Levels middle slider actually not a Gamma?

[shijan]

<p>Please take a look at this picture - left is the correction done with curves preset of negative gamma 3.0, right is the Levels middle slider more-less matched to this. Can anyone explain <strong>what's going on there</strong> and why so called Gamma adjustment with Levels tool produces absolutely different tonality result than Curves Gamma adjustment and never can be matched one to another? Corrections done in Photoshop as you can see.</p>

[shijan]

<p>ouch, here is the image:<br /><img src="http://i39.tinypic.com/25i3sia.jpg" alt="" /></p><div></div>

[Tim_Lookingbill]

<p>Where do you get the idea that word "Gamma" which isn't anywhere labeled in Level's interface describes how the middle slider functions? There's no correlation BTW.</p>

<p>You want to get a really dramatic effect with contrast changes convert your gamma encoded image into linear encoded and use the Exposure... tool under Image>Adjustment>Exposure... menu. Fascinating but completely useless if not a PITA to use.</p>

<p>What you're trying to accomplish from what I can see from your layers stack can be done in ACR/LR in linear space a lot faster than the time you've taken to make those layers and post this topic wondering about the differences between gamma with regard to curves and levels behavior.</p>

[digitaldog]

<p>We can (should) only use the term Gamma if the formula fits: </p>

 

 

 

<p>output = inputˆgamma </p>

 

 

 

<p>If we were to plot the tonal input value of a device in comparison to the tonal output value produced, and it followed the specific gamma formula, we’d have a true gamma curve and that curve would have a specific value (the result of the gamma calculation).</p>

 

 

 

<p> </p>

 

 

 

[shijan]

<p>Tim, the world Gamma regarding those slider mentioned in Adobe Photoshop Help <a href="http://helpx.adobe.com/photoshop/using/levels-adjustment.htm">http://helpx.adobe.com/photoshop/using/levels-adjustment.htm</a>l<br>

"The middle Input slider adjusts the gamma in the image. It moves the midtone (level 128) and changes the intensity values of the middle range of gray tones without dramatically altering the highlights and shadows"<br>

Thats why i'm so worry about this. They just could give notice that it is not equal to real gamma setting.<br>

btw, i try to emulate the tonality of levels middle slider with Curves and seems its near impossible, because the colors and tonality are tweaked in very very strange way there.</p>

[photomark]

<p>Dmitry, the middle slider in the levels dialog box <b>is</b> a gamma manipulation. As Andrew points out gamma has

a very simple, but very specific definition. The slider follows this formula precisely and it's pretty easy to test:</p>

 

<p>Make a grey gradient going from pure white to black. Apply the level adjustment and move the middle slider to a value.

If you picked 2.0 your new values will follow the formula: </p>

 

<p><code>output = (input/256)^(1/2.0) * 256. </code><br />

(you use the inverse of the gamma in the formula in this case but that's a technicality and you need to normalize the

numbers 0 through 255 to fractions by dividing them by 255 which will result in tones going from 0 to 1.0)</p>

 

<p>0 stays at 0, 255 stays 255, 128 becomes 181, 192 becomes 221, etc. </p>

 

<p>The curves do not follow the same formula and despite the name of the preset are not a true gamma correction. (BTW I

don't notice that preset in Photoshop CC, maybe they killed it). It's my (untested) understanding that curves are cubic

splines, which are an entirely different formula and will lead to different values. It would be an interesting exercise to try to demonstrate this. You could force the curves to produce an approximation of a gamma

correction, but you would need to use more than one control point. I'm guessing that preset is just a way of telling you how

it moves the middle value (i.e. 128 becomes 203 for a "gamma" of 3.0 but all the other values fall along the regular curve).</p>

[shijan]

<p>Very interesting...<br /> Those Curves presets created long time ago by Timo Autiokari using VisualBasic in Excel, as he wrote, and can be downloaded here from his archived website: <a href="http://web.archive.org/web/20080415220814/http://www.aim-dtp.net/aim/techniques/negative_film/index.htm">http://web.archive.org/web/20080415220814/http://www.aim-dtp.net/aim/techniques/negative_film/index.htm</a><br /> i'm not too powerful in theory but as i remember one time ago i disclose that my epson scanner on osx corrects images with gamma 1.8. so i attempt to correct it with levels to match the look of same image scanned on windows with gamma 2.2 and i got the same result as on my attached picture - the images never can be exactly matched. of course the difference was less dramatic because it was a small gamma shift.<br>

i think i'll need to test same levels middle slider behavior in other applications for sure.</p>

[Tim_Lookingbill]

<p>Does math know anything about making a pleasing image? No. So why bother with this level of precision? Math will always produce a crappy image. If it didn't then Auto button in ACR/LR would deliver perfect results.</p>

