Grade School: Mastering Qualifiers

foreign oh hello there grade school how are you today I'm excited to be with you guys it's gonna be a good session today um I'm doing pretty good I was just telling my buddy Rafa who's here with me hey Rafa hello Colin hello everyone um Ralph and I were just talking before we started uh the session here about uh my week which has been a little bit more like you know like uh color science heavy than I would prefer I kind of like to evenly divide my time between uh color grading and color science it's been skewing pretty color science heavy this week and kind of the tedious not like looked outside of color science so uh put it this way I'm really glad to be with you guys and talking and discussing creative aspects of what we do uh and not looking at a spreadsheet I've been in in Google Sheets and doing formulas and stuff a lot this week kind of fun after a while it gets a little bit old so I'm glad to be with you guys I hope you guys are doing awesome and we're gonna have a fun session today because we're talking about qualifiers and about the way that they really work and what they really are as I said in this week's pre-recorded video one of the things that uh we haven't discussed as much here on the channel even though I've talked about or dismissed qualifiers a lot is what they actually are and how they work so we got to touch on a good portion of that in the pre-recorded video and we're gonna get to go much deeper today um one thing that I thought would be an interesting to start off with before we dive into everybody's questions would be uh something that started for me as a a question that came up in my mind I don't know maybe maybe a couple years ago now but I I can remember like there was this odd interval for me uh as a color so realizing like coming to the belief that I have now which is that like well qualifiers have a role to play but they're a very specialized instrument and they need to be used with a fair amount of knowledge and they need to be applied to the right tasks they're not a broad tool they're not a tool with lots of different application they're kind of only good at one specific thing I can remember really realizing and becoming convicted of that after I was deepening my understanding of color science and then within that kind of same period looking up and realizing like wait there are all these colorists Who I Really respect who I really admire whose work I really really love who use qualifiers a lot so what gives with that how does that work how does that what does that say about my conclusion and one of the things that I realized was well two things first of all I realized that many of those color who I admire and who would claim the use of qualifiers when I started to look a little bit more deeply I noticed that they are for the most part adhering to one of the things you guys have heard me talk about with qualifiers which is that if you got to do one do it on one channel and if it you can get it done with any of the three channels being Hue saturation or luminance the best one to use is going to be luminance because it's going to be the cleanest right and I did indeed find okay most of these colors I'm talking about are using luminance as their Baseline and they may or may not be adding other channel uh kind of limiter qualifications on top of that but with that in mind I just want to show you guys something that I kind of like startled upon during that time that I thought you might find interesting um oh look I'm already in resolve I've already been looking at this image so let's just do a little thing here I'm going to grab I don't know if you guys know this about me I often move it during grade school but I'm actually I'm kind of like a I'm I like to use both hands I've got my pen here and I've got my mouse here it's useful for certain things um but what I want to look at right now is I want to look at a grayscale ramp my favorite uh sort of assessment tool we're going to bypass our color Management on this clip so we're just looking at this pure ramp on the waveform I'm going to go in and kill my look like so and let's just do a really simple thing this is like a classic example of when I talk about like Elite colorists who are using qualifiers really really well I'm going to show you one of the most common things I see them doing it's something like this I'm going to go in and I'm going to work my high uh not my high softness but my high range on my hsl qualifier here like so I'm using my control surface but hopefully you can see the little high end of the luminance thing changing I'm going to go to my highlight mode so it's really easy for me to see you know kind of where am I having my effect so somewhere you can see even here in the waveform it's this like bottom third of the final signal that I'm really kind of manipulating I'm going to introduce a little bit of softness here like so and now let's turn this highlight mode off and let's kind of get some drama about things I'm going to turn off this view for a second and just go back to like the full-on grade so here's an example like all I've done is qualify my low end of the image like this right and something that I see colorists who I like do all the time is they'll kind of soften things out down there so I'm just working my contrast and my Pivot down in that region so basically what it means is that I'm really really being gentle on my Shadows even as I uh kind of like maintain contrast elsewhere in the image or you know really I'm just kind of like finessing my my Shadows like that looks pretty good to me before after I'm just kind of going deeper but also like softer it just kind of feels good down there right this is like a move that I see colorist I dig use all the time it's like a really good move in general so like steal this move if this is all you get out of grade school today this move is worth it in my opinion but here's what I find interesting so like how do I if I go back to my original question how do I level this with what I all that trash that I talk about qualifiers and with my sort of conviction at the time Newfound conviction that like you know what qualifiers are really not the best way to achieve most goals and they're kind of sharp and narrow and and not the best tool how do I level that well let's take a look at what this is effectively doing to my image on a curved basis there it is right there it's just a curve that was like my light bulb like it's just a curve and if I do the qualifier poorly say for example if I go really hard on my softness if I go to a zero softness you can actually see the break in the curve right and even though none of us would probably make that big of a mistake we might do something like that which isn't great you know or maybe we're making too big of an adjustment and doing something funky like that that's really obvious when we look at it on a ramp might not be obvious depending on the image that we are actually grading at the time so that was my big aha that I thought I would share with you guys is you can think of like any qualifier you could re-envision it as like okay what is that in terms of an X versus y curve in terms of like an hsl curve and resolve and you can oftentimes get a better result by reimagining an adjustment in terms of what its effective curve is because the curve unlike the qualifier shows you when you're getting sharp and narrow Jagged broken whereas in the qualifier you have to do this little crazy experiment that I just did right here in order to get that lens on things but I just