Switch to English Language Passer en langue française Omschakelen naar Nederlandse Taal Wechseln Sie zu deutschen Sprache Passa alla lingua italiana
Members: 70,000   Posts: 1,524,324   Online: 738
      
Page 7 of 12 FirstFirst 123456789101112 LastLast
Results 61 to 70 of 117
  1. #61
    Bill Burk's Avatar
    Join Date
    Feb 2010
    Shooter
    4x5 Format
    Posts
    3,157
    Images
    46
    I don't think Giorgianni's "fourth quadrant" is inverted. Highlights are lower right on all the charts.

    I think the difference may come purely from the original preferred print studies being black and white and Giorgianni's study is color. There are probably different viewer preferences at play. (Nobody likes washed out color, and most people also don't like dark black and white prints).

  2. #62
    Stephen Benskin's Avatar
    Join Date
    Jan 2005
    Location
    Los Angeles
    Shooter
    4x5 Format
    Posts
    1,218
    The principles are about the same for all reflective material. This example isn't the easiest to read, but it shows how similar preferred reproduction curves for color and B&W prints are.

    Click image for larger version. 

Name:	Tone Reproduction of various materials.jpg 
Views:	6 
Size:	1.07 MB 
ID:	65435

    And I'm sorry, the inverted graph matches almost perfectly with mine and the classic preferred curve. Highlight separation is sacrificed for the mid-tone contrast. The tones are lighter than the original scene. It's easiest to miss mistakes in graphs, charts, and with math when proof reading.

    Does this look any better?

    Click image for larger version. 

Name:	Reproduction curve comparison a.jpg 
Views:	14 
Size:	503.8 KB 
ID:	65437
    Last edited by Stephen Benskin; 03-09-2013 at 01:21 AM. Click to view previous post history.

  3. #63
    Bill Burk's Avatar
    Join Date
    Feb 2010
    Shooter
    4x5 Format
    Posts
    3,157
    Images
    46
    Right! Those two graphs are very similar. The difference between them is the slight bias towards a "lighter" print for what I would call the black and white preferred print study. How much "lighter"? The difference you pointed out by drawing a 45-degree line in red.

    Here I am guessing the newer study of color reveals a preference for higher correlation to reality (in the mid-to-near-highlights), than in black and white which several authors have said should be "lighter". It makes sense to me that people would need "color" to be more accurate.

    I also know people prefer my lighter prints, though I occasionally choose darker for my own artistic reasons (as an artist, each of us is entitled to break the rules any way we want).

  4. #64

    Join Date
    Aug 2006
    Shooter
    Multi Format
    Posts
    321
    Quote Originally Posted by Bill Burk View Post
    I don't think Giorgianni's "fourth quadrant" is inverted. Highlights are lower right on all the charts.
    Exactly! The Giorgianni chart, as is, is directly comparable to Stephen's 4th Quadrant chart.

    I think the difference may come purely from the original preferred print studies being black and white and Giorgianni's study is color.
    I don't think so, other than the color materials seem to generally reach a higher max density.
    (Update: amended per my following post #65; I acknowledge that skin tones in a color print are not allowed to get as light as B&W can, because the color will be lost as paper-base white is approached. So, except for specular highlights, fleshtones should not be allowed to go light enough to have washed out color.)


    Quote Originally Posted by Stephen Benskin View Post
    Highlight separation is sacrificed for the mid-tone contrast. The tones are lighter than the original scene.
    The part I don't agree wih is "The tones are lighter than the original scene." I think this idea may be so fixed in your mind that you aren't hearing what I'm saying. The Giorgianni graph shows ALL of the print tones as either the SAME or DARKER than the original scene's "equivalent densities."


    Does this look any better?

    Click image for larger version. 

Name:	Reproduction curve comparison a.jpg 
Views:	14 
Size:	503.8 KB 
ID:	65437
    Not Giorgianni (look at the crazy print densities you're giving him - ~ 0.35 "white" to ~ 2.80 Dmax). You should be using the original chart, as is.