<p>The only thing functional about what I need to know about gamma is how it makes my monitor form a smooth gradient after calibration.</p>

<p>When it comes to editing an image in order to make it jump off the screen and come to life, considering my eyes' adaptive nature which constantly changes my perception of brightness, contrast, color cast and saturation editing on a bright, backlit display, achieving precisely defined gamma is the least of my worries.</p>

[digitaldog]

<blockquote>

<p>Math will always produce a crappy image.</p>

</blockquote>

<p>Yes and no <grin>. <em>Everything</em> done in Photoshop boils down to math. After all, your images and for that matter, everything you manipulate on a computer is just a big pile of 1's and zero's. But I agree on your point that it doesn't matter if the math is following a true gamma curve or not. Look at the image and make it look as you desire.</p>

[shijan]

<p>i accept both methods they are useful and can be used depending of situation and i'm not try to calculate math :)<br>

i just wonder why i can't get same result with curves. for example when i slide sliders in ACR i can see what happened with histogram and i can do same result with curves. with levels slider in photoshop i get something like brightness+gamma+saturation shift which happened in way that mind can't explain and represent clearly. but i like to understand how the tool works. </p>

[digitaldog]

<p>In Photoshop you are most certainly affecting saturation (hue) with curves as it was designed this way. If you don't want that, Fade>Luminosity after the curve move. LR and ACR don't operate this way, they protect hue with alteration of tone (again by design). And the data they operate on is vastly different (linear encoded vs. gamma encoded). </p>

[photomark]

<blockquote>Does math know anything about making a pleasing image? No. So why bother with this level of

precision? Math will always produce a crappy image. If it didn't then Auto button in ACR/LR would deliver perfect

results</blockquote>

 

<p>A remember a variant of this sentiment well from my time studying and later working at a music school. Many

students resisted the need for music theory. After all music is made of sound and time, not roman numerals. But

somehow the students who excelled at theory were able to find ways to apply it and were more successful

musicians for it. That's not to say you can't be a great musician without theory—there are plenty of examples—but

for most of us a solid foundation of theory only helps. </p>

 

<p>Whether we like it of not, math is the lingua franca of color theory. If we want to have common ground to talk

about color in a concrete way we need it. It may not directly make us better photographers, but it will give us a tool

to communicate with and learn from each other and that will make us better at our craft. This thread is a good

example. Without a mathematical description of what gamma actually is, it's really hard to talk about it. But once we

can express the underlying idea succinctly and accurately it becomes easy and we can avoid making demonstrably

false statements that are unhelpful.</p>

[Tim_Lookingbill]

<blockquote>

<p>—but for most of us a solid foundation of theory only helps.</p>

</blockquote>

<p>And how did you arrive at that consensus? With factual analytical data...uh, math?</p>

<p>Keep in mind the bigger the theory you're trying to prove the more precise you have to be according to theoretical physics. Proving how the human brain and "spirit" functions in how it understands and utilizes color and sound to produce something meaningful is a target bigger than the universe, so you're going to have to be quite a bit more precise in your theory there, Mark.</p>

<p>I thought the mention of the "Auto" button in ACR/LR that use mathematical algorithms to arrive at its results was enough context for most here to gleam what I meant concerning the dependency on math to explain everything. It doesn't explain why I like one image over another.</p>

<p>And I played music as well and was tied to understanding it and playing it strictly dependent on sheet music and the math it represented which pretty much crippled me from ever being a prolific improviser when it came to jazz improvisation as a trombone soloist. I was strictly a sectional player. But for some reason I could vocally scat and whistle jazz improv without seeing one note on a piece of sheet music. Didn't even know what key I was in. I just adjusted to what I heard. I'm quite good if I may say.</p>

<p>No math required. Hear the note, feel the note, see the note as my band director would instruct back in the day. Never mentioned how to do it with math.</p>

[photomark]

Tim, I'm not talking about automating our workflow or proving anything about the human spirit. I don't think

anyone is arguing that math will explain why you like one image over another—that's just a strawman

argument.

 

I'm simply arguing that we need math to talk about this stuff. Theory and math give us this tool and give

us a common framework to share ideas and understand abstract concepts. I really didn't realize that was

a controversial stance.

