It is also worth noting, that Cullen has previously indicated a preference for turning off the automatic S-curve option in Resolve colour management so that he has full control over toe and highlights without resolve generating roll off for him.
Thank you. Once again, you've taken one aspect of color grading and explained it in great detail... allowing us to better understand and utilize the tools we have to work with in Resolve. And thanks for the additional tools from CKC. A+++
I appreciated some of the use cases in the explanation. I'm starting to appreciate, how many times concepts are repeated throughout the pipeline to deal with unexpected, expected and implied things.
Clearest, fastest explanation I've seen in it's own video. This should be THE video on highlight rolloff. As it turns out, I already understood all of this, but it's good to know my understanding has been correct. You've got my subscription. I would love to hear your thoughts on the famed Alexa highlight rolloff compared to other high end cameras -- Do you think the rolloff is something that can be largely replicated in the grade, assuming the camera captured enough dynamic range? Or does the Alexa truly have an extra quality of highlight rolloff that simply isn't achievable with other sensors? So many professionals speak to the superior rolloff of the Alexa, and I genuinely can't tell if they're absolutely right, or if they don't understand that highlights can be rolled off the same way in any image that holds enough highlight /information/. I would love to believe that everyone in an industry knows exactly what they're talking about, but this is rarely true, so I'm forever scratching my head.
Even though I'm quite familiar with the concept, I've enjoyed your video as it's always nice to see same concepts explained to different audience - I work as a compositing artist and our approach in naming and using such things is more technical. So that's nice to hear you a pro colorist naming things for less technical inclined people - in more organic human friendly way.
When you add an anchor to the curves, how do you get it locked to the center line? Mine always pops and moves it slightly. Guessing there is some key command I am just missing.
I love your channel! I would love more information on grading for phones and RUclips. I suspect many of your viewers are like me and delivering content on RUclips that is mostly watched on phones. Most of my clients actually watch on great HDR phone screens but I can never seem to export a video for HDR that looks correct. I’m always getting massive strong gamma shift. Most of us have cameras (10 bit or even RAW is very accessible) that could take advantage of these screens but don’t use the capabilities. You should consider a Patreon. I would also pay for courses.
There's a lot of info and contrast in the door window when you pull it down real good at 9:02. Do you have suggestions on strategies to evening out the range as much as possible? I often find myself in this position and will have to use power windows and qualifying, but it's hard to make it look good. I really have to flatten the overall contrast of the entire image before being to grade the entire frame as a whole. Thanks for awesome content as always!
I find myself in situations like these, and most of the times when trying to show all available DR, I end up with a unnatural looking image! I use highlights /shadows a lot and compensate with curves. But sometimes I sacrifice a little DR for the sake of natural looking images.
If you are starting with a Rec709 LUT, they usually apply quite a lot of contrast around a pivot of 45 IRE. So you have to fight that baked in contrast first to get the DR back into the image. When you are starting with a RAW format, you have the option to do a tone mapping first without adding a default contrast and after that step add contrast to the perfect pivot point for the scene. Usually I find adding contrast around the pivot of the middle IRE value of skin represented in the scene gives the best results. This contrast has to be applied after the tone mapping stage to use the full dynamic you mapped into Rec709 before.
@@Andresvideo Yeah, but sometimes it doesn't look natural, just bad :D It's more about avoiding obvious clipping, showing stuff closer to the way our eyes would see the scene.
@@oleberek REDcode or slog3 debayered to DWG down to rec709 after, so your feedback is only valid on the assuption that I grade display referred, which I don't :)
Cullen. I'm using your DCTL tool, set exactly the same as in this video, and no matter what I cannot get my screen OR waveform scope to look like yours, to replicate. Why might that be?
What I get confused by, is people saying particular digital cameras (eg. Alexas and Blackmagics) have "good highlight rolloff". I think what I would want and expect from a digital camera is an accurate representation of the light hitting the sensor. Then rolloff can be "creatively" applied in post as per the video. I wouldn't want, or expect, a "professional" digital camera to be doing any kind of rolloff. Am I confused or are they?