    On your (4th Quad) graph, the added (red) reference line looks good - it now serves as a legitimate "original scene" density reference. My only problem now is that your tonal curve does NOT allow for specular reflection. It ends (lower right) at zero on the "Reflectance" scale, which I take as "equivalent density" of zero in the original scene, that is, 100% (diffuse) reflectance.

    In the original Giorgianni chart, you can see that the lower right (the highlight areas) extends about 0.30 log exposure units beyond "pure white" (which is "0" on the "Log scene luminance factor"). This allows about 1 f-stop equivalent of highlight detail to be reproduced, in a compressed manner, on the print.

    This ability to reproduce a nominal f-stop of specular reflection is the only (significant) difference remaining between the two graphs. Yes, there is some difference in the shape of the curves, but there is room for artistic intent there.

    FWIW, I have some experience in building tonal curves for use in portraiture. I found that I had to closely mimic the lower portion of the Giorgianni chart (the transition from specular reflections until meetup with the reference line) in order to handle both specular skin reflections and the slight folds and texture of a white dress shirt. I did a lot of fine tuning, but there is not much leeway for variation. Other subjects, such as landscape scenes, may find modification of this curve to work better, perhaps as in Stephen's graph (with some specular highlights built in). I can't say for sure.
    Last edited by Mr Bill; 03-09-2013 at 08:09 AM. Click to view previous post history.

  5. #65

    Join Date
    Aug 2006
    Shooter
    Multi Format
    Posts
    321
    Quote Originally Posted by Bill Burk View Post
    Right! Those two graphs are very similar.
    Bill, you're being taken in! Giorgianni has been turned end-over-end. Look at the print density axis, base density of about 0.30, more like a mid-gray, reflecting only half the light. On the high end, Dmax ~ 2.80? I've never seen a reflection print exceed ~ 2.50. The reference line isn't passing through 0, 0 anymore. The original graph needs to be used for these comparisons; it was already in the appropriate configuration.

    Here I am guessing the newer study of color reveals a preference for higher correlation to reality (in the mid-to-near-highlights), than in black and white which several authors have said should be "lighter". It makes sense to me that people would need "color" to be more accurate.
    Well, I would concur that color prints can't reproduce a flesh tone as light as B&W can, for the simple reason that the color saturation would go away. If you don't get some minimum amount of dye formed, the skin color will become "washed out." Whereas we don't see this, visually, when looking at real people.

    I think there's a bit of leeway talking about flesh tone reproduction in B&W. First, I don't know how to really measure flesh tones, expressed as B&W density, on a person. Well, I really sort of know how, but the translation to film density has some dependence on the film's spectral response, as well as the lighting used. So it could run overly light or overly dark, depending on conditions. In other words, I dont' know that one can precisely specify what an "exact" density match means.

    So yes, color has less leeway for interpretation.

  6. #66

    Join Date
    Aug 2006
    Shooter
    Multi Format
    Posts
    321
    This is getting exhausting! I finally looked through Theory of the Photographic Process, the later edition (James) as referenced by Mark Barendt.

    This idea that "the preferred print has less density than the subject" seems to be at the heart of these disagreements. Anyway, I DID NOT find this stated in Theory... However, it is possible to misinterpret this from a graph if one doesn't notice the base point of the reference lines.

    There is one place, page 539 (James' edition), that says, "It is significant that the densities of skin tones in the preferred prints were found to be equal to or slightly lower than the densities of real skin" [emphasis is mine]. Again, "equal to," or "slightly lower." In the Giorgianni graph, they are shown as roughly equal, as I pointed out in post #55.

    To conclude, I don't find any statements to the effect that a print needs to be lighter (less dense) than the original. (I agree there are cases where this works well, just not that it is stated to be a requirement for preferred prints.)

