Lightroom: Absolute Height Scale in Histogram to aid Comparisons, etc...

  • 5
  • Idea
  • Updated 7 years ago
  • (Edited)
Presently, the develop histogram is normalized such that the highest tone takes up the full vertical space - all other tones are scaled accordingly. Unless one is trying to compare histograms, this is kindof a good thing, since full advantage is taken of the limited vertical space allocated for the histogram. But it makes it hard to compare views histogram-wise, since the heights between views are squirreling around. I'd love to be able to switch back and forth between two views and not have the height re-normalized between views.

Put another way, I'd like to be able to compare views using the same height scale for both views, such that I can tell the differences between tonal distribution differences without having the height scale change between views.

I suppose one way to handle this might be an option to include histogram in before/after comparisons, say Alt-'\', in which case keep height scale same. But, this would only work for a single photo. Perhaps some way to maintain height scale when switching between photos too... - so one could better compare histogram between a photo and a virtual copy as well..., or a photo and a similar photo...

Summary:
-------------
A way to control histogram height for better photo histogram comparisons
(or so height on histogram is absolutely meaningful, not just relatively to other tones in same photo)

Actually, I can see this being useful when adjusting the tone of a single photo too, so height of a cross section of tones is meaningful (instead of always being relative to some changing "tallest").

Perhaps a solution where one can drag the bottom edge of the histogram down, or maybe double click somewhere, to expand to an absolute instead of relative height mode. Yeah, I like that - drag back up or double-click again to return to normal mode, if desired. - something like that...

Not one of my better worded Ideas, but hopefully you are getting the picture, so to speak:

Right now if the tallest part of the histogram is the highlights, and you knock 'em down, the dark tones raise up, even though you haven't changed the dark tones. Not only that, but the highlights don't even appear to be being knocked down until they are no longer the tallest, since the tallest tones are constantly rescaled to take the full vertical dimension.

Put another way, as it is now, when you lower tall tones, they don't get shorter. Instead short tones get taller. I'd like an option for a vertical scale that doesn't change, so absolute height on the histogram is meaningful.
Photo of Rob Cole

Rob Cole

  • 4831 Posts
  • 382 Reply Likes

Posted 7 years ago

  • 5
Photo of john beardsworth

john beardsworth

  • 1044 Posts
  • 241 Reply Likes
Maybe change the Compare and Before/After views so they display these absolute scale histograms nice and big? After all, don't most people judge their work by comparing histograms?
Photo of Rob Cole

Rob Cole

  • 4831 Posts
  • 382 Reply Likes
John - I'd like to suggest that if you don't understand the purpose of a feature request / idea, that you either ask, or simply refrain from commenting.
Photo of john beardsworth

john beardsworth

  • 1044 Posts
  • 241 Reply Likes
"John - I'd like to suggest that if you don't understand the purpose of a feature request / idea, that you either ask, or simply refrain from commenting. "

Please don't assume that your somewhat longwinded way of expressing yourself prevents others from understanding you, and responding accordingly.
Photo of Rob Cole

Rob Cole

  • 4831 Posts
  • 382 Reply Likes
It seemed clear from your response that you did not understand the purpose of the idea. If that was not the case, then please accept my apology.
Photo of TK

TK

  • 531 Posts
  • 110 Reply Likes
I have the histogram collapsed almost always all the time. The clipping indication is all I need most of the time.

I don't think that images should be judged by using histograms. That's like judging a song based on a frequency plot.

Sometimes histograms can be useful and I agree with the OP that an absolute scaling is often preferable. I'm supportive of the idea but don't think it is anywhere near critical path. Not as long as even much smaller requests are receiving the response that they would take away valuable developer time for much more important issues.
Photo of Rob Cole

Rob Cole

  • 4831 Posts
  • 382 Reply Likes
I think there are two kinds of digital photographers in the world:

1. Those who use histograms.
2. Those who don't.

I use them extensively. I'm not going to try and convince anyone about the value of it - that would probably be about as fruitful as trying to convince someone who doesn't like anchovies that they taste good.

PS - I *don't* use the clipping indicators very often - don't need them at all. But, I sure wouldn't argue for getting rid of them.

I guess we have even more proof that different people go about post-processing in different ways, even when preferring the same software.

This Idea was targeted more at the people who *do* use the histograms, since obviously if you don't use them much, then you don't care if the height scale is relative or absolute.