To avoid clipping, what you need is dynamic range. I'm sceptical that it makes sense to talk about digital cameras having an additional, separate property called "roll off".
I think there might need to be a distinction between what type of footage is being shot by these cameras. You're right that you'd want a high end digital camera to accurately record the light hitting the sensor, and if we're shooting RAW, that's exactly what we're getting. RAW, linear gamma files, just as light physically works in the real world. However, if we're shooting LOG footage, we are left with a digital file that no longer is representative of how light works in the physical world, we're left with a large gamma curve on our footage that is meant to maximize the recording of dynamic range of the scene. LOG gamma curves can be very different from manufacturer to manufacturer and I'm sure some some film makers and colorists have preferences to which LOG footage they prefer. So it's possible that "good highlight rolloff" is referring to the way the camera records it's LOG footage.
@@JimRobinson-colors Exactly. At the point of capture, it's about dynamic range, not "roll-off". It seems to me that roll-off isn't (or shouldn't be) something that happens as part of digital image capture. Unlike film, where roll-off is an inherent physical property of the film stock. Or maybe I'm confused somehow.
Unfortunately, I stop learning Davinci cause my cheap laptop was broken😢, if anyone in this comments section has a second laptop which you don't use it anymore and its still working, Im gonna appreciate if you donate it to me. Just let your responds here, and we can communicate further, and my base is here in Indonesia. Honestly, I want to learn seriously Davinci from this channel, and I cant buy the new laptop cause my budget is not enough. Thanks all
I think your approach is obviously correct, but the way in which you are explaining it is not as clear as when I saw Walter Volpatto explain it. Walter just put a stepped gradient up that represented even stops and identified the color space and middle grey. Like I have seen you do - then while all the stops are on the curve he would count 3 stops below and tag it and then count 3 stops above and tag it on the curve. Then lower the highlights on the curve and raise the shadows - it shows the compression of the stops forming the roll off on the highlights and a toe on the shadows. Without using the CST, the rolloff contrast worked on all the footage. I'm not comparing your teaching methods, but I think that your demo was over complicated for no real reason. The steps in linear then adding a CST with luminance mapping, is over complicating to me. It is possible that others might not see it. But the process that I explained above - you can easily see the middle stay consistent and see the luminance compression visually.
Join him live first thing Friday morning California time. Usually somewhere around 6am (varies, RUclips notifications however are fickle), then ask your questions. In Friday morning Grade School live Cullen discusses your questions, and adds extra insight to the previous weeks topics.
You are the greatest source of info on color grading on all of RUclips and you do it for free. Thank you for your service. You are greatly appreciated
It is also worth noting, that Cullen has previously indicated a preference for turning off the automatic S-curve option in Resolve colour management so that he has full control over toe and highlights without resolve generating roll off for him.
What exactly is this setting? In General Settings I found "Use S-Curve For Contrast" option. Is this it?
@@77dris Yes, that's it.
Thank you. Once again, you've taken one aspect of color grading and explained it in great detail... allowing us to better understand and utilize the tools we have to work with in Resolve. And thanks for the additional tools from CKC. A+++
Absolute GOLD! Second of your videos I've watched today and learned more in an hour than I have piddling around in Resolve for a year. Subscribed!
This is the first video I have ever seen that explains what exactly roll off is. Thanks a lot for making amazing videos
I appreciated some of the use cases in the explanation. I'm starting to appreciate, how many times concepts are repeated throughout the pipeline to deal with unexpected, expected and implied things.
I have been looking for this specific answer and demonstration for years Cullen! Thank you so much!
Awesome information here. Especially the explanation of tone mapping. Great work Cullen!
Thanks Cullen. I have the perfect spot to employ highlight rolloff. Your drill-down posts are fantastic.
