Skip navigation

Develop basic sliders - truly zeroed?

Jan 10, 2012 5:43 AM

  Latest reply: Cornelia-I, Mar 1, 2012 12:31 PM
Replies 1 2 3 4 Previous Next
  • Andrew Rodney
    1,388 posts
    Apr 16, 2009
    Currently Being Moderated
    Jan 22, 2012 1:17 PM   in reply to Butch_M

    Butch_M wrote:

     

    I saw it ... it's absence is of little consequence ... IMHO

    I saw it, it was a stupid thing to have posted and could easily have violated an NDA. Either he got smart and deleted it himself or someone else did.

     
    |
    Mark as:
  • Currently Being Moderated
    Jan 22, 2012 1:18 PM   in reply to johnhawk666

    OK, then it'll be Simon himself.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 20, 2012 5:50 PM   in reply to TK2142

    TK, I probably should just ignore this, but to me it seems that your analogy about the pool is't quite right.  You're not looking at the pool, seeing it's empty, and deciding not to jump in.  You're refusing to go outside because the pool might be empty. While Jeff's approach may be a bit terse, I have to agree that your totally negative reaction, based on zero knowledge of how the product actually works in practice, is a bit over the top.

     

    As for a truly "zeroed out" starting point, I don't get it.  I went into LR3, set all the editing controls to zero, and had a totally useless starting point.  The image was so dark as to be useless. It would take me more effort to start from there than from LR's starting points. Regardless of what's going on under the covers, LR makes it easier to get quality output from a raw file than any other program I've used, and based on the little bit of time I've spent with the beta, LR4 makes it even easier. To me, that's what it's all about.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 20, 2012 9:25 PM   in reply to bobmeyerweb

    jrsforums originally asked "Develop basic sliders - truly zeroed?"

     

    What about the photograph? are we not after the best possible image? Regardless of where our starting point is on a piece of software that is so powerful and complex and capable of making our photos something better than the previous iteration of itself?

     

     

    There are very few absolutes in nature, in life and in reality.

    • Even, -273.15º Kelvin where matter stops to move, is a theoretical zero.
    • e is another constant (maybe on this planet alone)
    • Planck has a constant, but that can only be measured so far, then errors come in from the actual measuring itsself.
    • Pi is a constant 3.1415..... when based on standard Euclidean Geometry. When measuring the pi on a curved surface, ie drawing a circle on a ball, the constant value of pi, call it the absolute zero is different.

     

    • There is absolute vodka, but that is something needed after some of the comments in the above thread.

     

    All of these are constants but relative to something, can any of them be truely zeroed?

     

    Its been mentioned above, that the 0 on highlights, exposure, shadows may well say 0 to begin with,  but they are zero relative to something, maybe the exposure that you took your photograph at, or maybe the colour / shade / hue of the subject matter of the photograph.

     

    As has also been said above, if you make your own profiles or presets, these set various points to a setting and then LR works as those as zero, so my LR4 shows zero for exposure, with my applied profile it may be 1/2 a stop underexposed, with a black point at +5, but the basic sliders say, they are zero. - I've set my own zeros.

     

     

    If you take that "absolute" zero and work from there, not pushing your shadows too high to introduce too  much noise, or similar for exposure, you should have a good picture. - zero defects?

    If you push your blacks or whites too high or low from that relative zero, you will potentially have an image that has over compressed blacks or whites and unless that is a style you are after, its probably not going to look too good. 0/10

     

     

    Most of the users of Lightroom are striving for their best possible photograph, whether  they are using their computers in mixed lighting, on a train, or like a few, in a perfect envrionment with zero distraction, zero incorrect colour balances, and perfect 20/20 vision.

    Then there photographeers with astigmatism and -3 myopia. Zero to them is relatively different, all their zeros or starting points are all different.

     

    I knew a bloke who shot for a whole day with sunglasses on, and exposed relative to what he saw through his sunglasses - true story. every frame a stop over exposed.

     

    At the end of the day, none of my settings may have a zero near them, but I've got a great image. That for me is ground zero, that is where I want to be.

    I'm after truly zero complaints from happy clients, my bills paid, food on the table and most defiintely not zero in my bank.

     

     

    hamish NIVEN Photography

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 20, 2012 11:25 PM   in reply to jrsforums

    What a strange argument being made here for a zero'd or "neutral" starting point. Has there ever been such a thing in the history of photography, regardless of the process?