If there are other histogram loving fools like myself, I'd love to hear from them, if not, I'll continue to use the histogram as is without further complaint. I do have other software capable of absolute scaling of the histogram (by design, for good reason, in my opinion), but for most of my purposes, it needs to be *in* Lightroom. If it would be a big deal to implement, or just too few would want it, then forget I brought it up. I certainly don't want to fight about it.

Examples of recent use cases:
--------------------------------------------------------------------------------------
The most recent application came as a result of trying to match, in Lightroom, the tone curve created externally via a DNG profile editor (in the interest of replicating DNG profile curve as preset) - much harder to do when the height scale continues to vary.

Off the top of my head, some other times when I've wanted it:
Trying to come up with general purpose relative presets that manipulate tone (for DevAdjust plugin). If/when Lightroom supports relative presets natively, you might find yourself using the histogram to aid in creation of relative toning presets, or you might not.

I could go on, but I assume at this point, you either get it or you don't...
--------------------------------------------------------------------------------------

Final thoughts:
-------------------
I'm not going to try and convince you that you should want the capability to lock histogram height as I do. I would appreciate if you don't try and convince me that I shouldn't want it, unless you've got some constructive ideas about how to accomplish the objectives that it serves in other ways.

Rob
Photo of Mark Sirota

Mark Sirota

  • 146 Posts
  • 29 Reply Likes
How about a view on the second monitor, like Loupe -- except instead of showing you a rendered image, it shows you the histogram of a rendered image.

This would have several advantages:

  • It's a lot wider than the panel -- you can fit more than 10 bits of resolution on the X axis, even 11 on the widest monitors.

  • There's plenty of space for controls, such as:

  • A control for locking the Y axis (perhaps a toggle between "locked" and "adaptive")

  • A toggle to change the Y axis scale between linear and logarithmic

  • A mechanism for zooming in to see the black point and white point up close

  • In Grid view or Survey view, enough room to display multiple histograms

  • In Compare view, perhaps a fancy overlay to make it easier to see the differences

  • The possibilities are endless for a true histogram fanatic. (I'm not one of them.)
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    Wow. Mark - you think big! - I like it!!...

    Yes, this would be a histogram-aholic's dream if there ever were one.

    I'm getting the feeling this shan't make top priority come Lr4, but I'd sure love to see image data exposed to 3rd party plugins that could implement such stuff - sorry for getting off topic ;-}

    I would definitely vote for a more robust histogram option for secondary monitor.
    Photo of sizzlingbadger

    sizzlingbadger

    • 31 Posts
    • 0 Reply Likes
    I think we should have the option for the histogram to be reversed for left handed people.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 376 Reply Likes
    Yes, my Nikon has an option to flip the exposure meter to match a Canon (or was it the other way around...) Its good for the dyslexic too...
    Photo of Geoff Walker

    Geoff Walker, Champion

    • 214 Posts
    • 42 Reply Likes
    I think Lightroom can do this already, just use the Mirror Image option - surely the histogram will be mirrored as well, if not that would be a bug!!
    Photo of TK

    TK

    • 531 Posts
    • 110 Reply Likes
    Rob, I'm sorry if you misunderstood my previous post as stating that your idea isn't needed. Note that I said that I am supportive of the idea.

    The sentence that triggered your reaction (e.g. you wrote "This Idea was targeted more at the people who *do* use the histograms,...) was probably my statement: "I don't think that images should be judged by using histograms.". This, however, was targeted at the statement "After all, don't most people judge their work by comparing histograms?" by JB.

    I'm sorry if you were annoyed by a statement I targeted at someone else. I considered using a comment (as opposed to using a reply) but I don't think comments work well for several reasons.

    So, again, I'm sorry if you have perceived my comment to the "After all, don't most people judge their work by comparing histograms?" remark as criticism of your idea.

    While I do think that your idea doesn't deserve critical attention, I am supportive in general.
    Photo of Andrew Rodney

    Andrew Rodney

    • 494 Posts
    • 86 Reply Likes
    >I'm not going to try and convince you that you should want the capability to lock histogram height as I do. I would appreciate if you don't try and convince me that I shouldn't want it, unless you've got some constructive ideas about how to accomplish the objectives that it serves in other ways.

    You have it backwards. Its YOUR job to sell the request to the engineering team both in terms of its usefulness and as something that needs to be put towards the top of a crowed feature list for the next rev.

