Reznick e-book Need Help

Discussion in 'General Technical Discussion' started by EdMac, Mar 21, 2005.

  1. I have listened to all of the comments expressed here on how much the R Reznick e-book has helped everyone improve their photography and post processing. The result is that I recently purchased the e-book.

    This past weekend, I set about printing out the book, and reading through it. Everything went well until I got to the post processing section using NC.

    The problems that I am having have to do with such things as gamma indicators, gamma ramp and gamma plateau. Are these terms unique to this e-book? I have used Google to search the Internet for a definition, and I have not found it. Is there a theory behind these so called indicators? There is no real definition of these terms in the e-book. You have take it on faith that the points indicated in the examples are in fact valid points. That is all well and good for the examples, but I am completely lost when it comes to a real world histogram of one of my pictures.

    What am I doing wrong? I have spent over 42 years as an engineer, and I have a degree in physics. Maybe I am expecting too much, but right now I am at a loss to apply these techniques with any confidence.

    I am sure that all of you folks that have the e-book have a good understanding of what all of this means. I would really appreciate a better explanation of just what all of this means. Any help will be appreciated. :(
     
  2. Ed you are not alone. I've gone througth that section three times now and I am still in the dark (pardon the pun). As an engineer I have a complusion to know why this is such and such. The eBook does not answer this and other question which, I think, are Ron's empiric observations converted to text. "Where does this stuff come from?" I'd ask myself. I suppose it would be clearer if one was in one of his classes.

    Perhaps when he's back from one of his trips he might offer to entertain some questions.

    Rich
     
  3. Flew

    Flew

    994
    Jan 25, 2005
    Alabama
    Ed & Rich,

    I also have the book, and I've been to two sessions with Ron. I had trouble seeing the gamma ramps in some cases while Ron was going through the live demos, so I can understand your confusion. Like both of you, I'm an engineer, and I have a (sometimes irrational) desire to understand the details of how something is supposed to work.

    Sometimes the gamma ramp is obvious, and sometimes it is not (at least to me). Maybe we can get Ron to post a set of rules for determining how to find the correct point even when it is not. For my own use, I have gotten to the point where I can usually get the gamma pretty close (at least I'm satisfied).

    Just about everything else in the book (and in the in-person classes) is very clear to me now. This is the one area where I still struggle a little. Maybe I'm just dense.... :wink:

    Regards,

    Frank
     
  4. MontyDog

    MontyDog

    Jan 30, 2005
    #1064 - You have an error in your SQL syntax;
     
  5. Monty you old salty dog, you have broken the code for those difficult images where the gamma point is not easly recognizable. I am doing something similar.
     
  6. I'm going to join the bandwagon for those who "cheat" by going straight to 1.08 and looking at where the gamma curve intersects the histogram to see what is there. That frequently works for me. Here's a couple other thoughts:

    - Sometimes I will do a very minor adjustment in Curves for gamma (or none at all) and go to the LCH Editor to look for the gamma indicators, which for some reason are more evident there in many cases (mental puzzle for you: figure out why -- I have no clue why there would be a difference)

    - A trick I learned when I was first going through the book was to WAAAAAYYYYY over-exaggerate the adjustment in both positive and negative directions temporarily to watch the shape of the histogram change. The motion of the histogram in this way will sometimes make a well-hidden indicator become a bit more obvious as you watch the shape of the curve change. You obviously don't want to leave it that way, but watching it go through the motions will help sometimes to coax a small indicator to the surface. This is also a good way to heat up your CPU if you need an additional space heater in your room, because it will tax the computational power of your PC... :)
     
  7. I don't know how many of you have looked at Ron's site lately but he now has an upgrade and explains the question that EdMac asked.

    Is that the explanation you were looking for?
     
  8. Answer

    Hi,
    The answer is no. It is the same answer that is quoted in Monty Dog's post above. The issue is the definition of the terms used. Ron Reznick has clarified his use of the terms in a reply to a post that I made on DPReview. Ron says that he created the terms to try and explain phenomenon that he has observed over his many years of working with photos.
     
  9. Hi gang,

    Here's the response I sent to Ed elsewhere:

    Often, it's difficult to describe a phenomenon unless you invent terms to discuss the issues. During the processing of several hundred thousand images, you tend to recognize patterns that emerge. Some of these patterns noticed were in the location of gamma settings for images. I have given you ranges to use when you can't see indicators for placement of gamma, but quite often there are indicators.