     

    Reminds me of early arguments made against digital, that only film gave a true and accurate representation of the scene. Customers would insist that a photograph shot on film was somehow completely accurate because it hadn't been "messed with" in the computer.

     

    As if!

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 20, 2012 11:34 PM   in reply to wtlloyd

    wtlloyd wrote:

     

    What a strange argument being made here for a zero'd or "neutral" starting point. Has there ever been such a thing in the history of photography, regardless of the process?

     

    Not realy...but it depends on whether you were a chrome vs neg shooter. If you shot color neg, then EVERYTHING was an interpretation of the original neg, If you were a chrome shooter then things were a bit altered...you tended to accept (and cling to) the chrome rendering...

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 12:35 AM   in reply to hamish niven

    hamish niven wrote:

     

    What about the photograph? are we not after the best possible image? Regardless of where our starting point is on a piece of software that is so powerful and complex and capable of making our photos something better than the previous iteration of itself?

     

     

    hi hamish,,

     

    startng point ?? yes,,  it's with the camera in hand !! that's the blank slate…it's as the canvas without it's first spash of oils… a vision only of the mind…what some photographers don't get and lean too heavily on technology to make…..so we are all schooled now on what is happening with lr4…so,,, start with the 'given' and move on !!!  use curves!!!

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 1:55 AM   in reply to decoyle

    decoyle wrote:

     

    hamish niven wrote:

     

    What about the photograph? are we not after the best possible image? Regardless of where our starting point is on a piece of software that is so powerful and complex and capable of making our photos something better than the previous iteration of itself?

     

     

    hi hamish,,

     

    startng point ?? yes,,  it's with the camera in hand !! that's the blank slate…it's as the canvas without it's first spash of oils… a vision only of the mind…what some photographers don't get and lean too heavily on technology to make…..so we are all schooled now on what is happening with lr4…so,,, start with the 'given' and move on !!!  use curves!!!

     

    use curves?

    use camera - properly and intelligently

     

    You can rescue a mediocre photo , but it will only very seldom be a good result

    but

      rubbish in = rubbish out

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 8:46 AM   in reply to hamish niven

    I assume that one of the focal points for this thread was mentioned in the sibling thread (http://forums.adobe.com/thread/948951?tstart=0 ):

    "Having adopted Michael Frye's workflow for landscape prints from his excellent eBook (http://craftandvision.com/books/light-and-land) this comes as a blow."

     

    I had not become really adept with LR3's basic sliders neither, so his sort-of-recipe was a welcome revelation for me:

    To get repeatable results, where different settings will not counteract unintentionally, set black + brightness + contrast to zero, tone curve selection to linear.

    Start from this rather flat dull image by going to the tone curve in point editing mode (all parameters also zero, of course): pull the black point in shortly before black clipping would start, dito with the white point in the right upper corner before highlight clipping would start. If this steeper line does not yet give you the desired punch make a darkening S-curve out of it, most of the time by just adding 2 more points.

    If needed adjust Recovery and Fill light from the basic panels, as you cannot reproduce their effect solely from the tone curve.

    [Note: This is MY short interpretation of Michael Frye's book. I hope he will not be astonished about too much misinterpretation, but this is within the realm of possible!]

     

    So what do you do when you'd like to still follow this recipe in LR4 with the sliders Fill Light and Recovery gone? When the rest of the former starting point is convoluted to restore?

     

    I opted for sheer trying... And found out, that often just by using the new Exposure and Contrast slider I achieved the same as before.

    For other fine-tuning and to rid my mind of seemingly *contradictory* slider settings other threads in this forum have helped, even if they have been at least as impolite as these two threads at times.

    So my notion of netiquette in this forum is sometimes not met,

    but I might get adept with basic sliders/LR4 at last...

     

    Thanks to all you nice guys,

    Cornelia

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 8:34 AM   in reply to hamish niven

    <annoying nerd mode>

     

    Hamish, I thought it was funny that in your list of non-absolute absolutes, you listed several things that were actually as absolute as things get:

     

    • 0 K (not -273 which is the number in Celsius) is a definition. It is not theoretical. No universe is possible in which there is not a zero point in temperature (kinetic energy/entropy). It would be a inherently unstable universe. As with all ideals, it is unreachable but its absoluteness is undisputable. Absolute zero is less theoretical than the existence of gravity.
    • e is a mathematical/geometrical construct. It is defined purely mathematically as the limit of (1+1/n)^n when n goes to infinity. This is again as absolute as it gets. On different planets the rules of mathematics/logic don't change.
    • Pi is a gemoetrical construct defined by gemoetry/logic/math and as such absolute. It is not a fuzzy concept at all and not something that you can mold to your desires. That you start with the construct of Eucllidian geometry is not something that makes it less "absolute"