    So far, most have simply asked you to provide some demonstrative methodology to back up the request and you seem to dismiss this request as if its neither appropriate nor necessary. As Victoria said, educate us (and the team).
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 379 Reply Likes
    Quoting Andrew Rodney: "You have it backwards..."
    - I made a judgement call that in this case, it would take more time & effort than I'm willing to spend on it.

    If you don't use the information that the entirety of the histogram provides, then this idea is not for you.

    To sell you on this idea, I would first have to convince you of the value of using the histogram as it is. I'm having the feeling this would be no easy chore...
    Photo of Andrew Rodney

    Andrew Rodney

    • 494 Posts
    • 86 Reply Likes
    >To sell you on this idea, I would first have to convince you of the value of using the histogram as it is. I'm having the feeling this would be no easy chore...

    Your call man. I’ve got an open mind, I’m willing to see what you have to demonstrate but you have to be willing to demonstrate it!

    I also question the value of any histogram that’s not based on what you’ll end up with once you render the data. My beef at this point is the current histogram isn’t based on anything you’ll get out of the LR environment.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 379 Reply Likes
    Quote: "My beef at this point is the current histogram isn’t based on anything you’ll get out of the LR environment" -
    - this sheds some light on some previous comments you made. I still haven't seen a new FR/Idea from you regarding this - only a small beef?
    Photo of Andrew Rodney

    Andrew Rodney

    • 494 Posts
    • 86 Reply Likes
    >I still haven't seen a new FR/Idea from you regarding this - only a small beef?

    There’s little reason to provide a FR here, I’ve expressed this over the years to the team “elsewhere”.
    Photo of Victoria Bampton - Lightroom Queen

    Victoria Bampton - Lightroom Queen, Champion

    • 4335 Posts
    • 1583 Reply Likes
    Rob, I can understand how you'd use this, but how do you imagine it being of use to the majority of users?
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    I imagine the majority of users don't use the histogram (except maybe for periodically checking the head and/or tail of it). If that's correct, then this would not be of use to the majority of users. I also imagine the majority of users don't shoot tethered, but that doesn't mean tethering improvements are a bad idea.
    Photo of Victoria Bampton - Lightroom Queen

    Victoria Bampton - Lightroom Queen, Champion

    • 4335 Posts
    • 1583 Reply Likes
    I agree the majority of people don't tether, although there is a significant percentage of the user base. But we're not talking about tethering - we're talking about your feature request.

    The thing is, I can't imagine any notable percentage of the user base needing this feature (or honestly, right now I can't imagine anyone but you), so I'm asking you to educate me.

    You've explained how you'd use it, which is great, but it sounds very specific and they're unlikely to build it just for you, so let me ask the question again another way... how would other people use it in their workflow?
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    Quoting Victoria Bampton: "I'm asking you to educate me."
    Sorry Victoria, but I no longer want to discuss this in this thread. Please contact me via PM to discuss further, if you want to.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    Thank you to all who've attempted an earnest response in this thread.

    I've presently said all I care to say about the subject. Maybe other histogram aficionados will chime in at some point, or I'll find some new inspiration...
    Photo of Andrew Rodney

    Andrew Rodney

    • 567 Posts
    • 94 Reply Likes
    >Rob, I can understand how you'd use this, but how do you imagine it being of use to the majority of users?

    Exactly, great question V.

    Its funny because the histogram Rob seem to think is so telling is based on a color space he’ll never get when he export the data! Its Melissa RGB.

    And other than the important info in a Histogram (saturation or tone clipping), just what else is being shown here that’s at all useful? So this is something the engineering team needs to spend time on? A compelling example of use is in order.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    There are plenty of worthwhile features in Lightroom that are not used by the majority.

    I use the histogram to help accomplish some of my goals (that's different than "seeming to think it's telling").

    The histogram is information, its usefulness varies depending on ones objectives (and their imagination, and their ability to interpret the information...)

    But, I have no desire to try and explain it to people who seem to have already made their minds up about it.
    Photo of TK

    TK

    • 531 Posts
    • 110 Reply Likes
    I don't think that people have necessarily made up their mind about it. But it seems clear that most don't miss absolute histogram scaling in their daily work. Therefore, it would be useful to hear about more use cases.

    Sometimes when people read about a particular use case they go "Oh, that's something I should try" and if it is clear that your FR is required to support the use case, surely they will vote for your FR.