    The amount of shadow information vs. midtone information will determine the strength and location of these indicators. If the shadows are weak with relation to the midtones, you will often see a ramp to the left of the shadow/mid-shadow data rise which sometimes lines up with a gamma indicator in the midtones, creating a virtual gamma ramp. If the shadows are stronger, your gamma indicators will form on the right of the mid-shadow/midtone peaks most of the time.

    The greater the amount of data in the midtones with relation to the shadows, the more apparent the gamma indicators will be.

    If the midtones are weak in comparison to the shadows, the midtones become compressed into the shadow/mid-shadow peak and gamma indicators can be difficult to see. In those cases, often you need to resort to the ranges I gave you and your visual evaluation of the image (select a range that looks right to you -- it's best to work on a calibrated monitor of course). As the midtones become stronger with relation to the shadows, first you will see a shelf in the midtone region (gamma plateau) -- the stronger the midtones are in relation to the shadows the closer to level this shelf will be. Weaker midtones will compress this shelf... before the shelf becomes completely compressed into the shadows it is reduced to a little bump in the curve (gamma point). When the midtones are really strong in relation to the shadows, the shelf will be higher on the right side (higher luminosity region) than the left side. The shelf then becomes a ramp. When you have weak shadows and strong midtones, you will very often see this midtone ramp line up with a ramp in the shadow end. When setting gamma, the diagonal gamma line/curve can be lined up exactly with this ramp.

    I had to come up with some way to describe this pattern of phenomena. Please excuse the obscurity of the terms I came up with to make my poor attempt to do so.

    Ron
     
  10. Flew

    Flew

    994
    Jan 25, 2005
    Alabama
    Guys,

    I just spoke to Ron and as a result, got the amplification above (which he had already posted over on DPR). As he says, finding the gamma points is a matter of processing thousands (or 10's or 100's of thousands) of images, and having the vision to see very subtle indications of these artifacts.

    Since most of us aren't Ron, and haven't processed as many shots as he has, we have even more difficulty. As I said above, I often can see the gamma points / curve, but many times I can't. While I had not worked out the slick technique that David has, I to have come up with a 'cheat', like Paul and others of you. When I can't spot the gamma points, I usually just set the middle value to 1.08, and adjust from there.

    I for one would like to continue this thread as each of us works to refine our skills in this somewhat esoteric area.

    Regards,

    Frank
     
  11. Ron's Response

    From the responses that I have seen here and over on DP Review, it is obvious to me that most of the users of the ebook have come up with a work-around for implementing this technique. No one has come forth and said that it is obvious that you do such and such to achieve the result. Yes it is a tricky area, and it is becoming obvious to me that there can be indicators to help guide the user. It is not obvious that there always are indicators, and that is the source of the dilemma.

    I would like to have more confidence in my ability to determine when my adjustment is correct. Finding a squiggle or a flat spot that may or may not exist in the distribution of pixel luminosity seems somewhat questionable.

    What I would like to know is what do other users use as their criteria to know when they have achieved the optimum setting of the gamma slider? From what I am reading it is very subjective. What are the penalties if I choose to set the slider at some other point? Is there a way to know if I have over achieved? Maybe there is no real answer. What do other users think? :)
     
  12. Well...

    Ed (et al):

    At the risk of over-simplifying this, just looking at the resulting image should tell you if the exposure is the way you remembered the scene, or more to the point: the way you want it to look.

    It is very easy to get lost in the process and forget that the entire point of all of this is ultimately to produce an image that represents the scene as you intended to capture it.

    In my experience, using the post-processing techniques Ron suggests will give you a foundation for any given picture that is most likely to optimize the captured data of the scene -- the better you exposed and composed in the picture, the better the results will be. And even if you goofed in the acquisition (as I all too frequently do), then the process will give you the best chance of recovering.

    Having said that, I try not to look at the histogram when adjusting curve gamma until I've gotten the image looking the way I think it should. Once I've gotten to that point, I will then look at the histogram for indicators as a "sanity check" to see if I was even in the ballpark. More often than not, I find that the gamma point hovers around 1.08 for my shots after setting the individual color curves. This is why I use that point as a default when I'm having a hard time deciding where to go. And occasionally (usually when I screwed up big time with the exposure or the scene was just too contrasty), I will find that the "correct" gamma is nowhere near that point.
     