     

    Now for the not-so-absolutes:

     

    • The Planck constant (usually no apostrophe is used) is very appropriate to this discussion. It is the proportionality constant between the energy and the frequency of a photon. It also gives, in the famous de Broglie relation, the proportionailty between the momentum and the wavelength of any particle, capturing the fundamental dual wave/particle nature of matter. While indeed it can only be measured, it represents one of the most fundamental insights in our uderstanding of nature. A universe with a different Planck constant would be ... interesting.
    • Absolute vodka is far from absolute. There are still way too many impurities in the ethanol ;-) Absolute ethanol even is not entirely pure.

     

    </annoying nerd mode>

     

     

    Anyway. I agree with the rest of your post. The point of the raw converter in software like Lightroom is to get a pleasing result as fast as possible. It is not meant to give a physically correct representation of the scene in front of the camera.

    Why not physically correctly interpret the data (even if this were really possible)? See the grey squares illusion. Perception is not reality.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 11:03 AM   in reply to hamish niven

    hamish niven wrote:

     

    use curves?

    use camera - properly and intelligently

    well the comment "use curves" was to be taken facetiously,,thus the ,, the thread seemed at times to be about the illusion of the "linear curve"

     

    the serious side was that the end result of editing, no matter the initial conditions of lr, is the vision of the photographer from the outset... get what you can from the camera and lr edits and create that vision.

     

     

     

    You can rescue a mediocre photo , but it will only very seldom be a good result

    but

      rubbish in = rubbish out

     

    not an abslolute…. maybe.. rubbish in => methane out   seemingly a contradiction to the absolute of entropy

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 11:07 AM   in reply to decoyle

    there is some inspired commentary in this thread,

    enjoying the entertainment as well as the seriousness of it.

     

     

    decoyle wrote:

     

    hamish niven wrote:

     

    You can rescue a mediocre photo , but it will only very seldom be a good result

    but

      rubbish in = rubbish out

     

     

    not an abslolute…. maybe.. rubbish in => methane out   seemingly a contradiction to the absolute of entropy

     

     

     
    |
    Mark as:
  • TK2142
    407 posts
    Jan 20, 2010
    Currently Being Moderated
    Feb 21, 2012 11:50 AM   in reply to Jao vdL

    Jao vdL wrote:

     

    The point of the raw converter in software like Lightroom is to get a pleasing result as fast as possible.

    I agree with the rest of your post but not quite with this statement. I don't want "a" pleasing result.

    I want a result that expresses my vision.

     

    Having to first fight default develop settings before I can start to develop my envisioned result is not helpful.

    Not knowing what kind of auto adjustments are being performed to what extent is not helpful (for my style of editing).

     

    Surely, I'll arrive at some version of the image by tweaking from an unknown starting point by using sliders with unknown effects. I personally would prefer to exactly know my starting point and exactly know what I'm doing with each slider (and with what amount of effect).

     

    BTW, the concept of "neutral" is not really that complicated as some people present it. A tone curve, for example, is either linear or it isn't. Only the linear form is neutral.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 12:13 PM   in reply to TK2142

    TK2142 wrote:

     

    BTW, the concept of "neutral" is not really that complicated as some people present it. A tone curve, for example, is either linear or it isn't. Only the linear form is neutral.

     

     

    See, that's where there's a problem...a truely linear tone curve from a raw file isn't really useful. The image below was processed out using a command line control of the DNG converter. It has had the demosiacing done but the tone curve is still an absolute linear curve. This is what a raw file looks like...

     

    2-04a-CR-Lin-processed.jpg

     

    And below is a default rendering from Camera Raw 6 in Photoshop CS5.

     

    2-04b-CR-processed-TC.jpg

     

    So, is the real linear file useful? I wouldn't call it "neutral". You need to apply a tone curve just to see the image.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 12:40 PM   in reply to TK2142

    What I mean with pleasing is pleasing to the person driving the converter, which is you as the photographer/artist. Expressing your vision is just a different way of saying that of course. What a raw converter should do is to allow you to reach the desired end result as efficiently and intuitively as possible. Apart from that rather generic statement, oftentimes tools become means in themselves. A good tool makes results possible that were not possible before - it's an enabler. They sculpt your vision. No tool is really dislodged from creative vision.

