How to Create Image Specific Tonal Curves
What I'm going to teach you how to do today, in theory, is how to create image specific tonal reproduction curves in which you can be image structure specific, so you gonna have total reproduction curves in total reproduction curves. Rgb is not a color, it is a formula to mix color gray is a color. It consists of equal values of red, green and blue. And on this printer, this printable print black light, black light, light, black, magenta and sigh or light magenta light sayin and I think yellow, other colors that it uses so it's using color inks in conjunction with the gray inks, to produce one of these images, that you're using colors to produce the images that it's equal values of grey that create the image. Do you see where I'm going with this? That the conception of what I hear frequently from people that teach photography, I've made it a point to watch every movie I can get read every book that I can read is, well, color doesn't matter it's contrast, it's all about lightness and da...
rkness, it's contrast or q, that doesn't matter, it's just saturation or my personal favorite is well, if you're trying to find a photograph to photograph in black and white, make sure that when you're out in the world, it has a lot of contrast secret to photographing incredible black and white is to photograph an incredible color image. And from that, you can then create the black and white image of your dreams were in the first time where if we can imagine it, we can create it. That's. The power of digital photography so, let's. The reason why you want to do it last has to do with this image right here. Would you agree that that's, out of focus? 00:02:00.42 --> 00:02:01. How about now? 00:02:05.66 --> 00:02:08. Tighten that up, right? I've sharpened this seven 00:02:08.53 --> 00:02:09. times. 00:02:10.82 --> 00:02:12. The reason why I can't get away with sharpening it 00:02:12.55 --> 00:02:15. seven times that I put all seven sharpening layers 00:02:15.41 --> 00:02:17. on different layers, and I use opacity to blend it, 00:02:18.13 --> 00:02:21. because all artifact ing is cumulative and maybe multiplication. 00:02:22.62 --> 00:02:25. The goal is to negate multiplication artifact ing 00:02:26.11 --> 00:02:29. and to minimize cumulation artifact ing. What is cumulation 00:02:29.74 --> 00:02:32. artifact ing look like? If I applied all of the seven 00:02:32.74 --> 00:02:35. sharp innings that I did to an image, that's, what 00:02:35.75 --> 00:02:36. it would look like. 00:02:38.42 --> 00:02:41. That's. Multiplication, artifact. The reason why you 00:02:41.77 --> 00:02:44. want to leave your black and white conversions to 00:02:44.06 --> 00:02:47. the end has to do with handling a raw file, which 00:02:47.39 --> 00:02:51. should be handled with a little touch on it as possible. 00:02:53.02 --> 00:02:55. How many people bake pie crust? 00:02:57.32 --> 00:03:00. Okay, well, you handle pie crust is little is possible. 00:03:00.23 --> 00:03:01. A raw files pie crust. 00:03:02.82 --> 00:03:05. Leave the heroic things where you have layers and 00:03:05.55 --> 00:03:08. you have opacity, and you have layer masks so that 00:03:08.29 --> 00:03:11. you can negate multiplication artifact ing and only 00:03:11.32 --> 00:03:13. have to deal with cumulation artifact. 00:03:17.46 --> 00:03:17. Okay. 00:03:25.22 --> 00:03:29. Let's, take a look at the lovely shao lin cates. 00:03:33.22 --> 00:03:36. This is the red channel, this is the green channel, 00:03:37.16 --> 00:03:38. this is the blue channel 00:03:39.72 --> 00:03:43. are these all by our definition, a chromatic grayscale 00:03:43.07 --> 00:03:45. or grayscale image. They have a black, they have a 00:03:45.03 --> 00:03:47. white and they have a grayscale ramp in between. 00:03:48.22 --> 00:03:52. They're all the same image structure, but this is 00:03:52.18 --> 00:03:53. different than that 00:03:54.38 --> 00:03:56. and this is different than that. 00:03:57.67 --> 00:04:00. But they all three meet our definition. 00:04:01.64 --> 00:04:04. What you're looking at is predictability. This is 00:04:04.14 --> 00:04:05. the red channel. This is the green channel. This is 00:04:05.87 --> 00:04:07. the blue channel. When you put them together, you 00:04:07.27 --> 00:04:07. get color. 00:04:09.39 --> 00:04:13. This is what film recorded. It recorded a collision 00:04:13.04 --> 00:04:16. between these three frequencies of light red, green 00:04:16.51 --> 00:04:18. and blue. And from that, they produced whatever the 00:04:18.28 --> 00:04:20. black and white was that you you bought. You picked 00:04:20.88 --> 00:04:24. portrait's for age h p five plus landscapes or pan 00:04:24.64 --> 00:04:28. atomic x or tri x three twenty. Based on the way in 00:04:28.21 --> 00:04:31. which it recorded things so let's, take a look at 00:04:31.52 --> 00:04:32. simply 00:04:33.8 --> 00:04:34. removing. 00:04:37.7 --> 00:04:38. All of the color. 00:04:39.64 --> 00:04:40. Does that meet our definition? 00:04:42.11 --> 00:04:44. Okay, do you see where the color got in the way of 00:04:44.19 --> 00:04:44. the picture? 