I´m not a subscriber either, but it seems that it is the delay between the first, preliminary histogram (without divisions or clipping indicators) and the final one that Lloyd considers a bug. I don´t think it is, really; I presume that the first version is based on the same jpeg rendering that is shown on the display (and so available right after the exposure), while the final version is indeed based on the finished dng file, and so takes a couple of seconds while the camera is processing the sensor´s native output data into a dng.

 

And that´s where I suspect there is indeed a bug; those divisions don´t behave according to Leica´s own description in the manual.

Link to post
Share on other sites

To make sense of the display have you tried shooting a grey card or similar full frame surface at various "zones" and seeing what they display in terms of widths etc?

 

BTW unrelated but in a way similar the M8 manual says the White balance Options available in the fixed menu are that, fixed, but in real life the Flash setting actually changes I assume trying to balance flash and ambient but fixed it is not. (On a Phase One back the flash WB setting is often recommended as a very neutral setting before processing individually whatever the scene.)

Edited by chris_livsey
  • Like 1
Link to post
Share on other sites

Advertisement (gone after registration)

To make sense of the display have you tried shooting a grey card or similar full frame surface at various "zones" and seeing what they display in terms of widths etc?

 

BTW unrelated but in a way similar the M8 manual says the White balance Options available in the fixed menu are that, fixed, but in real life the Flash setting actually changes I assume trying to balance flash and ambient but fixed it is not. (On a Phase One back the flash WB setting is often recommended as a very neutral setting before processing individually whatever the scene.)

 

Yes, I suppose some controlled experiments would give an inkling of what´s going on, but since I just got the camera, and since this phenomenon doesn´t actually influence the result, I will postpone them for a while in favour of actually taking pictures with this great camera!

 

And, as your M8 example shows, the manuals are sometimes wrong. For example, the Monochrom one states that the body top and bottom is "black or steel grey paint"; in reality it only comes in black chrome….

  • Like 2
Link to post
Share on other sites

Thanks for raising this issue. I have scratched my head about this on and off as the width does not increase monotonically when progressing to the right of the DNG histogram, suggesting that the intervals do not all correspond to complete zones.

 

Concerning his suggestion of a bug, perhaps diglloyd is confusing the initial JPG histogram from the one displayed subsequently for the DNG files. The second image shows a slight jump in width at the same time, when the representation of the DNG file is displayed. As he does not make his assertions in a public area of his web site, one can but speculate.

 

It would be useful if Leica could provide a brief technical summary of the DNG histogram, the scaling used and how it maps onto the initial JPEG histogram. Some guidance on the interpretation of the DNG histogram in terms of the Zone system would help, e.g., at least an example histogram with the zones 1 to 9 (or is it 10) added as labels. My Monochrom camera displays 9 intervals (like the one linked to on the diglloyd site above) suggesting the sequence I, II, III, IV, V, VI, VII, VIII, IX, but zone V is about 1/4 of the way across the plot spatially from the left. However, the second from right (here labelled zone VIII) appears to cover only about 3/5 of the width of the intervals adjacent to it.

 

Nick

Edited by Nick_S
  • Like 1
Link to post
Share on other sites

Here is a test that tries to make some sense of the divisions. I have photographed a Gretag Macbeth card. I have bracketed exposures to try to get the bottom left patch as close as possible to 100% luminosity. I have zoomed the photo on my Leica MM lcd to show one patch at at a time. I then measured the actual luminosity of the un-ajusted file in Lightroom. The results are as follows:

 

Histogram 1 - Actual luminosity : 99%

 

Histogram 2 - Actual luminosity : 96%

 

Histogram 3 - Actual luminosity : 89%

 

Histogram 4 - Actual luminosity : 76%

 

Histogram 5 - Actual luminosity : 56%

 

Histogram 6 - Actual luminosity : 33%

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

  • Like 1
Link to post
Share on other sites

I still don't see how it makes sense of the unequal divisions on the scale?

 

Zone V was always going to be more or less dead in the middle, but the other zones shouldn't be of different sizes, unless the sensor isn't linear and the segments represent points on a built in hard wired characteristic curve?