     

    Regarding the "linear tone curve", Jeff already answered this. You do not want linear. Although I do have to note that Jeff is cheating a little by not setting the white point on that linear conversion. If you do, it doesn't look quite as bad. Still terrible, but not as horribly dark. That said, as far as I know, there is no current commercial raw converter that maps scene luminosity directly using a linear scale to the same (or normalized between 0 and 1) luminosity in the output or reference space although probably dcraw can be tricked into doing that (and probably acr too if you know the appropriate commands). There is always some tone curve behind the scenes even if it presents you with a linear tone curve in its interface. In LR 3, the medium contrast tone curve that was presented was actually the neutral one giving you the rendering inherent to the camera calibration you selected and selecting linear in that interface gives you a toned down (sorry for the pun) version. LR 4 with PV 2012 provides less confusion by giving the "neutral" calibrated rendering with the linear tone curve selected. It is the same rendering (or approximately so considering some of the perceptual mapping and automatic highlight recovery going in in PV 2012) as the medium contrast curve in LR 3.

     

    Message was edited by: Jao vdL to remove email mangling.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 12:44 PM   in reply to Jeff Schewe

    Certainly not usefull in its basic form, but useful insofar as a - shall we say - "unbiased" (for lack of a better word) starting point is concerned.  If one should desire to go that route.  Fun and happy times on a bolus of images?  No.  But after editing and finding a shot that deserves special attention and wanting to run the numbers from scratch certainly.  If that's what your workflow (I again refer back to Michael Frye's technique here) requires.  It's not for everyone and nor should it be.

     

    I don't know how many times it's been said but I for one have never suggested that all the default renderings should look like this, that there is anything wrong with the interpreted develop that mimics the out-of-camera JPEG (that progress was worth the LR2 upgrade alone).  All I and others have suggested is that it would be nice to at least have this option for those who chose to process from scratch.  Clearly it can be done, albeit not by the casual user.  Turn that "DNG command line control" into a preset and this discussion is over and both sides can move on.

     
    |
    Mark as:
  • TK2142
    407 posts
    Jan 20, 2010
    Currently Being Moderated
    Feb 21, 2012 12:54 PM   in reply to Jeff Schewe

    Jeff Schewe wrote:

     

     

    See, that's where there's a problem...a truely linear tone curve from a raw file isn't really useful. The image below was processed out using a command line control of the DNG converter.

    When I say "linear tone curve" I do not mean a rendering of the RAW image without gamma encoding  (using a gamma value of "1" is the same as using no gamma encoding).

     

    Clearly, the display of images is predicated on the notion of gamma decoding at the display end and gamma encoding at the source end. Since RAW files contain a linear representation of brightness values, they need to undergo gamma decoding first, before they get a reasonable appearance. Looking at linear image data is akin to looking at a stream of characters where the value 64 has been substracted from every character value (not quite, it is just an analogy). This stream is not directly readable, unless I add 64 to every character first.

     

    I'm assuming that the image you posted has enjoyed no gamma encoding yet. That's not what I mean by "linear tone curve". The gamma encoding is necessary and standard (now that Apple also seem to have adopted 2.2 as an decoding gamma). There is no degree of artistic freedom on how to perform it.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 12:58 PM   in reply to TK2142

    Gamma = 2 and linear tone curve is NOT what you were getting with LR 3 either.

     
    |
    Mark as:
  • TK2142
    407 posts
    Jan 20, 2010
    Currently Being Moderated
    Feb 21, 2012 1:06 PM   in reply to Lee Jay

    Lee Jay wrote:

     

    Gamma = 2 and linear tone curve is NOT what you were getting with LR 3 either.

     

    Eric Chan seems to differ:

     

    === quote ===

    > Eric, can't you also get linear results (except for gamma
    > encoding) by simply setting all the values on the basic
    > ACR panel to zero and also setting the point tone curve to
    > zero?

    Yes, that works too and is equivalent to creating a "linear
    base tone curve" DNG profile using the DNG Profile Editor.


    === unquote ===

    (from http://forums.adobe.com/message/1210327#1210327, a discussion from pre LR4 days when "0" meant zero).

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 1:34 PM   in reply to Jao vdL

    Jao vdL wrote:

     

    Although I do have to note that Jeff is cheating a little by not setting the white point on that linear conversion. If you do, it doesn't look quite as bad. Still terrible, but not as horribly dark.