thought I'd share that with you guys as like you know first of all sort of my answer to my own question of a couple years ago like well what does that mean if like all these colorists who I love are using qualifiers well what it means is that they are using the qualifier to draw in this case a loom versus Loom curve that just happens to because they're experienced with the tool they know how to end up creating a nice soft gentle effective curve because they've got lots of practice and because they're good at what they do but in my opinion they're really just using uh maybe an odd tool to get at a really good result so that's my own answer to my own question that I brought into grade school today and I promise that's the only question that I'll ask today the rest of the questions can come from you guys let's hear what we got out there we got any good ones Rafa yeah we have a question straight up about what you just explained basically and what is the difference between between this method and what you can already do in the HDR Wheels so again I think like the you know you guys know as I already copped to today my way of understanding almost any operation in color grading is to evaluate it in terms of a curve so that can get more complicated depending on what you were doing and the Curve will not tell you everything but if we bypass our color management again here we can do the same experiment we can go to our HDR wheels and you can evaluate at least on a 1D basis what am I doing what is the net curve of my operation this is a great question to get into the habit of asking yourself either when you're trying a new tool when you are applying an adjustment to a particular shot and you want to get a sense for like is that nice and clean or you have a suspicion that it might be a little bit rough on the image throw up a grayscale ramp like I'm doing here we should make this a free download for you guys shouldn't we we're going to make that happen because this is a really useful tool you can also generate or generate it within resolve which I can show you how to do at some other point but looking at what you're doing on a grayscale ramp is a great practice to get into but you can see like when you look at your different ranges and moving them up or down like that's quite similar and uh short answer to the question that you just asked could you one to one match exactly what I just did in the qualifier by adjusting your values within the HDR wheels and adjusting your like Zone pivot and your falloffs and everything yeah yeah you absolutely could and it goes the other way too you could do something here and then decide I want to do it as a qualifier so really the question for me becomes like something that I talk about in my book the color is Ten Commandments how do I choose the right tools and when I say write tools I don't just mean what's the tool that can get the job done the reality is if you like you know like randomly picked 10 of the tools in resolve out of a hat and said this is all you get Colin I or any other good colorist would find a way to achieve my creative agenda with that randomly selected set of tools so it's not just about what tools can get the job done it's what what tools can get the job done in the very best most efficient most intuitive way and that answer May differ from one colorist to the next but that's kind of how I think about those things and uh you know like why I get so interested in figuring out like all right what's the best tool and then for me as a color scientist I'll go even further and say does that tool exist in resolve and if it doesn't why doesn't it and can it exist and then I'll often go in and create my own tool but it really that that's the chapter of my book that I would point you toward if you haven't read it already it's worth checking out it's called choose the right tools it's really all about figuring out what is not just a functional but what is the optimal tool to apply to the task at hand all right we also have a couple of questions about how qualifiers behave in basically large spaces like DaVinci intermediate basically the range for example for luminance starts like around at 50 percent uh until that every value that goes under that doesn't really have any information so uh is the qualifier Corner space aware how does it work in those colorless space in those big core space yeah you know what I'm really glad this came up because to be honest with you guys this is something that I've explored I think we've even explored it here in grade school before but it's been a minute and I want to see if it's changed so let's just do a quick experiment here let's do like the most simple experiment I can think of which is just to go from white Gammon intermediate to gamma 2 4. we're going to do my favorite exercise here we're going to make a sandwich so we're going to go from wide gamut intermediate to gamma 2 4. I'm going to leave tone mapping and Gamma mapping off just so it's easy to get back into the other space so let's label this we're going to say 2 7 or 9. and then we're going to do another one I'm fired up for the new version of resolve where we can do an automatic swap of input and output but I'm not brave enough to update my systems to a beta just yet have you guys done that any horror stories or is all thing is everything going smoothly let me know so if we look at this sandwich that I've made here we're going DaVinci y gamut Da Vinci intermediate to rec 709 gamma 2 4 all this junk turned off with the exception of white point which doesn't even need to be on it's not doing anything and then we're going to do the exact opposite 79 gamma 2 4 DaVinci y gamma intermediate all this junk turned off so as we're fond of doing here in grade school and on the channel we've created a do nothing sandwich this round trip results in no change to the image at all but this is really useful for our purposes because we want to evaluate what is the functional difference between the qualifier its ranges its feel in log Space versus in gamma space so here would be like this is the classic domain of the qualifier this is really what the engineers had in mind when they designed the qualifier way back when was for operating in Gamma or display space so the question is how does it work comparatively in log space like we are in here in DaVinci y gamma so let's just do a simple example Maybe will do too but for one since the question was about luminance ranges I'm just going to go in turn on my highlight mode and sweep until I'm looking at a particular selection of dark values like so so maybe something like that and do just a little bit of softness now let's just try something here I'm going to tap node number two here hit command C go to shot number four or node number four rather and hit command V and we can see kind of the difference between those two things like so now I'm actually curious I want to double check that my color management is set up properly which it is so if we look at the difference between these two things we're actually we're getting a bit counterintuitively because we're in log and everything is sort of pushed up we're actually getting more Shadow by qualifying in log than we are by qualifying the exact same range in our gamma space but let's also do another example this is the one that like always comes up for me that I always question let's try saturation so if we go into again here in log space and let's just sweep our saturation let's sweep our high saturations down and you're going to see I can just go down down down down down I'm almost