    I'm in that boat. If I could think of one example where the absolute scaling would be really helpful to me, I'd vote for it. I agree with all your points regarding the relative scaling being unhelpful at times, but it doesn't really affect me. I'm nevertheless open to learn as to why it might affect me in something I want to do with histograms.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    Sorry TK. Feel free to PM me if this subject interests you, but I no longer want to pursue it on the public forum.
    Photo of Andrew Rodney

    Andrew Rodney

    • 567 Posts
    • 94 Reply Likes
    To put a somewhat more positive spin on a Histogram request, and for those that presumably do find histograms useful (above and beyond highlight and saturation clipping), let me suggest we have the ability to select an encoding color space we will use for export and have that presented to us instead of Melissa RGB which we can’t really (easily) get out of the product. This would put the LR histogram on close parity with ACR where when setting one of the four options in the Workflow link, the Histogram updates. Being able to select ProPhoto RGB from sRGB makes a profound impact on the saturation clipping of the current image with the current settings.

    LR provides more options for an encoding color space than the four in ACR so we’d need some provisions for setting it as such. One idea would be a process of selecting profiles as we currently have in the Print module. I could see a user option clicking on the Histogram itself and having a Load dialog like that for output profiles in Print where you could populate a list with as many or few RGB color spaces as you wish, then toggle between them. When done, the Histogram would update.

    IF (and I certainly hope) LR4 has soft proofing, being able to select the output profile for the soft proof and updating the Histogram with that profile could be rather useful IMHO. In addition, the RGB percentage values should update, again giving us feedback based on a color space we intend to produce out of LR (Edit in PS or Export).
    Photo of Victoria Bampton - Lightroom Queen

    Victoria Bampton - Lightroom Queen, Champion

    • 4327 Posts
    • 1582 Reply Likes
    Now that one I would vote for! Write up a separate request Andrew, as it's sure to get lost here.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    I agree this should be a separate thread.
    Photo of TK

    TK

    • 531 Posts
    • 110 Reply Likes
    I agree that the ability to select colour spaces would be useful and it should be formulated as a separate FR. Judging from older LR podcasts involving the development team, I don't see a high likelihood of such a FR of making it, but it is worth trying.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    OK - y'all win:

    This excerpt I was going to post elsewhere, but (some of) you asked for it, and I have a feeling I may regret this...

    Some Lightroom users don't understand the value of the histogram, other than as clip warning. What else is it good for?

    Some users may glance at the head & tail from time to time, but don't pay much attention to the part in between - so why should they?

    My .02: Its not for everyone. I'm sure some people's style is entirely right-brain, and the histogram would serve no useful purpose except to interrupt them from their "zone".

    The histogram is information. Still begging the question what good is it.

    Unfortunately, I don't think there is any short answer to this question. But let me start with a couple of examples:

    - What is the difference between brightness and exposure? This question can be readily answered by cranking each one up in turn and watching the histogram. Armed with this new knowledge, am I better equipped to know when to use brightness and when to use exposure to lighten or darken my photo? It could, or it might not - that really depends on your style. So, regarding brightness vs. exposure, is the histogram useful - I can't speak for you. But for me: immensely.

    - What does contrast do? We all know intuitively, but the histogram can answer that question far more specifically: separates light and dark tones according to a mathematical formula. So what about the midtones? where is the split?, is the split point dependent on exposure? or other settings? Why do the hues change sometimes when adjusting contrast, and why does the saturation change sometimes, yet not others? How could I restore the original hues yet maintain contrast? Who cares? Again, the answer is personal - you may not know nor care, and that's fine, or you may have your own way of figuring stuff out, or dealing with it without figuring it out exactly. But most of these questions can be answered by observing the histogram.

    What is a histogram? its a graph, on the x-axis (horizontal) is a tone or color, and on the y-axis (vertical) is the number of pixels having it. How does one read it? Well that depends. If the height scale is absolute, then one can tell how many pixels are at each tone/color by its height. If the height scale is relative (e.g. Lightroom's) then the actual height in the histogram is meaningless. Does it matter? Maybe so, maybe not - again, I'm not qualified to speak for others, only myself. To me, sometimes it matters, and sometimes it does not matter.

    Enough theory! - what does all this mean in the real world?

    Well, what is the best way to recover detail in highlights (meaning in this context: to separate intra-highlight tones) using local brush? Answer: reduce contrast. Why? because contrast compresses highlights. How do I know that? because I've seen it on the histogram. Are there other ways to have arrived at the same conclusion? Absolutely, so what value was the histogram in this endeavor? For you: I don't know. For me: a lot. And bonus question: what if reduced contrast made the highlights too dark, what is the best way to brighten them back up? Answer: exposure, not brightness. How do I know? From watching the histogram. How do you know? I don't know.

