Alex Winker

Saturation vs. Luminance

Recommended Posts

I'm looking for the best way to affect an image the way a Saturation vs. Luminance curve would work in Resolve in order to approximate the way film deals with high saturation. My initial though would be a qualifier, but I wan't something more scalable than a SAT only qualifier with the offset rolled down, not to mention the range and fall off of saturation in the qualifiers in resolve are really small and can't be fine tuned as well as with a curve.

Does anyone have any tips or tricks on how they do something like this, or how you think this could be accomplished?

 

Also wondering if anyone has any insight into the maths or transforms Yedlin is doing here to control saturation and how they might be accomplished? seems to me that its similar to a SAT v LUM curve to a degree, but I'm not sure.

 http://www.yedlin.net/140914.html

SaturationMath.jpeg

 

Share this post


Link to post
Share on other sites

Any reason you can't use a lum vs sat curve for this? Lowering the overall density will get you access to more filmic red but you will need to twist it with a simple hue adjustment to create the specific red you would want. 

Share this post


Link to post
Share on other sites

The missing SAT v LUM curve can be simulated in Resolve by changing a node color space to HSL and lowering the G channel of the Blue Output (uncheck Preserve Luminance). The technic is listed in Juan Melara's video.

 

Share this post


Link to post
Share on other sites
18 hours ago, Tom Evans said:

Any reason you can't use a lum vs sat curve for this? Lowering the overall density will get you access to more filmic red but you will need to twist it with a simple hue adjustment to create the specific red you would want. 

I'm looking to affect the image in the inverse way that resolves LUM vs SAT curve works, which adjusts saturation based on luminance. I want to adjust luminance based on saturation as a way of managing highly saturated color for a more "filmic" response. 

Yeah I'm definitely thinking there is a hue shift involved, but If I were to try to make this a global correction, I don't think the desirable hue rotation would be uniform across colors. It almost looks like Yedlin's filmic desat is just a drop in the luma only channel.

 

51 minutes ago, Misha Denisov said:

The missing SAT v LUM curve can be simulated in Resolve by changing a node color space to HSL and lowering the G channel of the Blue Output (uncheck Preserve Luminance). The technic is listed in Juan Melara's video.

 

I've seen this - Thanks for the reminder! I need to do some testing to see if it that effects saturated areas uniformly. It would be nice to have a tool to affect it with as much control as a curve.

 

  • Like 1

Share this post


Link to post
Share on other sites
(edited)

@Alex Winker You can try to set node color space to HSL or to HSV. Then bypass channels 1 and 3 in that node and set Luma Mix to 0.
Now you can adjust saturation using Custom Curve. Add default points to fix Custom Curve position in lower level (at lower saturation level since color space is HSL (HSV). And then set the curve to something like highlights soft roll-off. This will compress highly saturated colors, because custom curve now act like SAT vs SAT curve.
Now you can switch HSL and HSV color spaces and see that one makes saturation compressed color brighter. while another makes it darker.
I don't remember, which color space exactly you should to use, so try both.

Edited by Anton Meleshkevich
  • Like 1

Share this post


Link to post
Share on other sites
(edited)

I would make sure that Luminance Mix is set to zero, even when only the G channel is selected in the node.

 

 

Edited by Cary Knoop

Share this post


Link to post
Share on other sites
(edited)

@Cary Knoop As I mentioned it :) Or you talking about something different?

Quote

Then bypass channels 1 and 3 in that node and set Luma Mix to 0.

 

Edited by Anton Meleshkevich

Share this post


Link to post
Share on other sites
(edited)

Bypassing the channels won't help to avoid the luminance mix interference (at least not in Resolve 16 beta).

Try it, make a bold change in the curves and now move the luminance mix from 100 to zero, you should not see a difference but it shows a difference.

 

 

 

Edited by Cary Knoop

Share this post


Link to post
Share on other sites

I know :) 
That's why I wrote about setting Luma Mix to 0 in my post. I wrote about exactly the same thing you are talking

  • Like 1

Share this post


Link to post
Share on other sites
Just now, Anton Meleshkevich said:

I know :) 
That's why I wrote about setting Luma Mix to 0 in my post. I wrote about exactly the same thing you are talking

Oh dear!  I am sorry, I completely missed that. 

  • Like 1

Share this post


Link to post
Share on other sites
(edited)

@Alex Winker Another interesting thing you can try is to add just one point to HUE vs LUM curve and drag it down a little. So the whole curve (a straight line) just move down. This won't make what you mentioned but it has interesting effect about sat/luma things.
It will make saturated pixels darker. I like how it looks on skin. 'Make awesome' magic button as for my eyes. But unfortunately increase noise. So it can't be applied 
every time.

This time node color space shouldn't be HSL or HSV of course.
No node color space and node gamma settings needed if your material is kind of gamma 2.4 and some rgb colorspace

Edited by Anton Meleshkevich
  • Like 1

Share this post


Link to post
Share on other sites

Another thing you could do is qualify a key on upper-mids and highlights, soften it to reduce artifacts, add a little NR, and desaturate there. The effect is different than a Lum vs. Sat curve and (to me) not as destructive if you're very careful. I do this all the time when I need to subtly go in the opposite direction and desaturate blacks without artifacts. But I try not to push it too hard.

Noted DP Steve Yedlin has some things to say about the "film look" and digital cameras, and he has some interesting theories and conclusions:

http://www.yedlin.net/OnColorScience/

  • Like 2

Share this post


Link to post
Share on other sites
(edited)
On 5/3/2019 at 3:02 AM, Anton Meleshkevich said:

to add just one point to SAT vs LUM curve

I meant HUE vs LUM curve of course.

2019-05-12 032545.png

Edited by Anton Meleshkevich
  • Like 1

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.