  7. #67
    Stephen Benskin's Avatar
    Join Date
    Jan 2005
    Location
    Los Angeles
    Shooter
    4x5 Format
    Posts
    1,218
    From Todd and Zakia’s Photographic Sensitometry: The Study of Tone Reproduction p. 286:

    “The densities of the aim reflection print are about 0.25 below the 45 degree reference line. The slope of the midtones is slightly greater than 1.0 – between 1.1 and 1.2. Prints with midtone slopes less than 1.1 are characterized by viewers as flat. The slope of the extreme shadows is nearly 1.0. These observations confirm the belief that shadow detail must be maintained for such (pictorial) subject matter if the print is to be satisfactory, whereas highlight detail is far less important.”

    “For portrait photography, the aim curve is to the right of the one shown in Fig XIII-3 and the slopes are increased in the highlight and midtone region, at the expense of the shadow reproduction. No doubt this change arises from the importance of skin tone reproduction.”

    Click image for larger version. 

Name:	Tone Reproduction - Photograhic Sensitometry.jpg 
Views:	2 
Size:	316.1 KB 
ID:	65453

    Right is darker and left is lighter.

    It would appear general pictorial and portrait photography require preferred reproduction curves that are almost opposite in appearance. So if the preferred curve in Digital Color Management is for portraiture, that would explain why I thought it looked inverted.

    The reference line in Digital Color Management does reflect the original subject values, but the reference line used with my program and in most books, isn’t exactly meant for the same purpose. Strobel et al Photographic Materials and Process, p. 356. “The 45 degree reference line is arbitrarily located so that its lower point corresponds to a diffuse white object in the scene and the minimum density of the photographic print material. If the luminance in the scene were produced exactly in the print, the resulting tone reproduction would have a slope of 1.00, matching that of the reference line. In fact, the average curve for first-choice prints is located 0.2 to 0.3 density units below the 45 degree line except in the highlight region, where the curve cannot go below the minimum density of the paper. The curve shows low slope in the highlight and shadow regions, but has a gradient of 1.1 to 1.2 in the midtone area. This indicates the print’s highlights and shadow regions are compressed compared to the original scene, while the midtones have been slightly expanded in contrast. The results in all cases were remarkably similar to that shown in Fig 11-1. Whenever the departure from the desired slopes was greater than 0.05, the observers invariably judged the prints as being unacceptable in contrast, indicating very narrow tolerance levels. Consequently, it appears that the desired objective tone reproduction in black and white reflection print are essentially independent of the characteristics of the original subject.

    Click image for larger version. 

Name:	Tone Reproduction - Photograhic Materials.jpg 
Views:	2 
Size:	312.0 KB 
ID:	65454 Click image for larger version. 

Name:	Tone Reproduction - Photograhic Materials 4 quad.jpg 
Views:	2 
Size:	707.1 KB 
ID:	65455

    And from Photographic Theory of the Photographic Process 3rd ed, p 467: “Whenever the density level of the prints was great enough so that the curves closely approached the 45 degree reference line, the prints were unanimously rejected because they appeared too dark.”
    Last edited by Stephen Benskin; 03-09-2013 at 09:56 AM. Click to view previous post history.

  8. #68
    Bill Burk's Avatar
    Join Date
    Feb 2010
    Shooter
    4x5 Format
    Posts
    3,157
    Images
    46
    Quote Originally Posted by Mr Bill View Post
    Bill, you're being taken in! Giorgianni has been turned end-over-end. ..
    So yes, color has less leeway for interpretation.
    Fortunately I cemented my idea on the original Giorgianni chart, and the assertion it supports that the print should be darker.

    I still have faith in the old chart that I've seen over the years.

    I'm inclined to believe more deeply that the main difference is what color needs versus what black and white needs.

    I can accept that Giorgianni says "make the upper-mid-to-highlights accurate". From there, you run parallel to the preferred curve, however you "had" to go darker to maintain linearity.

    While I still believe black and white needs lighter "upper-mid-to-highlights". And then from there you "have" to continue roughly linear toward the shadows.