    So, why would one want the height scale to be stable? I've already partially answered this above, but to recap:
    1. If you don't use the histogram much, then this would not be of much value.
    2. If you do use the histogram a lot, then you probably already know the answer.
    I mean, when you are just editing a single photo, then its usually no big deal, although even in this case, I'd sometimes prefer the absolute scale, so that when a portion of the curve is rising up, it means that more pixels are having the corresponding tone (presently it may just mean less pixels are having some other tone - impossible to tell). Why? I'm not sure I know how to answer that, except if the histogram information is valuable, then its more valuable when it represents what's actually happening. I'm not sure how to explain - I think you may have to experience the histogram a while, and then this may become apparent.

    The times when its most valuable are when its important to be able to tell exactly whats happening. Two use cases have already been mentioned above, to recap:
    - Creating a tone-curve preset that matches the tone curve in a DNG profile.
    --- While doing the job, what's important is the histogram, since trying to do it based on how the photo looks is too hard.
    - Creating a set of relative presets for toning.
    --- Again, the photos are not what's important while coming up with the presets, but tone curve response is (as indicated by histogram).
    - Normalizing tonal response between cameras and/or lenses.
    --- e.g. trying to get the Nikon Vivid look from your Canon. One aspect of this endeavor is tone matching.
    Although the DNG editor is an option for this, I don't like using it, because its not convenient - not integrated in Lightroom (and requires extra steps for non-DNG'rs). So, I use presets as much as possible. So matching tonal response means comparing a photo taken with my Nikon, assigning it the Vivid profile, then comparing the histogram to the same photo taken with a Canon - tweaking tone curve in Lightroom until they match. Much easier with a height-stable histogram.
    --- The same would apply if you ventured to normalize your lenses tonal responses, although I haven't done that (I'm not even sure if that makes sense, but if it did...).
    - Oh yeah another time I wanted this was when developing the DevConvert plugin - trying to automatically convert photos from PV2003 to PV2010 without altering tone. Since fill-light tonal response changed, tone was different for all photos that had fill light applied. So, I was trying to match the distribution of tones in PV2003 photos with the same photos converted to PV2010. A height-accurate histogram would have been preferred for this endeavor (PS - I gave up before accomplishing this goal in a satisfactory way, but still - the height-squirreling histogram didn't help).
    - There is an exposure matching function in Lightroom, but it only matches exposure, it does not attempt to match contrast. I've got to stop here but hopefully you see that using the histogram to help contrast-match a set of similar photos, a height-stable histogram would be better.

    Summary:
    ========
    Although it would be beneficial to anyone who prefers to view a more height-accurate histogram while they work, the occasions when it is *really* helpful are when you are trying to do something with tone that involves more than just tweaking the present photo to taste.

    Final thoughts:
    ===============
    I have been reluctant to post this because it seems people sometimes prefer to refute (often without first understanding), than to grasp the points being made, or worse still, resort to personal attacks on me or my character instead of addressing subject content.
    Photo of Andrew Rodney

    Andrew Rodney

    • 567 Posts
    • 94 Reply Likes
    >>If the height scale is relative (e.g. Lightroom's) then the actual height in the histogram is meaningless. Does it matter? Maybe so, maybe not - again, I'm not qualified to speak for others, only myself. To me, sometimes it matters, and sometimes it does not matter. <<

    How and why is the $64K question. And when (?) since you say it 'depends'. Example images and decoding of the usefulness (when and when not meaningful) is what I'd like to see demonstrated.

    >>Absolutely, so what value was the histogram in this endeavor? For you: I don't know. For me: a lot. And bonus question: what if reduced contrast made the highlights too dark, what is the best way to brighten them back up? Answer: exposure, not brightness. How do I know? From watching the histogram. How do you know? I don't know. <<

    If one understands what Expsorue does (its a white clipping control as described on page 35 of RWACR by Fraser and Schewe), if you look at the clipping indicators (alt/option) and the image itself (or the numeric values), how is looking at the hight of Histogram better, necessary or preferable? It shows clipping in both saturation and tone without the need to see the height. What am I missing here? So that’s how I know. Is that not adequate and if so, why?

