Lab Curves not precise RESOLVE 14 [BUG] UPDATE (Also in .8)

Get answers to your questions about color grading, editing and finishing with DaVinci Resolve.
  • Author
  • Message
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Lab Curves not precise RESOLVE 14 [BUG] UPDATE (Also in .8)

PostWed Jul 05, 2017 5:52 pm

UPDATE in version .8 the problem continues to exist


I found out the Lab curves in Resolve are not correct/simmetrical.

Procedure:

Create solid color with generator. Create a perfect neutral.

With the help of color space transform and splitter combiner reveal the the Lab channels.

Change simmetrically the slope, moving the endpoints only, in channels "a" and "b". A perfectly simmetrical action would keep the value of the color generated neutral.

With the help of RGB picker I achieve neutral. (Helped with Histogram Scopes).
When a neutral value is achieved, in all the 255 values of luminosity, the curves are blatantly not simmetrical.
Last edited by Jean Paul Sneider on Thu Aug 24, 2017 6:41 pm, edited 2 times in total.
Offline

Tom Early

  • Posts: 2664
  • Joined: Wed Jul 17, 2013 11:01 am

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostWed Jul 05, 2017 9:16 pm

It's the same for rgb, yuv and hsl. Top anchor point is wrong too. Same for all versions of resolve
MBP2021 M1 Max 64GB, macOS 14.4, Resolve Studio 18.6.6 build 7
Output: UltraStudio 4K Mini, Desktop Video 12.7
Offline
User avatar

waltervolpatto

  • Posts: 10502
  • Joined: Thu Feb 07, 2013 5:07 pm
  • Location: 1146 North Las Palmas Ave. Hollywood, California 90038 USA

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostWed Jul 05, 2017 11:53 pm

Why you're expecting to be symmetrical?
W10-19043.1645- Supermicro MB C9X299-PGF - RAM 128GB CPU i9-10980XE 16c 4.3GHz (Oc) Water cooled
Decklink Studio 4K (12.3)
Resolve 18.5.1 / fusion studio 18
GPU 3090ti drivers 512.59 studio
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostThu Jul 06, 2017 12:25 am

Tom Early wrote:It's the same for rgb, yuv and hsl. Top anchor point is wrong too. Same for all versions of resolve

Didn't know, wasn't able to check it. For older version, I just recently moved deep into Lab colorspace use.

waltervolpatto wrote:Why you're expecting to be symmetrical?


Well.. cause they are supposed to be. With neutral point in the middle between green/magenta and blue/yellow.
Offline

Rohit Gupta

Blackmagic Design

  • Posts: 1628
  • Joined: Wed Aug 22, 2012 5:00 am

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostThu Jul 06, 2017 12:24 pm

Set the Make Lum control is Primary palette to 0. This let’s you grade in 3 channel mode vs 4 channel mode. Can you confirm if this gives you the behaviour you expect?
Rohit Gupta

DaVinci Resolve Software Development
Blackmagic Design
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostThu Jul 06, 2017 6:20 pm

Rohit Gupta wrote:Set the Make Lum control is Primary palette to 0. This let’s you grade in 3 channel mode vs 4 channel mode. Can you confirm if this gives you the behaviour you expect?


Not sure what you mean,
Offline
User avatar

Dwaine Maggart

Blackmagic Design

  • Posts: 11241
  • Joined: Wed Aug 22, 2012 2:53 pm

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostThu Jul 06, 2017 9:29 pm

On the Color Wheels palette in the lower left corner of the Color page, in either the Primaries Wheels or Primaries Bars mode, in the bottom right corner of the palette is a setting called Lum Mix. The default value is 100. Change that to 0.
Dwaine Maggart
Blackmagic Design DaVinci Support
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostFri Jul 07, 2017 9:19 am

Dwaine Maggart wrote:On the Color Wheels palette in the lower left corner of the Color page, in either the Primaries Wheels or Primaries Bars mode, in the bottom right corner of the palette is a setting called Lum Mix. The default value is 100. Change that to 0.