Thank you for giving me the language to talk about an issue I faced a week ago as a beginner 😍
Man you're amazing. Quite good humor there in all your videos.
The best channel on color grading. Thank you thank you thank you
Its just perfect content for simple but complicated subject. Thank you!
Clearest, fastest explanation I've seen in it's own video. This should be THE video on highlight rolloff. As it turns out, I already understood all of this, but it's good to know my understanding has been correct. You've got my subscription. I would love to hear your thoughts on the famed Alexa highlight rolloff compared to other high end cameras -- Do you think the rolloff is something that can be largely replicated in the grade, assuming the camera captured enough dynamic range? Or does the Alexa truly have an extra quality of highlight rolloff that simply isn't achievable with other sensors? So many professionals speak to the superior rolloff of the Alexa, and I genuinely can't tell if they're absolutely right, or if they don't understand that highlights can be rolled off the same way in any image that holds enough highlight /information/. I would love to believe that everyone in an industry knows exactly what they're talking about, but this is rarely true, so I'm forever scratching my head.
thank you for this video! now I have a better understanding on this subject
Just 3 words. YOU. ARE. AWESOME
If you make a course. Trust me I am selling all I got to be able to afford it. You are amazing Sir. Thank you.
wonderfull explaination .
Really interesting stuff. Something I never really understood. I appreciate your time in explaining this!
Thank you! I really like your teaching style
Brilliant! Thank you! Happy Easter.
One more topic demystified. Thank you again Cullen.
Thank u so much sir for putting so much effort🙏
Really valuable info! Super thorough 👏
Even though I'm quite familiar with the concept, I've enjoyed your video as it's always nice to see same concepts explained to different audience - I work as a compositing artist and our approach in naming and using such things is more technical. So that's nice to hear you a pro colorist naming things for less technical inclined people - in more organic human friendly way.
Great video, as always. Thank you!
Could you say rolloff is the same as shoulder and knee?
When you add an anchor to the curves, how do you get it locked to the center line? Mine always pops and moves it slightly. Guessing there is some key command I am just missing.
Hold Alt (Windows) or Options (Mac) key while moving the control points
@@invariant7905 thanks!
very nice video, thank you!
I love your channel! I would love more information on grading for phones and RUclips. I suspect many of your viewers are like me and delivering content on RUclips that is mostly watched on phones. Most of my clients actually watch on great HDR phone screens but I can never seem to export a video for HDR that looks correct. I’m always getting massive strong gamma shift. Most of us have cameras (10 bit or even RAW is very accessible) that could take advantage of these screens but don’t use the capabilities. You should consider a Patreon. I would also pay for courses.
Gracias!
There's a lot of info and contrast in the door window when you pull it down real good at 9:02. Do you have suggestions on strategies to evening out the range as much as possible? I often find myself in this position and will have to use power windows and qualifying, but it's hard to make it look good. I really have to flatten the overall contrast of the entire image before being to grade the entire frame as a whole. Thanks for awesome content as always!
I find myself in situations like these, and most of the times when trying to show all available DR, I end up with a unnatural looking image! I use highlights /shadows a lot and compensate with curves. But sometimes I sacrifice a little DR for the sake of natural looking images.
If you are starting with a Rec709 LUT, they usually apply quite a lot of contrast around a pivot of 45 IRE. So you have to fight that baked in contrast first to get the DR back into the image. When you are starting with a RAW format, you have the option to do a tone mapping first without adding a default contrast and after that step add contrast to the perfect pivot point for the scene. Usually I find adding contrast around the pivot of the middle IRE value of skin represented in the scene gives the best results. This contrast has to be applied after the tone mapping stage to use the full dynamic you mapped into Rec709 before.
@@Andresvideo Yeah, but sometimes it doesn't look natural, just bad :D It's more about avoiding obvious clipping, showing stuff closer to the way our eyes would see the scene.