    >>So, why would one want the height scale to be stable? I've already partially answered this above, but to recap:
    1. If you don't use the histogram much, then this would not be of much value. <<

    The question is WHY would you use the histogram on the vertical axis? Knowing more or less pixels are affected in this axis is useful because?

    >>I'd sometimes prefer the absolute scale, so that when a portion of the curve is rising up, it means that more pixels are having the corresponding tone (presently it may just mean less pixels are having some other tone - impossible to tell). Why? I'm not sure I know how to answer that, except if the histogram information is valuable... <<

    I think that sentence sums up the frustration and desire of those of us asking for your guidance in this feature request.

    >>The times when its most valuable are when its important to be able to tell exactly whats happening. Two use cases have already been mentioned above, to recap:
    - Creating a tone-curve preset that matches the tone curve in a DNG profile. <<

    So the two tone curves in differing locations are the same? Used the same why?

    >>I have been reluctant to post this because it seems people sometimes prefer to refute (often without first understanding), than to grasp the points being made, or worse still, resort to personal attacks on me or my character instead of addressing subject content. <<

    Where are these personal attacks?
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    I'm not sure I'll be able to explain this any better (except maybe by walking through specific examples complete with visual media accompaniment - but that's just too much for me to do...). Try some of the things I've mentioned and you'll see for yourself. Again, if you never do the things where it matters, it doesn't matter. I've given 4 cases from my experience.

    I think you need to be a person who is experienced using the histogram to aid in "unusual" toning endeavors to really appreciate this idea. When simply developing a single photo, the constant re-scaling of the histogram is usually OK, often even preferable since it takes full advantage of limited vertical space, without vertical clipping (I could also argue for a margin that extends leftward and rightward beyond the horizontal clipping points, to give a better feel for what is being clipped, but that would take more work than this, for the lr-developers, and more CPU cycles in real-time, and has nothing to do with this FR/Idea).

    I'm sure the majority of users don't care about this. A better question is: do enough users care about it to make it worthwhile given the development effort. My estimate of the development effort: Small. My estimate of the number of users who would appreciate it: Small. Its up to Adobe to decide what to do and what not to do.

    There have been numerous personal attacks on me and others in this forum, I'm guessing you know that. Perhaps it suffices to say it makes me reticent. Even when attacks are not personal, sometimes people tend to react to a (potentially good) idea with contempt due to their own personal biases... (first reply in this thread was a scathing dig which was subsequently removed). And sometimes the more one says, the bigger the target and the easier the shooting...

    PS - I look at the histogram while developing photos like I look in my mirrors when driving. I know exactly how what I see in the photo maps to what I see in the histogram. Why do I do this? partly just because I'm curious and I find it interesting, but partly because it helps me to know which combination of exposure, brightness, and contrast to use in a local to affect a desired change in some region. The histogram may not be necessary to be able to do this, but its helpful to me.

    In my earliest of editing days in Lightroom, I used to try and add contrast using locals to make something pop. I didn't understand why it had the exact opposite effect in shadows and highlights, and had no effect whatsoever at what I now call the "contrast split point". I now understand this thoroughly, and the histogram was instrumental for me in coming to this understanding. I suspect there are other people who have come to the same understanding without ever looking at the histogram. Peoples minds work in different ways.

    The histogram is there for the minority of users who find it valuable, and this FR/Idea is for when constant vertical axis scale factor would be preferred.

    Regarding specific question above: "So the two tone curves in differing locations are the same? Used the same why?"

    sizzlingbadger came up with a great DNG profile based on Adobe Standard, it included a tone curve which debrightens the midtones and preserves detail in highlights. I sometimes prefer this toning and sometimes not, but I wanted to be able to apply it to other camera profiles without having to create a second version of each profile - so I created a tone-curve preset which I can use with any profile. And, if I decide to tweak it, or fork another version of it, I can do so without resorting to the DNG profile editor, and without further cluttering my camera calibration profile list, which does not support folders nor maintenance in Lightroom the way presets do.

    Question: How does one know when the tone curve in the preset is the same as in the DNG profile?
    Answer: There's probably more than one way, but what I did is continue to tweak the tone curve of a photo (with same base profile and color tweaks), in Lightroom, until its histogram matched the histogram of the same photo but with the aforementioned DNG profile. Just try this and you'll see how its harder to know what to do, and tell whether you're getting closer... when the vertical scale of the histogram keeps changing.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    Another case came up recently where histogram matching would be desirable (and would be facilitated by consistent y-axis scaling):

    Coming up with brush presets to emulate the various vignette styles for the purpose of applying custom vignettes using local adjustments.