  13. Re: Well...

    It would have been helpful if terms and the basic approach had been explained prior to the examples. Also if you recollect the first exercise with the pelican starts without an explanation of the approach. In other words if the book had something like:

    "First we're going to adjust the White balance, then the exposure level, and then the shadows.

    We're going to try several methods of finding the white balance. We need to because there are certain circumstances which necessitate a few different approaches.

    - When x we'll explore method a
    - When y we'll explore method b
    - When z we'll explore method c

    the same approach for exposure level and shadows." It would have a bit more continuity.

    More explanation should be given about the shadows. For example I don't recall if there is an explanation why the shadow exercise is so important. For example, an engineer might know about "normalizing" the gamma. Most don't. It would help to explain that as well. It doesn't matter if Ron's terms are his "own" if he provided profuse explanations all this argy bargy might be avoided.

    Then we need to be reminded what gamma means (in Ron's terms). What is the significance between high medium and low gamma? What is so important about a ramp, why would you need to know about it? Instead of the long summary after each exercise, why not put it before the exercise so we can try and share Ron's vision?

    Ron, if you're reading this message, I think this part of the book which causes so much confusion would be better understood if you addressed the above items. I've followed both threads here and in DPReview and kept silent 'till now. I wanted to provide positive suggestions instead of complaining about the shortcomings of the teaching technique.

    Rich
     
  14. Disclaimer:

    I speak only for myself. In no way, shape, or form do I claim to have any inside or enhanced knowledge of Ron's techniques, and I'm sure he'll tell me to shut up soon if I keep this up... :)

    First, when I read back over Ed's questions that prompted my last post, I realized that I didn't respond to one of his questions. Specifically, what happens if the gamma isn't set "correctly"?

    In my experience, pushing the gamma too high (ie. sliding the control to the left) tends to make an image too "muddy" -- the relative color levels in the mid-tones get brightened too much, which generally does two things: makes the image looked "washed out" (losing contrast and detail), and the colors start to look de-saturated. Pushing it too low will make the image generally darker, and increase the contrast between the lightest areas and everything else. Experiment a bit with it and you'll see what I mean. Very rarely will either extreme look "good".

    Here's where I'm probably going to get myself in big trouble: I suspect that the ranges Ron has suggested for low-medium-high gamma adjustments have come simply from observation of thousands of pictures (taken with specific cameras) during processing more than any inherent technical underpinnings that dictate using these levels of adjustments on all photos and equipment combinations. This implies to me that gamma adjustment is, to some extent, a correction that you apply to an image to account for assumptions that are designed into the camera itself (perhaps to optimize dynamic range, for example). Part of the reason that I say this is that I'm starting to suspect that the D2x has some differences in how it captures the mid-tones of an image. Maybe it's just me, but I'm finding that D2x shots don't seem to need as much gamma pushing as my D100 shots did (on average). That's not based on any scientific testing, just early observations with snapshots. I am definitely curious as to whether other D2x users are seeing similar results.

    Rich:

    You made some wonderful suggestions. Although from the sound of things I doubt Ron will have the time in the near future to work on "eBook Rev. 3", your ideas would improve the flow (IMHO). :)
     
  15. PJohnP

    PJohnP

    Feb 5, 2005
    Re: Well...

    Rich :

    Without question, there's a lot of 'art" to the curves management process. Like many, I have some problems with spotting the gamma points, ramps, inflections, and such like.

    And I have degrees in engineering and science, so graphical representations are not exactly a foreign concept for me.

    However, after my second class with Sensei Ron, I sat down and made up a flowchart of the process. Unless Ron has an objection, I'd be pleased to share this, and then to modify and update this with the ideas of Cafe members. The flowchart doesn't address all of the definition issues, but it offers a workflow concept.

    Note that this is my take on Ron's approach and that he's not responsible for my idiocies.

    However, the chart's in either Visio or PDF format.

    Let me know how I can upload this to the forum (my photosite, PhotoShare, doesn't have an option for such files, and I don't have a website currently), and I'll do this after I get from Big Bend NP this weekend.

    And with that, I'm heading out for the weekend. See all of you around here next week.



    John P.
     
  16. I've noticed in a lot of situations that if you hit people with too much at once they have more trouble handling the material than if you give the information in smaller amounts with examples, allow them to learn that as a base level, then build upon it.