at a saturation high value of like one before anything starts to go away yeah I mean it take I have to get to like a 0.7 to qualify out a good chunk of my image like so so if we have that as the qualification here in our log space and let's just keep a completely hard Edge on this let's do the same experiment go over here and hit command V so those are somewhat different now the part that I'm not sure about that I'm kind of thinking through on the Fly is the color space awareness of this image so in other words if I am let's put it this way if I if I want to use any kind of color management at all let's just try something here if I change my timeline color space is the qualifier's behavior going to change I'm going to change this to rec 709 gamma 2 4 to reflect what's actually happening here so I'm going to hit save and it doesn't actually look like the matte itself is changing let's go to the Highlight the black and white mode here and try that again so we're here no that actually is changing so what does that mean what does that tell us I think what that tells us is put it this way if we look here this I'm going to expect is going to exactly match this is this is a a cheap trick but it will work for us I'm just going to grab a screenshot of this because I want to see what this mat looks like with that setting turned on versus my log mat in this domain like so so let's go back well let's see here sorry let's try like this I'm going to change my timeline color space back to 709 gamma 2 4 here watch that image change so this is 709 gamma 2 4. and I am qualifying like so so I'm just qualifying with the expectation that that's the domain of my image which is effectively what I'm doing here and if we grab a still of that well let's see yeah interestingly those are well I think we need to actually have that on you know what I'm gonna be honest with you guys I'm getting I'm like getting a little tailspun here short answer I don't know if unless we are explicitly changing the domain of our color space let's just simplify a little bit let's go back to our stock settings here and hit save and let's go back to our normal highlight mode so we can see that indeed when we change when we use a color space transform to change our color space we apparently can get a slightly different uh qualification range now it's not much like the biggest area that I see uh problem with the original question or or with the original uh you know like issue that was being raised is like it sure would be nice if I could separate out low medium and high saturations with more than one percent of my 100 range here so by moving into seven or nine gamma 2 4 instead of having you know things pivot around like a 0.7 they go up to around say a 1.2 or so so it's not like night and day but it does seem that if you wanted to go to this trouble within log of kind of like round tripping into another space that there might be a benefit to that or maybe you could just simply change your color space and Gamma here in your tag let's see if that works before I round out my very long Meandering answer to this question uh like so and that produces yet another result like yeah so that that one actually kind of puts me back into like if I copy and paste my adjustments here and change so right now my qualifier is identical and I change my color space here yeah that has no effect so that's not actually changing the qualifier so that won't do you any good so essentially all that you have to change the way the qualifier behaves is the actual state of the image that you are feeding into it like so so when I move into seven or nine I can pull a slightly different qualifier maybe have a slightly easier time building a qualifier and then I can move back into DaVinci y gamut or whatever my log space is that took a minute that was an interesting exploration I hope that gives some some uh Clarity or at least a line of further inquiry into the behavior of qualifiers in different color spaces let's talk about other kind of qualifiers for a second you only mentioned in your video the 87.5 what do you think about the other type like the RGB or the 3D qualifier yeah so those are definitely worth talking about I was hoping we get to cover them today so RGB and hsl qualifier let's wipe out some of this junk we don't need this where's my pretty look okay it's back and we're no longer bypassed okay good so let's look at our other qualifiers the short answer to the question can you sometimes get better results or different results with the RGB qualifier or with the uh 3D qualifier than with the hsl qualifier yeah yeah you absolutely can here would be my thing what I like about the hsl qualifier the number one thing is that I can go in with an explicit criteria and say I want this is a simple example I can say I want to select all of my colors that are low in saturation and low in luminance that would be like a sensible qualifier to pull and other than the problems that we were just talking about with like the saturation range being so funky that would be very easy for me to do I'm going to say well if I want to qualify those low things then I'm going to go into my qualifier and I'm going to sweep my luminance down so that I'm only leaving in the darker values let's turn the Highlight mode back on again using of course my my image as the guide for what that for what low means and then when I look at my saturation same thing I'm going to go in and sweep the high and say like I want to maybe get that red scarf uh out of the bargain if I can yeah something like that so like that is even with like the the odd sort of ranges that we sometimes see in the qualifier here that's quite intuitive to me because I have dedicated channels for the thing that the things that I'm conceiving of qualifying based on so I want to qualify based on luminance and saturation because those are intuitive to me and I've got channels for that so in that sense it's fairly intuitive now let's just for fun try the same experiment using the RGB qualifier if I go in here and just say like all right I want to pull you know whatever these low values but I want to leave the red out so I've got a baseline that's close-ish to where I was before now what do I do if I want to refine this unless I just want to keep dragging around the image I'm like well I guess I could reduce my red high end yeah but now that's also changing the luminance level of what I'm qualifying because luminance is coming from a combination of red green and blue right so that's a really good example to me of like if you press me right now there's not a single knob that I can know to adjust that is going to allow me to exclude the Red Scarf where back in the hsl qualifier there totally was I was like oh go to that high sat pull it in and you're good you know so it doesn't mean this can't work but it means you're left to kind of like just feeling around or you know doing stuff like if I went to this like minus button I could say oh like let's leave that out of the deal but that's as you can see like not necessarily netting the best results so it turns a even if it's not even if the ranges and stuff are kind of tricky it turns something predictable into something that I just kind of have to keep grabbing until like grabbing for possible solutions until I find one so that's the reality with the RGB for me doesn't mean you can't or shouldn't use it it's just a reason for me that I I'm like well that it might get me a better result but how long is it going to take me to get there and how am I going to know it's going to get me a better