     

    Actually, as far as I know, when running the image through the SDK in command line, there is no fuction to set the white point...what the process does is simply save the image with no changes but demosiacing. So, this is the result of a true linear process. Even when setting the LR/ACR tone curve to linear and putting all other settings to zero, there is still a mapping going on–it's not truely linear. The mapping is set by LR/ACR based on the camera metadata and some scaling is going on. Running an image through the SDK doesn't, so I think it's truely linear (as far as I can tell).

     

    There's also a command for processing the file out with no demosiacing as shown below. You can still see the Bayer pattern. Note it's a tiny crop of the yellow center of purple flower.

     

    2-10b-undemosiaced-lin-flower.jpg

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 2:05 PM   in reply to TK2142

    TK2142 wrote:

     

    Lee Jay wrote:

     

    Gamma = 2 and linear tone curve is NOT what you were getting with LR 3 either.

     

    Eric Chan seems to differ:

     

    === quote ===

    > Eric, can't you also get linear results (except for gamma
    > encoding) by simply setting all the values on the basic
    > ACR panel to zero and also setting the point tone curve to
    > zero?

    Yes, that works too and is equivalent to creating a "linear
    base tone curve" DNG profile using the DNG Profile Editor.


    === unquote ===

    (from http://forums.adobe.com/message/1210327#1210327, a discussion from pre LR4 days when "0" meant zero).

     

    I was talking about LR 3 with defaults except tone curve = linear.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 2:20 PM   in reply to Jeff Schewe

    Actually, as far as I know, when running the image through the SDK in command line, there is no fuction to set the white point.

    You're right. I checked a little and ran a single file through dcraw using a completely linear tone curve and only a gamma correction using gamma 1/0.45 (the dcraw/ufraw default) and it comes out very dark just like your image. This is my daughter on the prow when heading out of Alameda Estuary.

     

    linear-vs-standard.jpg

    left is default PV 2010 with default settings and Adobe Standard, middle is PV 2010 with all settings zeroed (so non default but what people for some reason think gives "neutral" rendering) and "linear" tone curve. Right is a purely linear conversion in dcraw with only the standard gamma mapping applied. You see the actual linear rendering is far darker and contrasty than what people think is supposed to be "linear" or neutral in LR 3 (the middle or the left image).

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 21, 2012 3:15 PM   in reply to Jao vdL

    Jao vdL wrote:

    You see the actual linear rendering is far darker and contrasty than what people think is supposed to be "linear" or neutral in LR 3 (the middle or the left image).

     

    Yep...and I just recreated a linear curve in DNG Profile Editor-started with Adobe Standard, set the curve to linear and saved the profile. Then in LR4 I updated a raw file to PV 2012 and selected the linear profile in Camera Calibrations and set the Curve to linear, and with the image with zeroed settings in Basic, it matches the PV 2010 with a linear curve and zeroed settings. Be sure to quit and restart LR4 after saving the profile.

     

    So, if people want that old zeroed look, that's how you do it. Took all of 3 minutes. I had already downloaded and installed DNG Profile Editor. Course, you'll need to do that for each of your cameras...is it EXACTLY the same? Nope, but it's real close...the image adaptive zeroed setting actually tend to make the image a bit flatter.

     

    PV2010-linear-zeroed.jpg

    The image above was processed zeroed in PV 2010

     

    PV2012-linear-zeroed.jpg

    The image above was processed using a linear DNG profile with curve set to linear and zeroed settings in PV 2012.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 28, 2012 1:11 PM   in reply to Jeff Schewe

    Your "raw" image looks dark because it is presumably encoded with a gamma of 1.0 and has not been tagged with the proper profile (such as LinearRIMM-RGB) or is being viewed in a non-color managed application that does not recognize the profile. If you have a linear file so tagged and viewed in Photoshop, it will look normal. However, editing such a file can be problematic, since a gamma one space is not perceptually uniform.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 28, 2012 1:15 PM   in reply to Bill_Janes

    Bill_Janes wrote:

     

    Your "raw" image looks dark because it is presumably encoded with a gamma of 1.0 and has not been tagged with the proper profile (such as LinearRIMM-RGB) or is being viewed in a non-color managed application that does not recognize the profile.

     

     

    Actually, the raw images as shown above are the way the zeroed images with a linear curve look in Camera Raw and Lightroom. The images I posted here are indeed processed out as sRGB but that was simply to post them to the web.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 28, 2012 1:24 PM   in reply to Jeff Schewe

    Jeff Schewe wrote:

     

    Bill_Janes wrote:

     

    Your "raw" image looks dark because it is presumably encoded with a gamma of 1.0 and has not been tagged with the proper profile (such as LinearRIMM-RGB) or is being viewed in a non-color managed application that does not recognize the profile.