    First, apply brush strokes until coverage matches Lightroom vignette (by eye). Then adjust brightness, exposure, and contrast until histogram matches what you do with the Lightroom built-in vignettes, then you know your local presets for vignetting match the Lightroom styles. Even if you were not entirely successful in this venture, it would still be instrumental in learning to apply custom vignettes via local adjustment to satisfy your fancy, using exposure & brightness & contrast..., instead of just exposure.
    Photo of Andrew Rodney

    Andrew Rodney

    • 567 Posts
    • 94 Reply Likes
    Again, why is this y-axis (height) useful? The vertical axis (the height of points on the graph) shows how much of the image is found at any particular brightness level. The level itself, yes that’s useful. The amount of pixels at that level?
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    Try it, then tell me if you think it would be easier if the y-axis didn't change as you make adjustments, or if you think it wouldn't matter.

    I mean the answer is: its easier to compare histograms when y-axis is same on both histograms, and isn't changing every time you make an adjustment. But, I already said that. I really think you need to try this stuff and it will become clear.

    As you said: "The level itself, yes that’s useful" - right, but you can't tell what the level is because the scale of the y-axis itself keeps changing - its constantly being normalized so the tone with the greatest population takes up 100% of the vertical axis.
    Photo of Andrew Rodney

    Andrew Rodney

    • 567 Posts
    • 94 Reply Likes
    The value in these contexts are meaninglessly as far as I can understand, why would I need to look at something that has no meaning (unless you can explain the meaning).

    What’s really easier for me is to look at the image.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    I'm sure many people get by without ever doing histogram matching, or even histogram observing. It was never my intention to convince people they should do it. I look at this from an opposite point of view: when coming up with tone-matching presets, the photo doesn't matter - only the histogram (in fact, I've thought of coming up with test-pattern photos to take the "photo" out of the equation). When using and applying and coming up with preferred variations of presets, just the opposite, (what the photo looks like is the only thing that matters) but that's not what I'm talking about here.

    PS - there's a difference between being "not valuable or meaningful", and not being something you'd ever care to do.

    Really, this FR/Idea is for people who do use the histogram to do the kinds of things where being able to interpret what's actually happening when adjustments are made is of value, to them.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    I came up with this FR/Idea after the Nth time I tried to do something using the histogram and had a hard time because of the height re-normalizing. If I thought it would be a big deal to implement, I wouldn't have brought it up. Its an easily doable thing that would be preferable for some of us, especially in unusual circumstances.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    There is more than one thing that have become issues now:

    1. What good is the histogram, besides to show how close to the endpoints one is.
    2. What good is the height of the histogram.
    3. What good is constant height histogram capability.

    Regarding #1 - uses are many and varied, and subject to personal inclinations. For me, I track two things: What exactly do the tools do, histogram-wise, and how exactly is my photo represented, histogram-wise. A little game I play: trying to come up with exactly the right combination of exposure, brightness, and contrast for a local adjustment to do exactly what I want, with no trial and error, by analyzing the histogram, and surveying the tones in the region to be adjusted. Its not for everyone... - that's just an example. There are other uses for the info too...

    Regarding #2 - Height on histogram indicates relative number of pixels with specified tone or color.

    Regarding #3 - Constant height histogram only matters when changing adjustments. If scale varies, then its harder to tell what's actually happening as adjustments are being made, since increasing/decreasing height at a specified point does not necessarily mean there is an increasing/decreasing number of pixels having that tone or color, respectively, since it could also be due to a decreasing/increasing number of pixels at some other tone, or a combination of the two.
    Photo of Andrew Rodney

    Andrew Rodney

    • 567 Posts
    • 94 Reply Likes
    I (and I suspect most here) know what 1&2 mean in terms of a histogram.

    The question remains why is the height important. You “Regarding #3“ doesn’t clear this up for me. Knowing there are 10,000 pixels of Level 85 or 1000 pixels at level 85 tells me what about the work I’m doing? And since its totally image specific (and in this context is based on a histogram of data I‘m not getting), its even more mysterious to me what the height (number of pixels) is saying. There’s no scale as well for this Y-axis. So I don’t even know if the height shown is 1000 or 100,000 or said pixels (assuming knowing this number were useful). If you open a 3mb JPEG or a 40 megapixel raw, the height is the height.

    So again, just confused by the info presented for the FR.