result until I spend all that time on pursuing it so it's kind of a an unknown and then if we look at the 3D qualifier here would be my suggestion on the 3D qualifier by the time you're doing a 3D qualification which is pretty complex in character what I would recommend at this point would be the use of your magic mask and we haven't talked much about the magic mask here on the channel because I feel like it's still been evolving I mean honestly that's the best version of the key I've been trying to pull that I've seen yet so you know maybe that's serviceable I don't know um but for me by the time I'm into a 3D qualifier and in this case I'm looking at yuv you could look at different versions of this this could be maybe I'll revise what I'm saying here actually I feel like this might be more interesting potentially than the RGB trying these different color models and qualifying within them and and like drawing shapes within them to see if you can get something interesting but at a certain point like the the tldr here would be that like if I can't like ideally I can use hsl because I know the exact those are criteria that I think in terms of very few of us think of like a low saturation shadow in terms of like how much red green and blue is in that you know what I mean whereas with hsl that's quite intuitive for me to go like well that's low end of the luminance low end of the saturation curve so if I have to go much beyond that and things are not intuitive to me or it's not I don't know ahead of time I'm going to be able to pull a really good key that to me makes uh the image a really good candidate for using the magic mask as opposed to any of the qualifiers if that makes sense a little insight that deem shared with us in the comments basically we haven't seen the qualifiers changing before because we you have selected the basically using color space aware tools but it will disable that everything should change between all the spaces really yeah that is interesting Jim get up here dude I'm gonna sit down I'm just gonna follow along man that's gold if that's the way it's actually working here so if we were to I mean we were yeah like so I I'm assuming the easiest way to test your idea here Jim would be like all right if we sweep our luminance range like so now that I've turned color space so where grading tools off which by the way was in my project settings color management color space aware grading tools now if I change my gamma like to linear I should get a different result because before I was seeing a change with my uh when I moved color spaces like let's just do a quick experiment here when I go from I'm just going to say use timeline even though I usually discourage you guys from doing that because I don't trust it for the sake of speed I am going to use it so if I go from uh DaVinci y gamma to 709 gamma 2 4 I've got here's here's the same qualifier before and after the image and let's turn this yeah so this is off for now so even with that turned off I am seeing different ranges being qualified here so how does that change if I turn it on doesn't actually seem to affect the qualifier yeah I don't I don't have an exact answer for like that there I'll put it to you guys this way since I think results 17 there's been a claim in the documentation for result that's like hey the qualifier is now color space aware but I've it's I've kind of been trying to like pin that down what that means and whether it's true for a while now and I'm still not sure it's actually a good reminder to me I should reach out to our friends at Blackmagic and just confirm what the intended and actual behavior of the qualifier is because that color space will wear a grading tools thing I'm I'm sure would apply to like the HDR zones but it doesn't seem at least uh with me thinking on the fly right now I can't seem to think of a way to make it affect the behavior of the qualifier but maybe I'm missing something and let me know if so Jim a little insight that I have to share is that at least in my testing this works very well in Fusion as well because I use the linear color space to basically qualify Skies all the time for the sky Replacements sure because you're going to have like lots more lots more color separation in there right yeah exactly yeah interesting interesting yeah it's so funny like you know within within resolve here like yeah that's a that is a good that's actually a really good strategy we could do something like that of course then you're going to have some harder edges to contend with but that can be at least you've got the region of interest that you need yeah exactly yeah okay cool well we're all we're all learning things and and finding more Mysteries as we go along what else we got we have a question here from greenstone um what kinds of food that you would know you should avoid pulling a qualifier and I want to add that question since I know you what kind of footage should you qualify oh we're getting right to the meat of the matter this is the probably the most important question that we are going to cover today so what footage should I or shouldn't I use a qualifier on I want to rephrase that question a little bit to me it's not a question of what footage should I use a qualifier on there is no footage that I'm like oh this is 8-bit as an easy example so I will not use a qualifier there's also no footage that I'm like oh this is pristine Arie raw 16 bit right off the sensor so I'm going to use qualifiers because I know I can it's not that simple it's more a matter of what am I trying to do and what's the simplest cleanest way to achieve it and if that's a qualifier does that work or do I you know like ultimately we may want to do things to our footage that our footage just won't let us do if you expose something four stops under and you want to go for like you know like a completely level open Airy exposure level there there is no tool that's going to do that it doesn't matter what like we talked about choosing the right tools before to you know think about that simple example there's no tool it doesn't matter how perfectly you select the some particular tool and how artfully you twist the knob for it it's not going to work that that goal isn't supported by the material so there is that reality but let's set that aside for a moment and just say like across the board we're all we can all recognize that we can all recognize like oh that's great that you want to open it up five stops it's not going to work so you need to calibrate recalibrate your expectations short of that what are our principles for when we use a qualifier and when we don't here's what I'm going to say this is again something right out of my book the color is Ten Commandments broad beats narrow that's one of the chapters of the color is ten commandments so what that means is we want to find the broadest possible tool for accomplishing our goal so if we for example want to get you know we kind of use this as a demonstration at the beginning of grade school today if we want to get thicker Shadows what is the absolute easiest way to get thicker shadows in resolve what is the cleanest broadest simplest way to get thicker shadows in resolve I probably won't wait for the answers to flip through on the chat because of the darn lag on YouTube I wish we were in a zoom call for moments like this but I want you to think about it for your own sake what's the simplest way for me to get thicker Shadows you have an answer in your mind it's simpler than you think whatever your answer is it's probably simpler than that offset that's