00:04:46.84 --> 00:04:49. So now this meets my definition but let's look at 00:04:49.47 --> 00:04:52. the channels that's the red channel that's a green 00:04:52.74 --> 00:04:55. channel that's a blue channel what do you observe 00:04:57.19 --> 00:05:00. nothing happened they're all the same so the problem 00:05:00.97 --> 00:05:03. is that I've lost the relationship between red green 00:05:03.74 --> 00:05:08. and blue what I have left is loom in its in rgb 00:05:09.74 --> 00:05:10. so 00:05:11.54 --> 00:05:12. I don't want to do that 00:05:14.04 --> 00:05:14. right 00:05:16.04 --> 00:05:18. so let me ask you this you've got clouds clouds are 00:05:18.73 --> 00:05:19. basically what 00:05:20.37 --> 00:05:23. great and they pick up the reflection of the ground 00:05:23.95 --> 00:05:26. beneath them and the ground beneath them is brown 00:05:26.41 --> 00:05:28. and so you have brown and gray clouds do you do a 00:05:28.75 --> 00:05:32. multi channel mixer approach or multi plugging approach 00:05:32.34 --> 00:05:35. or you simply do de saturation de saturate the color 00:05:35.53 --> 00:05:38. out of the clouds and just brush in the clouds yes 00:05:39.98 --> 00:05:41. simplest way is always the easiest 00:05:42.6 --> 00:05:46. oh did I not introduce this class welcome to it makes 00:05:46.47 --> 00:05:50. absolutely no sense would you like me to show you 00:05:50.08 --> 00:05:50. that 00:05:51.83 --> 00:05:54. okay it's ground hog day what every like for example 00:05:54.95 --> 00:05:56. the first thing you're talking photography get it 00:05:56.58 --> 00:06:00. right in the camera right yes so that says that you 00:06:00.01 --> 00:06:02. should set your camera to monochrome because you're getting it right in the camera right no the reason why you don't want to do that is because cameras take the green channel which is where illuminates is in midtown and replicate to the red and the blue so what happens is you have all the headroom of whatever your camera is, but with none of the benefit of the collision between red, green and blue. So, yes, you want to get it is right is possible. But no, you don't want to do that. It will all make sense to at least me when I'm done so. Okay, so let's, take a look at let's. Create a serious of files here so that we can look at them. Duplicate this call. This l a b one. Pitch these two things. It's. The one joy of working no. With projectors is that I have no screen acreage left. And let's. Duplicate this again. Call this lab, too. And then let's duplicate this. Again, we'll call this gs let's start with convert to gray scale. So the first thing I'm going to do, and I want you to. Watch this number down here. How much is that file number? What's that number thirty? Okay. First thing I'm going to do is I'm gonna go to image mode, gray scale. What's that number now, what did I just do? I just dumped two thirds of my data. That means that those two thirds of data that you could have access that were in there, held in the red and the blue, and those two channels are gone to you. You've lost the relationship between red, green and blue, and, as you can see, they're different. The red was lighter and creamier. The green had more mid tones in the blue tended to be darker and deeper. Okay, so let's, zoom back. So this is our grayscale image will put this over here. Call this decent. We'll call this gs. Let's, go to lab let's, convert. Get back here, let's, convert this one to lab, so I'm gonna go image mode. L, a, b and what you should see is absolutely nothing happened. So let's, take a look at our channels, that is. The l channel. Does that meet our definition? Yes, and there is usable information in there. Yes, this is the h m. Is there anything that you visually configure out how to use for your chromatic grayscale conversion in that channel? Really? No, you're shaking your head. I'm like do whoa! This is the beast channel again. So what happens here is that, yes, it may contain all the components that will produce a full, visible spectrum color image, but the access to the data eludes you. Now what happens in the first conversion process? This is called the classic black and white conversion technique, because sense, it's log, rhythmic, it replicates the way in which film would have recorded it. Okay, so what I'm gonna do here is, I'm going to throw away my two channels. What number? Do I have there? So what did I do? Dumped two thirds of my data. Okay, though this technique then requires that you go to grey scale. Okay, then you go to rgb. What did I get back? I have all of the head room of that file. But none of the collision between red, green and blue. I do not have access to the channels. So all I did was deflate the tire, take all the important stuff out, and made the tire bigger. But with nothing in it. Worked, didn't it? I guess you had to be here for that. All right, so let's, take this and call this no one. And let's drag l one over here to work comparison files close out of that. We don't need it now. After there was some hullabaloo about that, just basically being you just doing the luminous channel, the next one was convert to lab, okay, nothing happened, convert to gray scale, okay, discard the information, and if we look at this again, what number do I have back here? Okay, so once again, I wiped out and then convert to rgb to call. This fell, too. All right, so let's, take a look at these. Get out of this. This is l one and that cell, too. Lab conversion one laugh converge into do you see a difference? Okay, this is convert to gray scale and lab one. Do you see a difference? Okay, this is convert to gray scale in lab to do you see a difference. So the second newer version of the classic black and white conversion technique, the neo classic black on white conversion technique basically does with photoshopped does in convert to grayscale accept ads, a couple of extra steps. So if you have nothing better to do with your time than to just run through photo shop, this is the technique for you. This is di saturation of the image. The saturation of the image actually renders us a closer, more usable picture than either one of the lab conversions. The reason why you don't want to consider using the lab color spaces because you can't predict it. You do not have access to the red, green and blue channels of the file.