    The X-Axis is well defined and understood although one could suggest we are working on a high bit file with a Histogram who’s encoding isn’t high bit. Making BOTH scales odd at best.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    You can't tell what exactly has happened when you make an adjustment, unless the vertical scale factor hasn't changed - its really that simple. Put another way, as presently implemented in Lightroom, increased height doesn't necessarily mean an increased number of pixels at that tone, but it could - no way to tell from the histogram, whereas if the scale factor is fixed, one can tell exactly what's happening as adjustments are made: if its getting taller, then more pixels are having that tone, if its getting shorter, then less pixels are having that tone. Its not the exact number of pixels that matter - I don't care if the axis is labeled, its about how the adjustments are affecting the tonal distribution: in one case you can't tell, in the other case you can. Again, whether this matters depends on the person and what they're trying to accomplish, and how they're trying to accomplish it. Try the things I've suggested and you'll see.
    Photo of Mark Sirota

    Mark Sirota

    • 146 Posts
    • 29 Reply Likes
    I do understand what you're saying about the fixed height making it clearer about what's happening. If the Y axis were fixed and you're just looking at, say, the 3/4 tones, you don't need to worry if some adjustment adjusted the peak which happened to be on some other part of the histogram.

    Andrew, it doesn't matter if it's a 3 megapixel or 40 megapixel image -- the height is what *percentage* of the pixels are at that value, rather than what *number* are at that value.

    So I agree that it would be easier to see what the tools are doing to the histogram if you could lock the Y axis scale. I still question whether that's of *significant* value, though.

    I'd also be concerned that if you lock the Y axis scale at some point and then make an adjustment that drives the peak above the top of the scale, you'd no longer be able to assess the effect of the change. Some visual indication that this has happened would be needed, followed by a user action to reset the Y axis scale. Sounds cumbersome when you're trying to focus on what you're doing. Can you think of a solution?
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    You got it. And, how much value it has really depends on what one is trying to do, and how one is going about doing it. I noticed this phenomenon in the earliest of my Lightroom editing days, and didn't think much of it, except: "I see why they are doing it - full use of available space with no clipping". I'm not trying to suggest what Adobe did was anything but sensible, just that a tweak for us weirdos if not too much trouble would help sometimes. One of the reasons I suggested using a pull down bottom edge is so user could control the scale to eliminate clipping if need be. Also the scale would not need to be linear, since the objective is generally "correct trends" and "comparable results", not "correct amounts".
    Photo of Andrew Rodney

    Andrew Rodney

    • 567 Posts
    • 94 Reply Likes
    >You can't tell what exactly has happened when you make an adjustment, unless the vertical scale factor hasn't changed.

    I can look at the image or view the values (percentages)...
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    Its a lot harder to do tone curve matching (and such stuff) by looking at the image than by comparing histograms - but you're getting the idea...

    And, you are absolutely correct: the percentages would tell you exactly the same information, but you'd have to scan the entire range of tones after each adjustment to get the same info you'd get at a glance with a height-stable histogram. Plus you'd have to compute luminance from R,G, & B - this is entirely prohibitive.

    All you have to do is try one of the things I've suggested, and you'll see. Put another way, if you really want to understand the purpose of this FR/Idea, just try one of the things I've suggested. In other words, without the experience of having tried to do some things like what I've suggested, its nearly impossible to understand the value of a height-stable histogram to aid in such endeavors.
    Photo of Andrew Rodney

    Andrew Rodney

    • 567 Posts
    • 94 Reply Likes
    >Andrew, it doesn't matter if it's a 3 megapixel or 40 megapixel image -- the height is what *percentage* of the pixels are at that value, rather than what *number* are at that value.

    The percentage of data, number of pixels, in the two differing captures is itself different, but even so, the questions about the usefulness of the height still remain in my mind.

    >I still question whether that's of *significant* value, though.

    Exactly. And again, the values, percentages or otherwise are not what you get from the rendered image. Until we have the option to set the color space encoding, the current histogram’s usefulness is questionable, even in areas most users understand; saturation and depending on the TRC, tone. I’d have a far better time accepting the histogram once that’s taken into account.

    In terms of the FR, my suggestion to Rob is to ask Thomas to think about it for ACR. At least that histogram is based on something we’ll get. And if Thomas agrees its a useful FR, far more likely the LR team would.
    Photo of Rob Cole

    Rob Cole

    • 4831 Posts
    • 382 Reply Likes
    Thanks all...