the simplest way for me to get thicker Shadows Shadows thicker you cannot tell me I'm wrong my Shadows are indeed thicker now is that the right tool well that depends on my creative agenda because this is such a broad tool this is in fact the broadest possible tool that I can use in resolve so in addition to getting thicker shadows I'm getting a darker image aren't I I'm also getting uh weaker highlights so this might be too broad of a tool so the idea of broadbeats narrow isn't you should choose a tool that does things to parts of the image you don't want to affect but the premise is you should choose the broadest possible tool that does accomplish only the part of your goal that you're concerned with okay so what would the next one down from there be well that would be lift wouldn't it and that's probably in this case going to be pretty sufficient like so like let's just say we want to get those kind of like thicker Shadows but again it depends on what your goal is like if your goal is just to have some thicker Shadows then that's probably a decent start especially if you pair this with a little bit of gain so you're going down and left and up and gain so that you're still keeping those Peaks kind of at a similar level you can see it even here on my waveform I'm just getting thicker down in the bottom that's a good example of like all right that's still a super broad tool that could be a perfectly viable option for making your adjustment now if we look at like all right well what if we wanted to do this as a qualification well we could do it in a couple different ways we could use the qualifier or for me the next broadest tool would be to use either like HDR zones or log Wheels but those are again kind of tough to know what your effective range is so instead we could do something like our custom curves and you could do something like this we could say like I'm going to use my DaVinci wide gamut Gray this is just mid gray in my DaVinci wide gamut intermediate color space I'm going to option click my custom curve so I've got a control point there and I'm going to say my or actually let me do that in a separate node like so and I'm going to say now instead of a lift operation which by the way if you want to see lift on the curves this is lift deeper left right or left okay so now I'm saying well instead of doing my lift in that broad of a fashion I'm going to mess with the bottom end of the image but I'm only going to have effect below middle gray so that's a narrower tool right that is not as broad as lift which is not as broad as offset but you can hopefully see the kind of ladder of complexity That I'm Climbing up only as my creative needs demand so reason I go into all this is because the qualifier kind of sits at the top of that ladder like it's a very narrow tool now you can be broader or narrower even within the qualifier but fundamentally it is a narrower tool the only reason you would use it is to make narrower selections then you are able to get with something like gain or lift or Gamma or even your custom curves so when should you use a qualifier when there is no broader tool which serves your creative agenda so that's kind of the simple answer there and another great alternative because so far we've talked about a lot of like luminance adjustments within qualifiers but of course we qualify like you know the the biggest like misuse of qualifiers that I see out there among new colorists is qualifying skin tone well there's a number of problems with that but let's just say for a moment that uh we do indeed want to explicitly change the skin tone of our subject in some fashion let's say we want to saturate it a good alternative to qualifying my subject skin tone would be to do a broader version of that where I go to a hue versus SAT and I eyedropper this range and I drive that value upward like so that's very similar to what the qualifier will do but again as I I mentioned at the beginning of the episode today here I can see how broad or tight my curvature is here and if I do something dumb like that where it's like oh that's like impossibly tight and Jagged and you can see the detrimental effect that's having on the image here it's obvious I don't need to see this image to know that this sucks you know what I mean in the qualifier that's not so easy to determine always because it's just a bunch of numbers so this has the potential to be broader and softer it generally is and it also will tell you when you are getting too narrow in a way that the qualifier typically won't so that's how I would think about the qualifier is like is there any possible tool that is broader softer gentler that I could use to accomplish my goal than uh the qualifier and if not then I want to use the qualifier itself with those same principles want to work as broadly and softly and simply as possible we talked about this in the video this week within the qualifier one channel if you can luminance if you can get it done with luminance two channels if one won't do try to avoid three channels try to introduce lots of softness at the edges of your ranges so that you're taking out an insurance policy against you know like a qualifier you guys have probably heard me make this analogy before a qualifier is akin to like a surgeon's scalpel we would not want to have no surgery in modern medicine surgery is a very like beneficial Medical Practice in certain situations but if you can get it done with diet and exercise that's probably going to be better than going under the knife right same thing with our images all right let's take a race from qualifiers for a second I love using power windows for making my grades but sometimes the foreground elements get in the way of the background so any tips on how to work around this sometimes the foreground elements get in the way of the background so if I'm sort of envisioning the right thing in my mind I'm seeing a situation where maybe you're knocking down your background but then your foreground is occluding or coming into that power window range I hope I'm getting that right here's one thing you could try do you have any motion elements got a bunch of static shots in here this one might work so you know like it really just depends on the specific example but actually what you just described one solution depending on the specifics could be a qualifier so if you were to do something like you know what if I want to let's just wipe out my template or no it's fine I'll leave this up what if down here in my secondaries branch of my node graph what if I'm doing something like I mean I have no idea how far off the reservation of the kind of adjustments that you're talking about this is but let's say you know like in the background I'm really wanting to I don't know like warm up this highlight like so I'm really give it kind of like a nice yellowy character and it's nice and soft like that that's kind of pretty right maybe that's the thing now do I want what's happening on her it's actually not bad but maybe I don't maybe this is an example of the foreground kind of stepping into something like if she wasn't here and I just wanted to control this uh light pouring in from overhead this would be the shape and the position of my window in an ideal scenario but I've got a person who's stepping into the frame here so this is something you I often use qualifiers for and is right in line with that principle I just talked about of using one channel and letting that channel be luminance if possible I'm just going to go in turn my highlight mode on and sweep my luminance range from the