@@oleberek REDcode or slog3 debayered to DWG down to rec709 after, so your feedback is only valid on the assuption that I grade display referred, which I don't :)
Cullen. I'm using your DCTL tool, set exactly the same as in this video, and no matter what I cannot get my screen OR waveform scope to look like yours, to replicate. Why might that be?
Hello Cullen, thanks I feel less dumb now hehe I also have one request, can you make a video on how to grade F-log (Fuji) please?
What I get confused by, is people saying particular digital cameras (eg. Alexas and Blackmagics) have "good highlight rolloff". I think what I would want and expect from a digital camera is an accurate representation of the light hitting the sensor. Then rolloff can be "creatively" applied in post as per the video. I wouldn't want, or expect, a "professional" digital camera to be doing any kind of rolloff. Am I confused or are they?
The difference is clipping, without the roll off on capture it will hard clip and all detail ( stops ) will be gone forever.
To avoid clipping, what you need is dynamic range. I'm sceptical that it makes sense to talk about digital cameras having an additional, separate property called "roll off".
I think there might need to be a distinction between what type of footage is being shot by these cameras. You're right that you'd want a high end digital camera to accurately record the light hitting the sensor, and if we're shooting RAW, that's exactly what we're getting. RAW, linear gamma files, just as light physically works in the real world. However, if we're shooting LOG footage, we are left with a digital file that no longer is representative of how light works in the physical world, we're left with a large gamma curve on our footage that is meant to maximize the recording of dynamic range of the scene. LOG gamma curves can be very different from manufacturer to manufacturer and I'm sure some some film makers and colorists have preferences to which LOG footage they prefer. So it's possible that "good highlight rolloff" is referring to the way the camera records it's LOG footage.
@@verybutton If you have 22 stops in front of you and the camera captures at best 12 stops. It's gonna clip.
@@JimRobinson-colors Exactly. At the point of capture, it's about dynamic range, not "roll-off". It seems to me that roll-off isn't (or shouldn't be) something that happens as part of digital image capture. Unlike film, where roll-off is an inherent physical property of the film stock.
Or maybe I'm confused somehow.
Fantastic explanation.
Unfortunately, I stop learning Davinci cause my cheap laptop was broken😢, if anyone in this comments section has a second laptop which you don't use it anymore and its still working, Im gonna appreciate if you donate it to me. Just let your responds here, and we can communicate further, and my base is here in Indonesia. Honestly, I want to learn seriously Davinci from this channel, and I cant buy the new laptop cause my budget is not enough. Thanks all
informative ty!
I think your approach is obviously correct, but the way in which you are explaining it is not as clear as when I saw Walter Volpatto explain it. Walter just put a stepped gradient up that represented even stops and identified the color space and middle grey. Like I have seen you do - then while all the stops are on the curve he would count 3 stops below and tag it and then count 3 stops above and tag it on the curve. Then lower the highlights on the curve and raise the shadows - it shows the compression of the stops forming the roll off on the highlights and a toe on the shadows.
Without using the CST, the rolloff contrast worked on all the footage. I'm not comparing your teaching methods, but I think that your demo was over complicated for no real reason. The steps in linear then adding a CST with luminance mapping, is over complicating to me. It is possible that others might not see it. But the process that I explained above - you can easily see the middle stay consistent and see the luminance compression visually.
Thanks for sharing another method, Jim. I think both approaches help conceptualize the concept, and then give you strategies to put it into practice.
thx 🤟🏼
Don't think i can get tired of seeing Cullen's face lol.
I wish I could join this live. I'll see if I can because I have some interesting questions about highlight roleoff and color management.
Join him live first thing Friday morning California time. Usually somewhere around 6am (varies, RUclips notifications however are fickle), then ask your questions.
In Friday morning Grade School live Cullen discusses your questions, and adds extra insight to the previous weeks topics.
Lol. Yeah, that's me.
FTW
So....an S curve basically