    I was intrigued by reading a book about Polaroid, Instant by Christopher Bonanos, which described Land's Retinex experiments. I think "Color Constancy" has a lot to do with this.

  9. #69
    Stephen Benskin's Avatar
    Join Date
    Jan 2005
    Location
    Los Angeles
    Shooter
    4x5 Format
    Posts
    1,218
    Here’s an interesting experiment with the reproduction of the original luminance values in the print. Both 4 quad curves represent an average scene luminance range of 2.20 with one stop of flare. The negative, processed to CI 0.56 has a negative density range of 1.03 which closely matched the print’s LER of 1.06.

    The red highlighted reference line in the first example indicates the metered exposure which is at 12% reflectance. But notice how the original subjects tones at 12% reflectance become 18% in the print. The value of the metered luminance is lighter in the print.

    Click image for larger version. 

Name:	4 quad - normal 12 percent.jpg 
Views:	4 
Size:	521.7 KB 
ID:	65468 Click image for larger version. 

Name:	Reproduction curve for 12 percent.jpg 
Views:	3 
Size:	255.8 KB 
ID:	65470

    Now, what would happen if something is metered slightly lighter than 12% reflectance like an 18% gray card and then printed it to a reflection density of 0.74?

    Click image for larger version. 

Name:	4 Quad, 18 percent target.jpg 
Views:	3 
Size:	551.2 KB 
ID:	65471 Click image for larger version. 

Name:	Tone Reproduction - 18 percent keyed.jpg 
Views:	3 
Size:	128.2 KB 
ID:	65472

    The print tone matches the original luminance value, but not without a cost to the reproduction curve. Interesting enough, from what I’ve recently discovered, this curve looks similar to the preferred portraiture curve.

  10. #70

    Join Date
    Aug 2006
    Shooter
    Multi Format
    Posts
    321
    You guys are getting ahead of me, but here's what I've been writing, on and off.

    [Todd & Zakia] “The densities of the aim reflection print are about 0.25 below the 45 degree reference line..."
    Ah, therein lies the rub, where is the reference line? And further, why is it there? Is it as a reference for the slope, or as a hard density reference? Although Todd and Zakia describe (roughly) where the tonal curve is with respect to the reference, I think they are just describing their published graph, not saying that it is an absolute. (If they wanted a hard density-to-density comparison, they would not have based their reference on paper-base white, would they?)


    Below is the only thing I take exception to, from post #42
    Along with the mid-tone gradient, one other critical condition is required. The print tones need to be lighter than the original subject.
    I don't believe this is authoritively stated in the literature, and it is an untenable position to say that it is crucial in all cases.


    Giorgianni's (and Madden's) book isn't about portraiture, the few sample images include scenic views. They are just establishing the general tone reproduction requirements for video, transparencies, reflection prints, and negatives. So I think that they are showing the same fundamental tone reproduction principles as Theory..., but with less rigor. I don't see that the graphs you've posted vs Giorgianni's are fundamentally different, only the strength of the curves.

    By the way, even the portrait curves don't always behave as shown. They work well for a light-skinned subject, where many tones print darker than the original. But in the case of a dark-skinned subject, normal procedure is to print lighter, until the flesh tones come up to a mid-range. So in that case, despite the graph I showed, most tones will print lighter than the original. (By the way, I also designed a different response curve where I didn't let the slope climb until mid-way; this worked very well for dark-skinned subjects with a white shirt or blouse. So this tone curve was visually very different, although it followed the same basic principles)

    I think the same situation will exist in landscape photography and other situations. The same basic tone curves always work, but sometimes they should print darker and sometimes lighter than the original.

    Photography is loaded with things like this, where initially it seems very clear how something should work. But when examined very closely, things get much deeper than first met the eye.



 

APUG PARTNERS EQUALLY FUNDING OUR COMMUNITY:



Contact Us  |  Support Us!  |  Advertise  |  Site Terms  |  Archive  —   Search  |  Mobile Device Access  |  RSS  |  Facebook  |  Linkedin