bottom and say like yeah none of that stuff gets admitted into the party and then go nice and soft on the low end again maybe sweep back up a little bit more like this nice and soft again and now if I go off and on that same window is having a nice soft gentle effect on the area I want to influence and it's having no effect on her whatsoever if I go in here and I turn this qualifier off here's where it was before here's where it is with that turned on so like that's one very arbitrary example but that's something you can often use to your advantage if you've got foreground occluding background that's a great place to think about maybe I can use a qualifier do those things have a distinction in terms of the aluminum so are they well separated in terms of luminance or are they well separated in terms of hue that would be the backup option saturation being a distant third but if they're well separated in terms of their Hue or their luminance then you could look at using a qualifier to say hey if it's not inside the sweet spot for the Hue range of that background thing that I'm trying to affect with lots of softness built in then leave it out of the party and introduce that exclusion by using a qualifier so combining qualifiers and power windows is a really good way of netting out to a focused but nice and gentle uh sort of like selective grade of particular elements in the frame have we solved the mysteries of color grading one day we're going to come into grade school and there's going to be unanimous like you're done dude we don't need anything else we've we we all know we need to know we're ready to go I hope that's not today uh yeah I got you now all right perfect um let's go to a similar question basically uh how do you handle the scenarios where it's hard to get a qualify on the skin or in the background because the background and the skin are very similar yeah yeah yeah yeah okay so I love talking about this actually we're on a great shot to evaluate this on so like we're lit like this shot looks really really nice especially under this look that I've got happening here but we're definitely kind of like we're all in this upper left warm quadrant right now aren't we and there's not a ton of separation between her face and the background like she's more saturated but like kind of similar Hue Vector you can even see it here on the vectorscope so how do you deal with wanting to qualify one from the other well this is a good example to me of like asking better questions leading to better outcomes so really if you're saying you're trying to qualify your subject distinctly from a background for example well I'm going to assume that you are wanting to do that because you're feeling there's not enough separation like what are you going to do once you pull that qualification nothing no you're going to do something to try to move the background or move the subject I'm going to assume right otherwise what's the point of pulling your qualifier so if we think about it from that POV like all right what are you going to do what are you looking to achieve after you pull this qualification let's infer that you are looking to achieve greater color separation so the number one thing like first of all color separation that's something you should be thinking about like like just as a through line when you're color grading it's very very rare to be in a situation where you're like gosh I just can't get the color separation tamed in in these images did anybody have ever have that experience I haven't like you I'm not saying there's never a situation where you want less color separation but that's easy isn't it you trim yourself saturation back you move things into the cool quadrant you move things into the warm quadrant that's kind of easy what can be really tricky as this question implies is like yeah but what if I want more separation because separation is such a desirable trait in an image and not all images have a lot of innate separation what do we do then how do I get more separation into the image so that should be something that we're always thinking about and this is a great example to me of wanting to make sure that we are using the broadest softest tools that we can even if they're not getting us all the way there if it's getting us 50 60 70 percent of the way there that's now less work than a qualifier or something narrower has to do for us so in a shot like this I instead of trying to go in and qualify her as separate from the background which like let's just do that at face value for a second here if I try I'm gonna have trouble right like I can go in and I can fuss with these things and maybe like take that problematic saturation Channel no it's it's still gonna hold like I'm not gonna be able to get much better separation between the background and the foreground I could go in and throw a power window around her and limit things that way that might be a good solution depending on the way that the frame actually works in motion but the bigger thing that I can do before any of this qualifier would be to go back to my Basics go back to that balance note go back to that offset and do what I'm always talking about here on the channel that I'm always trying to do with my balance which is to max out color separation max out skin tone and max out color separation so if you just go in here like that may not feel like night and day but you can see even on my Vector scope I've got greater separation between her skin and the background than I did before and this also means if I want to go in and do other things if I want to do like let's use our our favorite trick this like HSV sat and just turn off channels one and three because I am not a fan of like standard saturation as you guys know if I want to go in and saturate things I can and I could even go back to my balance here let's turn this highlight mode off oh there we go now we're under our look and things are looking really tasty so once I've cooled things off here like look how much more separation I got off on off on and if I wanted to go in and add more saturation off on off on and let's do this let's just look at a before and after on this so we were here now we're there I'm going to go like this and let's go full full frame with this so before after before after this is the kind of thing I see this so much out there like it's again as I said with new colors so I don't mean to dog on those of us who are learning like we're trying to do good work but like when when we see images like this one where it's like oh there's so much richness and separation uh you know between background and foreground and my eye just feels really easily LED where it wants to go the first intuition if I were to tell you hey take this and make it look like that the first intuition for a lot of new colors so be the polar qualifier right off the bat right and if you grab it and you brute force it enough you can probably get it close enough on that individual frame but it's going to be dirty it's going to take a lot of work to get there when I did something totally different I used offset and I used overall saturation to create more separation between subject and foreground so when you're trying to I I don't mean to completely skirt your question but the reality is like I'm gonna go ahead and make this a rule for you don't qualify skin tone your image should be balanced based on what optimizes skin tone and then you can uh like play with elements around skin as needed so I wouldn't qualify skin tone in general and if your explicit goal is like well I'm trying to qualify