[Tim_Lookingbill]

<blockquote>

<p>I'm simply arguing that we need math to talk about this stuff.</p>

</blockquote>

<p>Why? How will it help make a better looking image? What numbers are relied upon and from what measured reference and who gets to choose it so it correlates to better photographic methods we can all use?</p>

<p>The automating is the proof you can't rely on math for discussing how to make better photographs. Why can't you understand the context of that statement in relation to this subject? </p>

<p>Math is not magically going to point you in the right direction toward that goal. Math is just used as backend infrastructure supporting the software to give us an interface so we humans can create better looking images without thinking about the math underneath. You paid for the software to do this for you. Let it do its job.</p>

[photomark]

<blockquote>>I'm simply arguing that we need math to talk about this stuff<br />

>>Why?</blockquote>

 

<p>For starters, to avoid making demonstrably false statements like this:</p>

 

<blockquote>Where do you get the idea that word "Gamma" which isn't anywhere labeled in Level's

interface describes how the middle slider functions? There's no correlation BTW.</blockquote>

 

<p>The minute you speak about correlation between one function and another, you are using math. If you don't want anything to do with it, fine, but why then are you contributing to threads like this one?</p>

 

<p>This is getting tedious. Open any book on color it's full of math. Look at any specification for colorspaces, printing processes, exposure, depth of field, dynamic range. It's all math. You don't have to understand it if you don't want to, you can let the software and camera interfaces hide all the numbers if you want, but you should probably refrain from talking about technical matters and stick to platitudes on the human spirit.</p>

[shijan]

<p>seems i find a problem) those gamma correction curves actually were done to be used with film negative invert curve, so they are some kind of flipped gamma settings. and they begin to produce the result similar to levels gamma slider when used negative invert curve. finally the result looks like the <a href="http://en.wikipedia.org/wiki/Gamma_correction">real gamma correction curve</a><br>

<img src="http://i41.tinypic.com/29zbkep.jpg" alt="" /></p>

[Tim_Lookingbill]

<blockquote>

<p>This is getting tedious. Open any book on color it's full of math. Look at any specification for colorspaces, printing processes, exposure, depth of field, dynamic range. It's all math.</p>

</blockquote>

<p>Yeah, it's getting tedious because you still haven't answered the question on how and why knowing all the math is going to help you make a better looking image and become a better photographer.</p>

<p>However, if it's just for fun and interesting FYI then I'ld understand that, but you haven't indicated that yet.</p>

<p>The impetus of this topic was to point out an issue with the visual behavior of the Levels middle slider that doesn't follow some mathematical definition of gamma that suggests this is a problem looking for a solution. How do you fix the Levels middle slider so it functions exactly like a gamma adjustment as claimed by an entry in an Adobe Help section but not labeled specifically anywhere in Levels interface.</p>

<p>Wouldn't you think that Adobe engineers are the folks to be contacted in order to mitigate this problem? How are we as end user photographers suppose to solve this on our own? It ain't gonna' get solved no matter how much math we throw at it.</p>

[photomark]

Tim, I did answer your question, you just didn't like the answer. At any rate, I'm done taking your bait. If

you want a discussion about why ignorance is better than knowledge, I suggest starting a new thread in

the philosophy forum.

 

Dmitry, glad you've figured it out. Sorry about the off-topic naval gazing. This is normally a more-focused

forum.

[shijan]

<p>what can i say, both of you are correct. when you do art better use intuition, because if you will calculate things with math the result will be perfectly calculated but same time will be clinically death without author's soul. but digital world got its own rules - when you need to understand complicated things for example how the color spaces works, or how works the tool in photoshop you better understand its theory or math or visual model. and after you understand it you begin feel it in same intuitive way as you feel other things.<br /> unfortunately today people more often forget the intuition and rely on math (or on new cool camera, cool vfx, helicopter or other hi tech equipment) more than on intuition, mind and feeling and so we get a tons of pretty looking but dump pictures, films, music and so on. sometimes people even can't imagine that it is possible to take pictures not for work, for sale, for client wish, for laugh or eye candy but for art, art not as form but as content.</p>

[robert_k1]

<blockquote>

<p>In Photoshop you are most certainly affecting saturation (hue) with curves as it was designed this way. If you don't want that, Fade>Luminosity after the curve move.</p>

</blockquote>

<p>For the purpose of only changing the tones, is there a difference between Fade>Luminosity and Blend>Luminosity after the curve move? Actually after any layer adjustment.</p>

[digitaldog]

<blockquote>

<p>For the purpose of only changing the tones, is there a difference between Fade>Luminosity and Blend>Luminosity after the curve move? Actually after any layer adjustment.</p>

</blockquote>

<p>Nope, not that I am aware of. You'd be better off applying the curve on a layer if you wanted that extra flexibility and with Fade, you have one opportunity to use it before the next command. </p>

[Tim_Lookingbill]

<p>Frankly, I'm surprise no one brought up fractal art as an argument in this thread which IS made with mathematical algorithms.</p>

<p>Ignorant is as ignorant does I guess.</p>

<p>http://en.wikipedia.org/wiki/Fractal_art</p>

[robert_k1]

<p>I prefer using Blend>Luminosity in adjustment layers. The Blend mode selection is right there, and takes only two mouse clicks. And I can see the difference in the preview.<br>

https://www.photoshopessentials.com/photo-editing/layer-blend-modes/luminosity/</p>