 

Steve

Link to post
Share on other sites

I think the lines are pretty useless in there current format. It looks like some attempt to show the zones for a linear DNG. The only positive is that the last zone seems to be more or less representative of the last stop available on the sensor and in reality that is the only part of the histogram that really matters.

Link to post
Share on other sites

Here is a test that tries to make some sense of the divisions. I have photographed a Gretag Macbeth card. I have bracketed exposures to try to get the bottom left patch as close as possible to 100% luminosity. I have zoomed the photo on my Leica MM lcd to show one patch at at a time. ….

 

Thanks! An interesting experiment, although I´m still confused….:confused:

 

Still, since the divisions are identically placed on all your detail histograms from the same file, at least they don´t vary between parts of the image…

 

I´m planning another experiment, but I am busy the next few days, so I´d better come back about it (I´m planning a series of images, all containing the same 3 zones (and identically exposed for Zone V placement, but with varying image areas for these zones. They SHOULD then generate histograms with 3 peaks in the same positions, only of varying height. If the dividing lines are not equally placed in all of them, something beyond the zone system is taking place…).

Link to post
Share on other sites

I still don't see how it makes sense of the unequal divisions on the scale?

 

Zone V was always going to be more or less dead in the middle, but the other zones shouldn't be of different sizes, unless the sensor isn't linear and the segments represent points on a built in hard wired characteristic curve?

 

Steve

Zone V is not in the middle of a linear scale. As the zone system is logarithmic, zone V is 18% reflectance. I think the MM histogram reflects this.

Edited by jaapv
  • Like 1
Link to post
Share on other sites

Zone V is not in the middle of a linear scale. As the zone system is logarithmic, zone V is 18% reflactance. I think the MM histogra, reflects this.

 

But then, why aren´t the widths of the divisions monotonically increasing? Besides, after seeing quite a few histograms of "real images" by now, they don´t look the least like those from a linear raw file without gamma correction. Those are totally bundled up to the left end of the scale; the MM raw ones look like "normal" histograms.

Edited by elgenper
Link to post
Share on other sites

Zone V is not in the middle of a linear scale. As the zone system is logarithmic, zone V is 18% reflectance. I think the MM histogram reflects this.

 

Thanks, yes you are correct, I meant the sensor isn't linear, not the Zone System, when looking at a Zone scale all the darker tones are not bunched up down one end.

 

I think the segments/lines are points on a characteristic curve, the uneven spacing being where the curve changes shape on the X/Y axis

 

Steve

Link to post
Share on other sites

The DNG histograms from wjkotze's tests with a Gretag Macbeth card, reported above, have only eight intervals to the histogram, compared to nine that I obtain in my photographs and the link to diglloyd example, indicating that the horizontal scale varies.

 

Nick

Edited by Nick_S
Link to post
Share on other sites

The DNG histograms from wjkotze's tests with a Gretag Macbeth card, reported above, have only eight intervals to the histogram, compared to nine that I obtain in my photographs and the link to diglloyd example, indicating that the horizontal scale varies.

 

Nick

 

Indeed. Since the card, under even illumination, must be considered a low contrast subject, it is likely that the algorithm for these divisions is image dependent. The obvious question is: why? It is contrary to the ideas behind the zone system (unless there is some automatic analogy to compensating development in there...).

Link to post
Share on other sites

Indeed. Since the card, under even illumination, must be considered a low contrast subject, it is likely that the algorithm for these divisions is image dependent. The obvious question is: why? It is contrary to the ideas behind the zone system (unless there is some automatic analogy to compensating development in there...).

 

This can't be the answer.

 

I have just shot a three image test, low, medium, and high contrast and the divisions are all in exactly the same place, and it's the same place they have always been. I get eight segments, and counting the end lines of the histogram nine lines.

 

I've never seen nine segments as Nick S reports, I've never seen them move, they are always in the same place as wjkotze's examples above. It cannot be contrast dependant where the lines/segments are. The only consistency is that photographing a mid grey card the peak of the histogram is always on the sixth line from the left (or the fourth line from the right, including the end lines of the scale).

 

Steve

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...