skin tone so that I can get more separation between skin tone and background usually there's a broader way of going about that again going back to the broadbeat's narrow principle there's usually a broader way to go about achieving that goal without having to resort to the qualifier in the first place super fun question here if we move to hsl lav HSV or whatever space that is not RGB and we use the RGB qualifier thus that behave properly oh that's a fun question uh yes that will behave properly so if you now properly is of course a relative term so like the here's the reality like I've been dogging on like the stature if the question is like oh if I I'll Focus this a little bit let's say we go into hsl and let's say we go into RGB is this H is this s and is this L yes yes it is will it behave properly yes it will in the sense that these channels are now hsnl now watch what happens if we sweep if we try to qualify our saturation we are not necessarily going to have any easier of a time than we did uh before because saturations are all going to be living in a very finite range and in fact what we might need to do because I think we discovered earlier that these tags are not taken into account when we use the qualifier you might actually need to do an old school like sandwich on these things so let's just do that real quick and we're going to go to hsl I'm going to do this the really kind of fast lazy way and then on this side we're just going to come from hsl we're making sandwiches it'll be almost lunch time when we're done with grade school here in La so maybe I can go have a sandwich love a good sandwich so now we can see like all right we're we are now operating on our s Channel essentially yeah so let's let's open up our low range to zero and our high range like so interesting well okay that no that does make sense we're just here's the the challenge here is going to be that you're getting like a pretty distorted view in this highlight mode so you would want to look at it probably more like this but yeah like that will that will work the only catch is like you're still gonna have like the ranges are still going to be kind of funky not identical to what we were seeing what we'd see in an hsl qualifier but you're not necessarily going to have an easier time separating out different levels of saturation here than you would uh with the hsl qualifier because just to like give Blackmagic their credit an S Channel that's just a computed Channel when you're operating in log or in any color space for that model um an S channel is weak like just by Nature it doesn't reach it most of its values are like way down there so it's not just a question of Blackmagic like implementing a really bad qualifier it's just kind of a reality of the S channel in models like hsl HSV lab Etc the amplitude the strength of those channels is going to be pretty weak which is why it's so difficult to pluck out discrete levels of saturation from them but yes to answer the question uh totally we've moved into hsl here we're moving back out of it like so and let me just make sure I set that timeline color space back up correctly here which I did so yeah now in my RGB let's just prove this out with one more test here and say like all right well we want to sweep our high end of our red green and blue like so yeah that seems to be that seems to be behaving properly I can't swear a thousand percent that there's nothing not something that uh not not something else happening here with the way these red green and blue channels are being used but I don't think so so I think this is a perfectly sound uh sound IDF way to think about it do you use qualifiers when working with Brands who demand an exact color reproduction in their brand colors typically no if it's got to be that way it can but here's here here would be like something really good to keep in mind I used to for sure but here's something to keep in mind when it's you're talking about a situation where the client is going to be hyper critical of that product and of its color right so the last thing that you want to do is inadvertently introduce some chatter or noise or artifacting into that object and that's what the qualifier is great at you guys have heard me talk before about how qualifiers you you kind of have to babysit them for the rest of their life because like you can never be a thousand percent sure that some change other elsewhere in your node tree isn't going to bump it in fact that's why I pull qualifiers down here so it's the qualification itself is going to be not affected by any changes I make up here so I almost would be more averse to qualifying in that situation because it's such a you know like I can lose trust with a client instantly if they see that kind of chatter on the thing that's most important that they are paying me to make look amazing that's a situation where I would 100 percent use my hsl curves as my go-to I mean the first thing I would do is use power is uh not power windows use my balance and then in appropriate cases use power windows to say like hey here's my overall balance now let's throw a window on the burger or whatever it is and or on the you know uh on the the box for the product or whatever that may be track the window if I need to and finesse the balance with in that power window because again that's a nice broad simple soft approach but if I wanted to or if I needed to like you could also look at like this is the shot that comes to mind here and we're like Hey we're in the teapot business we make teapots so one thing I would also be mindful of at a system level like right now I'm on a look that I just know off the top of my head is doing stuff to this red before after I would make sure my look is doing things that support what we want to do to that teapot and if not I would change it but let's say that we're looking at this and they're like all right that there are our red teapot is actually more colorful than that it really really needs to pop you know rather than pulling a qualifier I would go to my Hue versus set sweep it nice and Broad and pump it up like so see what I mean that's still nice and localized at least in this image and that works really well and here's another great example of where we can pair uh hsl curves and or qualifiers with the power window in this case I don't need to but in other cases I might want to limit that adjustment to the interior of a power window and then track that window if you need to that way I'm not like a power window is not going to introduce artifacts as long as it's nice and soft an hsl curve is not going to introduce artifacts as not as long as it's nice and soft and if I combine them they're not gonna they're not going to be any more likely to introduce artifacts because they're combined does that make sense as opposed to a qualifier where it's like all right I've got a general good range here but I need to do something else in my ranges to keep this adjustment from affecting some other portion of the image now you are increasing your risk of introducing artifacts as opposed to combining clean operations or clean qualifications such as those from a power window and a hsl curve or a power window and a qualifier a clean qualifier that makes sense all right we got time for a couple more uh question here not about qualifiers but about overfix basically the question is that SoMo will fix work better using linear quality space but most ofx have a range from zero to one and sometimes in linear you have values over one so how do you deal with that problem yeah you know it is a