No it doesn't solves the problem.
To be more precise if I use the lateral slider that inverts the curve, that should keep neutrals as they are changes them a lot.
Offline

PeterMoretti

  • Posts: 928
  • Joined: Sat Aug 03, 2013 12:12 am

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostFri Jul 07, 2017 9:48 am

Tom Early wrote:It's the same for rgb, yuv and hsl. Top anchor point is wrong too. Same for all versions of resolve


Tom, can you be a bit more descriptive? While it sounds important, I really don't understand what you're saying.
Resolve 14.3 Studio. GTX 970 with GeForce 390.77 driver. Desktop Video 10.9.10. Intensity Shuttle USB 3.0. Windows 10 Pro.
Offline

PeterMoretti

  • Posts: 928
  • Joined: Sat Aug 03, 2013 12:12 am

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostSat Jul 08, 2017 5:46 am

Jean Paul Sneider wrote:I found out the Lab curves in Resolve are not correct/simmetrical.

Procedure:

Create solid color with generator. Create a perfect neutral.

With the help of color space transform and splitter combiner reveal the the Lab channels.

Change simmetrically the slope, moving the endpoints only, in channels "a" and "b". A perfectly simmetrical action would keep the value of the color generated neutral.

With the help of RGB picker I achieve neutral. (Helped with Histogram Scopes).
When a neutral value is achieved, in all the 255 values of luminosity, the curves are blatantly not simmetrical.


Jean, can you explain more what you are saying? Starting with how you are color correcting a generator, AFAIK Resolve does not allow that? And maybe include some screen captures?

Thanks!
Resolve 14.3 Studio. GTX 970 with GeForce 390.77 driver. Desktop Video 10.9.10. Intensity Shuttle USB 3.0. Windows 10 Pro.
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostMon Jul 10, 2017 3:16 pm

PeterMoretti wrote:
Jean, can you explain more what you are saying? Starting with how you are color correcting a generator, AFAIK Resolve does not allow that? And maybe include some screen captures?

Thanks!


Sure. So:

Neutral and space transform.png
Neutral and space transform.png (182.88 KiB) Viewed 4463 times


I choose a random clip.

Node 9 I apply a color generator 120 gray. (as shown by the RGB picker value.

Node 2 Color space transform from 709 to Lab.

Node 8 the opposite action Lab to 709

In between a splitter combine with channels L, a and b.

MOdification  b Channel  keeping neutral.png
MOdification b Channel keeping neutral.png (178.05 KiB) Viewed 4463 times


With the help of the histograms and the rgb picker value I modify endpoints of the curve keeping the neutral. As shown it isn't symmetrical.

Non neutral.png
Non neutral.png (177.47 KiB) Viewed 4463 times


Trying to keep the symmetry it creates an obvious cast
Offline

PeterMoretti

  • Posts: 928
  • Joined: Sat Aug 03, 2013 12:12 am

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostMon Jul 17, 2017 3:33 am

Jean Paul, I've replicated what you're describing. Thank you.

I have to say that I'm not familiar with the Color Space Transform FX. It might be doing something wrong, as it adds a cast as soon as I apply it, but I really don't know. Because when I add it again to reverse the transform, the image again looks normal.

But I've found when just changing a node's color space to LAB and not using a splitter combiner, that no cast issues seem to present themselves.

I am curious as to why you aren't using that approach. I guess I don't understood why people use splitter combiners with LAB. Don't the individual R, G and B curves map to L, a and b anyway? Why do people go through the trouble of splitting out and disabling the channels?

And thanks again for the detailed reply above!
Attachments
LABSymetry.png
LAB a and b changes look symmetrical w/o adding color cast.
LABSymetry.png (100.75 KiB) Viewed 4384 times
Resolve 14.3 Studio. GTX 970 with GeForce 390.77 driver. Desktop Video 10.9.10. Intensity Shuttle USB 3.0. Windows 10 Pro.
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostMon Jul 17, 2017 3:04 pm

Like sometimes you want to work on a single channell in RGB, for example for controlled noise reduction, or recover details mixing channels for a bad video, or targeted masks, or use channels as masks.
In general the splitter combiner gives you more flexibility and powerful tools, but they need to act reliably, even more in lab than in RGB.