    You should know the several methods of WB adjustment to allow you to correct images in as many situations as possible. When there is a neutral it's easy to do using either direct Kelvin adjustment or Set Grey. When there isn't a neutral it's extremely useful to know about the trick of looking for maximum amplitude and optimum symmetry. I know that some folks don't like to be reminded that I've processed a tremendous number of images, but the point of that is the experience has allowed me to recognize that when WB is correctly adjusted based on the former conditions the latter conditions (when data is present in the histogram) are also achieved. This is how I found out about that trick. Why not pass it on?

    Balancing the shadows allows you to maximize contrast with a minimum of shadow blocking.

    Gamma is midtone luminosity. You have to increase gamma from the arbitrary "Low" level when dynamic range limitations of the camera cause the midtones to be buried in shadows (or the image is radically underexposed).

    Gamma adjustments require the ability to analyze the histogram if you want to do it in a manner that is rapid and repeatable. You can often do it by using a calibrated monitor and evaluating the overall character of the image, and in some cases this is the best option (if the indicators are compressed into the shadow data due to the lack of midtone information). This is less repeatable. By defining gamma locations based upon a careful adjustment of gamma using the visual method, to my best abilities to accurately represent the scene as I saw it, then analyzing the histograms of many, many images, I learned that in situations where certain details in the histogram were present, the correct gamma position (based on my best recollection of the scene, often right after shooting it) was aligned with one indicator or another. This is sort of like the WB situation outlined above. If you can see these indicators and you know that the correct gamma is almost always aligned with them, it's faster and easier to align the gamma with the indicator. I gave you the shortcut that took me years to find on my own. I had several years to figure it out. After examining that many histograms it's easier for me to see these indicators than it is for someone who is just beginning to study them. I've seen how a gamma plateau flattens out into a little bump, then a point, or a stronger plateau becomes a ramp. I've seen how ramps in the shadows align with indicators in the midtones. I can rapidly see where these would be just because of the physical alignment of small irregularities in the histogram... primarily because I've seen so darned many histograms. I passed on the shortcuts so you will know what to look for, but some experience and practice is required to gain enough familiarity with histograms to make the indicators easy to see. Even then, it can be difficult when the midtones are compressed into the shadows, or downright impossible if the midtones are severely compressed. Then, it's useful to know the ranges that were developed by seeing that a vast number of images tended to fall within one range or another.

    I wrote this eBook because people were begging me to do so for months, and because there were no written precedents for some of the material the best thing I could do was attempt to explain things in a way that would not overwhelm the reader but still get the information across. I am aware that a better job can be done in explaining this material, Rich. If nothing else, these discussions have pounded that into my brain.

    It may put some people off when you say that you have a statistical database of xxxxxx number of images to build upon, but if something holds true many, many, MANY times it's silly to ignore that experience. If you don't have that experience, but someone tells you the shortcuts learned from gaining that experience, it gives you a leg up and reduces the learning curve. That's what I was trying to do with the eBook. Give you a leg up. I didn't explain things as well as I could. If I decide to rewrite it again in a few months I'll do my best to explain them better (some of the things said on DPR have been making me consider using the time for other things and letting the eBook disappear).

    Ron
     
  17. MontyDog

    MontyDog

    Jan 30, 2005
    #1064 - You have an error in your SQL syntax;
     
  18. Paul, I really like people and teaching them is rewarding for me, but this recent fiasco has been difficult to take. It has also destroyed eBook sales. If that keeps up I have zero incentive to spend 5 or 6 solid weeks rewriting it at the expense of losing all kinds of other business. Teaching time itself has to be supported by a tremendous amount of unpaid email, organization, etc., and has yielded a grand total of $16 per hour. Based upon a careful analysis last night, I've made a grand total of a little over $11 per hour on the eBook, not counting time spent answering emails and telephone calls. It certainly does seem as if it's not worth the effort, except when someone like you tells me how much the time spent has helped them enjoy photography and improved their results.

    I like putting smiles on people's faces. I like hearing how their work has improved. I really like seeing the results and looking back to see how much their work has improved. It makes me feel good. This recent fiasco hasn't though.

    Ron
     
Loading...
Similar Threads Forum Date
8 or 16 bit - Ron Reznick says save 8 bit. General Technical Discussion Sep 13, 2006
Ron Reznick and WB - help General Technical Discussion Jul 18, 2006
Nikon Capture Question (from Reznick e-book) General Technical Discussion Nov 9, 2005
By George (or Ron Reznick) I think I've got it! General Technical Discussion Apr 2, 2005
ATTN; MR. RON REZNICK General Technical Discussion Mar 11, 2005