problem uh that I was actually chatting with uh some some smart color science folks this week that like yeah linear domain as you guys may know like in log color space if you know like whatever like this image was captured in area log C3 uh and so it doesn't have any any code values that go higher than one because it all got packed into a zero to one range within log C3 if I linearize this image I'm going to get values in the case of Arie in case in the case of log C3 I'm going to get values that can go up as high as 55 and in other log curves it's even higher I can go to 100 or 200 or even you know like into those ranges so it's a good question like all right if we want to operate in a linear domain what's going to happen to those values between 1 and 55 if the tool is set to ignore or to clip or to clamp values above one the short answer is nothing good um so that's not that that's something that you kind of have to discover by trial and error and if you discover this tool expects zero to one and I but it does more of what I want in linear there's not much you can do except to try to like essentially roll and unroll the image which is not maybe the best thing but you could look at like tone mapping I'm trying to think of an easy image to easy tool to demo this on this is maybe a good one to wrap us up here um you know we could look at um let's see here we could look at our at our texture uh pop I think I can get that to misbehave if I flip it into linear by affecting my let's go to our Advanced operating mode here maybe not yeah I mean I I wish I had a handy uh example I could think of off the top of my head but I would say like if you want to if you absolutely want to use a tool in linear and it is clamping your values above one you could do this eventually gamut DaVinci intermediate DaVinci UI gamut Da Vinci intermediate luminance mapping custom Max of ten thousand custom out of 100 essentially and that is going to limit you down into a zero to one range within DaVinci wide gamut intermediate and then I would just make this kind of like your last stop before going forward through your system although yeah it gets a little it would be tricky to do with project level color management like we typically do on here on the channel so your best second option would be to try to invert it like so and these probably won't be identical although they're close depending on what you did in the tool this this round trip won't be lossless but you could try something like this this is going to compress you down into a range of zero to one and then you can do your thing here and then stretch back out to your full range of zero two in the case of DaVinci intermediate zero to a hundred um something worth trying anyway depends on what the tool is that might not even work depending on what it's doing in there but it would be worth giving a try with but it's a good good question and like I said I was talking about it this week with some color science friends of like well how should tools behave should all scene space color grading tools operate with no expectation of an upper boundary and some of us felt like yeah that would be ideal but uh it's not the reality that we live in today um we could do one more if we got another good one yeah we have a very good question here and basically the question is that using only the Hue uh section in the qualifier results into uh basically uh blocky qualification compared to what the RGB the hsl course do only using the Hue as a qualified why does that happen yeah yeah I mean that that's a really good question I haven't done like a a scientific analysis of One Versus the other I have a feeling you could get the exact same effective selection by setting your params effectively in uh your Hue you can even go to your uh six Vector these things were like really popular uh like when I first started color grading so let's say we want to do you know like just a qualification of my Reds this is our six vector and we can actually turn off these two guys completely so there's my Reds and maybe that luminous thing this is really designed more for seven or nine uh for like a display type of color space but something like this is going to be close to what you would get with uh you know like an hsl qualifier like if we do this and then we saturate those areas you know like it depends on what you're trying to do but saturation even if we wanted to do like Hue like let's do kind of a a crazy Hue rotation you know that's pretty clean that's not bad it's a broad uh and soft enough selection you can kind of see it here on you can kind of see some fringing here on the color chart a little bit but honestly that's pretty clean that may actually be just the nature of the you know if I if I went here and then maybe widened this out even further uh like so now I that's just because that center part is I suppose uh more of a let's see here yeah that's that's helping with that so you know like you you could get a nice broad gentle qualification within the qualifier but like is it going to be one to one as clean as your uh Hugh versus Huey Hugh versus SAT I actually don't know but I would also say like you know there's I don't see any advantage to doing it here if the Hue versus Hue or the Hue versus Assad is giving you what you want I would just go with that and then you know like you can actually this is another fun thing you can do if you wanted to do a let's say we want to do a hue versus Hue okay we want to take our Reds and we want to bring a little more kind of yellowy like so something like that but we only want to do that with like very saturated Reds you could actually like if it gets you the result that you want you could actually combine this with a saturation qualifier like so and say only do that on the more saturated stuff and you can see here because I've got a hard Edge on that qualification there's a some really bad artifacting but something like that is feeling interesting and if I flip this off and then on you can see the subtler effect that I'm having on the image by doing this or actually I guess that would be the better point of comparison there so anyway you could even combine qualifier parameters and hsl curves if you wanted to do that if you felt like there was something you like about your hsl curves but now you need to go in and delimit your saturation or your luminance you could actually do that so that's another another fun option to explore a few are feeling that way um all right guys that was a really good session on qualifiers uh we got to I feel like I was uh optimistic about getting to the bottom of that whole business with qualifying in different color spaces we definitely leave with more answers than we started with here today but I also don't feel like we have all the answers so we'll have to keep exploring and this is a good reminder for me to reach out to Blackmagic and ask some of these questions uh directly if I can get some answers from the engineering team and if I do I'll be sure to share them with you guys because it's good stuff to be armed with in our color grading Adventures um you guys are awesome thanks for being here thanks to my buddy Rafa for co-hosting with me always love seeing you guys on Friday mornings and exploring uh the craft and the science of color grading so thank you all as always I hope you have an awesome Friday I hope you have an awesome weekend and I will see you next week here on the channel take care

Channel: Cullen Kelly

Views: 3,219

Keywords:

Id: kdSDhBJf1iU

Length: 65min 45sec (3945 seconds)

Published: Fri May 05 2023