     

     

    Actually, the raw images as shown above are the way the zeroed images with a linear curve look in Camera Raw and Lightroom. The images I posted here are indeed processed out as sRGB but that was simply to post them to the web.

     

    Actually, I was talking about your reply No 92 where you posted an image rendered with the DNG converter using a command line as quoted below.

     

    "See, that's where there's a problem...a truely linear tone curve from a raw file isn't really useful. The image below was processed out using a command line control of the DNG converter. It has had the demosiacing done but the tone curve is still an absolute linear curve. This is what a raw file looks like...'

     

    I don't think that file is in sRGB.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 28, 2012 1:32 PM   in reply to Bill_Janes

    And my raw image converted using ufraw/dcraw above was displayed fully

    color managed and only as a final step converted to sRGB for web display.

    It looks exactly like the image looks in Photoshop on my display. The

    luminosity you see on the display is the luminosity encoded in the raw

    file. This is just what raw images look like when you're only doing a

    straight (but color managed) conversion without any tone curves applied,

    meaning that the luminosity in the output file and in the output color

    space is equal to the luminosity that the camera recorded. Now if you send

    the linear RGB values from the raw file direct to the display without gamma

    correction/color management, the image looks even darker (basically almost

    all black except for the extreme highlights) but that is not the reason why

    Jeff's and my image look dark.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 28, 2012 1:40 PM   in reply to Jao vdL

    Jao vdL wrote:

     

     

    I think you misunderstand how raw conversion works though. You start with a capture that is inherently linear. A CCD or CMOS chip is basically a photon counting device. The signal stored in the raw file represents the linear light intensity present in the capture. Since our eyes and brain operate completely differently and work logarithmically and dynamically interpret scenes, an actual linear presentation of this would look very very dark, like nothing you have ever seen. It also has a Bayer mosaic applied but let's ignore that. In order to get any reasonable representation, you have to both gamma correct and apply tone curves. In the ACR engine, a tone curve gets in the camera profile (Adobe standard or Camera profiles). This happens in a gamma corrected space. You can see the tone curve that gets applied at this stage by opening the profile in DNG profile editor and going to tone curve and checking "Show base tone curve". This tone curve is applied behind the scenes and different versions (depending on the profile you select in camera calibration) are ALWAYS applied. There is no profile that has a linear tone curve. They all have some curving to them (pun intended). This all happens before you even get to the tone curve in the Lightroom interface.

     

     

    Our vision is logarithmic as you state, but images are not gamma encoded to compensate for the characteristics of human vision but rather to allow more precision in encoding the shadows and to achieve a more percetually uniform space for editing. An 8 bit linear file would show posterization in the shadows.

     

    When the image is viewed, an inverse gamma function is applied to undo the orginal gamma compression and reproduce the original luminance.  Then there is a more or less one to one relationship between the scene luminosity and the viewed luminosity. Additional tone mapping is usually necessary to compress the lumionosity to a range that can be displayed on screen or in a print. However, if we had a true high dynamic range display device, a scene referred image would look the same as the original. Indeed, most HDR images are in linear encoding.

     

    An example of a profile with a linear tone curve is linear_RIMM-RGB_v4.icc. See:

     

    http://www.color.org/scene-referred.xalter

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 28, 2012 1:49 PM   in reply to Bill_Janes

    I know that. All that is controlled for. The screenshot shows a luminosity

    on screen that is the same as what the camera recorded. It is ONLY white

    balance corrected. It is completely color managed. So this is as close to

    scene referred or neutral as you can get. All the intermediary steps with

    different encoding gammas are taken care of by the color management system

    and are basically transparent. The only thing that is missing from the dark

    image is the toning that Lightroom and ACR do by default behind the scenes,

    even when in the interface everything is set to linear. The point is that

    the image that people think is neutral in Lightroom is not neutral at all.

    It is actually toned down and compressed.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 28, 2012 2:01 PM   in reply to Bill_Janes

    http://www.color.org/scene-referred.xalter

    In these instructions, step 2 is wrong (and specifically the linked page here). This does not give you a scene referred image using current ACR/Lightroom implementations as there is another hidden tone curve that is inherent in the camera calibration that they are missing. This is what Jeff and I have been saying. Those instructions are probably quite old. The reference to CS3 sort of confirms that but even then we had multiple camera profiles if I remember correctly. Jeff probably knows far more about this than I do.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 28, 2012 5:13 PM   in reply to Jao vdL

    Jao vdL wrote:

     

    http://www.color.org/scene-referred.xalter

    In these instructions, step 2 is wrong (and specifically the linked page here). This does not give you a scene referred image using current ACR/Lightroom implementations as there is another hidden tone curve that is inherent in the camera calibration that they are missing. This is what Jeff and I have been saying. Those instructions are probably quite old. The reference to CS3 sort of confirms that but even then we had multiple camera profiles if I remember correctly. Jeff probably knows far more about this than I do.

     

    I was not aware that the behavior of the sliders had changed since CS3 and the corresponding ACR version. I'm not a frequent user of LR, but apparently it is using an adaptive algorithm. In your own quote of Eric Chan, he states that with ACR leaving the tone curve in the PE alone and setting the ACR sliders to zero, the default ACR tone curve rendering is turned off. I would presume the result would be a linear rendering encoded with whatever gamma one is using in ACR. Converting to LinearRIMM then convert to a gamma of one. If Jeff, Eric or anyone else has information, I would appreciate it. At times a linear file is needed for photometric measurements or post processing best accomplished with linear files.

     

    http://www.luminous-landscape.com/forum/index.php?topic=59688.msg48310 6#msg483106

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 28, 2012 7:05 PM   in reply to Bill_Janes

    Bill, in general I don't recommend using Lr to produce images intended to be used for photometric or scientific analyses.  This is because there are a lot of steps applied internally, such as the color profile, which cannot be disabled.  (Keep in mind that ACR/LR are really designed for photographic workflows ...)  Some of these steps can involve non-linearities, e.g., due to clipping, tone curves, color mapping/compression, etc.).  Yes, a very carefully constructed color profile can avoid some of these issues, but not all.

     

    Speaking more generally:  I do agree that a linear scene-referred image will appear of a normal intensity on a display (not overly dark) if processed through a standard color management system, with a suitable color profile (linear gamma).  One way to do this is simply to edit the working space RGB in Photoshop and customize the Gamma to be 1.0.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 28, 2012 9:26 PM   in reply to MadManChan2000

    MadManChan2000 wrote:

     

    Speaking more generally:  I do agree that a linear scene-referred image will appear of a normal intensity on a display (not overly dark) if processed through a standard color management system, with a suitable color profile (linear gamma).  One way to do this is simply to edit the working space RGB in Photoshop and customize the Gamma to be 1.0.

     

     

    But that's not what the thrust of this thread is about...it's about some users wanting an altered reality (zeroed) because they "think" it gives them a better "neutral" starting point–which is not something I agree with, but different strokes for different strokes. The bottom line is with LR4 and DNG Profile Editor, those people can have their cake and eat it too...

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 29, 2012 10:11 AM   in reply to Jeff Schewe

    Jeff Schewe wrote:

     

    MadManChan2000 wrote:

     

    Speaking more generally:  I do agree that a linear scene-referred image will appear of a normal intensity on a display (not overly dark) if processed through a standard color management system, with a suitable color profile (linear gamma).  One way to do this is simply to edit the working space RGB in Photoshop and customize the Gamma to be 1.0.

     

     

    But that's not what the thrust of this thread is about...it's about some users wanting an altered reality (zeroed) because they "think" it gives them a better "neutral" starting point–which is not something I agree with, but different strokes for different strokes. The bottom line is with LR4 and DNG Profile Editor, those people can have their cake and eat it too...

     

    That is not the main thrust of this thread, but it is the thrust of a subtopic that we have been discussing. The fact is that linear rendering of a raw file does not appear dark if a proper profile is attached and the file is viewed in a color managed application.

     

    Shown below are images of a Stouffer wedge with the same raw image rendered into ProPhotoRGB using ACR with the sliders set to zero (linear tone curve) and with the same ProPhotoRGB image converted into LinearRIMM (gamma 1.0). The image is a screen capture converted to sRGB for web display. The gamma 1 image is not darker, which does not seem to be what you have been asserting.

     

    LinearComposite.png

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 29, 2012 10:40 AM   in reply to Bill_Janes

    Bill_Janes wrote:

     

    The gamma 1 image is not darker, which does not seem to be what you have been asserting.

     

    I think the presumption there was that you were rendering out into an sRGB situation, in which case it would be dark.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 29, 2012 10:44 AM   in reply to Bill_Janes

    Bill,

     

    the darker boat image I show above IS displayed through a proper linear profile. The direct render from the raw using dcraw (which is the only way I know to absolutely certainly do a linear conversion) IS different then using the supposedly linear approach in ACR/Lightroom. My guess actually is what I said to Jeff above that there is a hidden exposure adjustment in the ACR/Lightroom pipeline that probably is camera dependent. For the boat image above, if you add one full stop of exposure, it looks fairly similar to the ACR "linear" image. The point is that there are hidden adjustements that are inherent to the calibration profile in the ACR engine.

     

    Here is the proof.

     

    Conversions.jpg

     

    Left top is ACR converted exactly to the color.org instructions using the Adobe Standard profile for this camera. Everything is zero and a linear tone curve. Right top is the really linear dcraw conversion taken into Photoshop and assigned a linear (gamma=1) profile. The color management makes sure the tone displayed is correct. Compare this with the right bottom where I did not assign a profile and where the image is just interpreted as if it is in sRGB. Much darker. The left bottom is the actually linear conversion from right top where I added a +1 EV exposure compensation in LR 4 beta which I used to make this comparison display. Safe some differences in hue, this is fairly similar to the ACR "linear" conversion. This is what I meant with that the white point was different in Jeff's pepper comparison.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 29, 2012 10:58 AM   in reply to MadManChan2000

    MadManChan2000 wrote:

     

    Bill, in general I don't recommend using Lr to produce images intended to be used for photometric or scientific analyses.  This is because there are a lot of steps applied internally, such as the color profile, which cannot be disabled.  (Keep in mind that ACR/LR are really designed for photographic workflows ...)  Some of these steps can involve non-linearities, e.g., due to clipping, tone curves, color mapping/compression, etc.).  Yes, a very carefully constructed color profile can avoid some of these issues, but not all.

     

    Eric, I tested the method recommended by the ICC for obtaining scene referred files (the reference was for PSCS3), using ACR ver 6.6 using the AdobeStandard profile with process 2010 by photographing a Stouffer wedge and rendering into ProPhotoRGB with a linear tone curve (sliders set to zero). An exposure adjustment of -0.5 EV was used to compensate for the BaselineExposure offset.

     

    The Imatest LogLog plot is reasonable linear, with some lifting of the shadows by flare.

     

    05_ACR_Lin_expMinus0_55_LinearRIMM_Step_2.png

    Since LogLog plote are linear even with a gamma curve, I plotted the values on a linear graph to check that the gamma was near 1.0 rather than the 1.8 of ProPhotoRGB.

     

    05_ACR_ProPhoto.png

    05_ACR_LinearRIMM.png

    The results are reasonably linear with some deviations in the highlights and shadows, but may be sufficient for government work.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 29, 2012 11:05 AM   in reply to Jao vdL

    Jao vdL wrote:

     

    Bill,

     

    the darker boat image I show above IS displayed through a proper linear profile. The direct render from the raw using dcraw (which is the only way I know to absolutely certainly do a linear conversion) IS different then using the supposedly linear approach in ACR/Lightroom. My guess actually is what I said to Jeff above that there is a hidden exposure adjustment in the ACR/Lightroom pipeline that probably is camera dependent. For the boat image above, if you add one full stop of exposure, it looks fairly similar to the ACR "linear" image. The point is that there are hidden adjustements that are inherent to the calibration profile in the ACR engine.

     

     

    Jao,

    I don't know how to explain the difference between our results, but one possibility is setting of the white point. ACR uses a baseline offset, whith is +0.5 EV for the Nikon D3 which I was using. If this offset is not used, the ACR renderings appear to light, and one may assume that there is overexposure when in fact none has ocurred.

     

    Perhaps Eric can comment, but I accept his explanation that a linear file can appear normal if a proper profile is used.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 29, 2012 11:29 AM   in reply to Bill_Janes

    This will work in some cases, Bill, but not all.  This is why I cannot generally recommend it.  Even highly skilled users would find it nearly impossible to keep track of the cases when it would work, and not work.  This is the nature of using a system to do sometime outside the scope of its original design.

     

    That said, if this approach satisfies your needs, then by all means continue to use it!    Just be aware that the required steps could change at any time.

     
    |
    Mark as:
  • Currently Being Moderated
    Feb 29, 2012 12:43 PM   in reply to Bill_Janes

    The difference is just the baseline offset, which appears to be +1EV for

    the D300 that I used in that image. So the white point setting (which is

    equivalent to exposure adjustment) is what is causing these linear images

    to be dark in my case so we are in agreement.

     
    |
    Mark as:
Actions

More Like This

  • Retrieving data ...

Bookmarked By (0)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points