Too bad Dwaine Maggart seems to have left the conversation.
Offline

PeterMoretti

  • Posts: 928
  • Joined: Sat Aug 03, 2013 12:12 am

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostMon Jul 17, 2017 9:55 pm

But what does using a splitter combiner give you that just using unganged R, G and B curves don't give you?

(My hesitancy with using splitter combiner with LAB and disabling channels is that I don't know changes in the in Y curve will map in the a and b channels' nodes. As I'm not sure what that means to increase the luminance of an a or b channel when luminance in LAB is a separate channel and defined differently than it is in RGB or YUV. )
Resolve 14.3 Studio. GTX 970 with GeForce 390.77 driver. Desktop Video 10.9.10. Intensity Shuttle USB 3.0. Windows 10 Pro.
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostTue Jul 18, 2017 12:03 am

For example selective control on opacity, or channel mixing to restore heavily casted videos. Being able to apply diffent masks to L and ab channels.
You can do that with Some parallel combiner, but you get a messier node tree and still wouldn't be the same then a splitter combiner.

The Y works a a master control since the a and b become black and white.
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Re: Lab Curves not precise RESOLVE 14 [BUG]

PostWed Aug 16, 2017 11:52 am

In the Version .7 of the beta the problem still persist
Offline

Peter Chamberlain

Blackmagic Design

  • Posts: 13875
  • Joined: Wed Aug 22, 2012 7:08 am

Re: Lab Curves not precise RESOLVE 14 [BUG] (Also in .7)

PostWed Aug 16, 2017 1:44 pm

Jean Paul, what's the functional problem your encounter from what you found and what do you do to work around it?
DaVinci Resolve Product Manager
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Re: Lab Curves not precise RESOLVE 14 [BUG] (Also in .7)

PostWed Aug 16, 2017 4:20 pm

Mostly:
Every time I have to improve colors through steepening the curves I have to generate a neutral color. Steepen the curves as much as needed. Do test at various luminosity, take out the color generator see if it is what I want, otherwise again generate color, steepen or moderate it and do all over again. I have a power grade that I append at the end of the graph and start from that.

Generally I started to give up on the approach. Once it will be trustable I would appreciate a default add control point to the middle and the "invert" slider that now goes only "up and down".

In general not being sure that I can control consistently the neutral makes me shy away from using Lab controls.
Offline
User avatar

Jean Claude

  • Posts: 2973
  • Joined: Sun Jun 28, 2015 4:41 pm
  • Location: France

Re: Lab Curves not precise RESOLVE 14 [BUG] (Also in .7)

PostWed Aug 16, 2017 6:04 pm

There's bound to be something I did not care about in Jean Paul Sneider's approach but I tried something simple and I did not find the mistake.

Project REC709 from untitled project

A RGB clip-on 128-128-128 => compound clip

1_generator_compound.jpg



Color tab:
NOde 01: none
02: OFX color transform REC709 to LAB
Splitter
Combine
08: OFX color transform LAB to REC709


lab.jpg


Delivery: as is in TIFF
Import in PS
Source / delivery comparison

PS.jpg


(I have something wrong?)
"Saying it is good, but doing it is better! "
Win10-1809 | Resolve Studio V16.1 | Fusion Studio V16.1 | Decklink 4K Extreme 6G | RTX 2080Ti 431.86 NSD driver! |
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Re: Lab Curves not precise RESOLVE 14 [BUG] (Also in .7)

PostWed Aug 16, 2017 7:12 pm

You made no correction to the curves.
If you look ad ma posts with attachments I make the curve steeper. And the steeper it is the curve the more evident is the discrepancy.

PeterMoretti tried a different approch, and in the blue curve you can start to see, limited but clearly, that the two edge are not simmetrical to obtain the neutrality.
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Re: Lab Curves not precise RESOLVE 14 [BUG] UPDATE (Also in

PostThu Aug 24, 2017 6:42 pm

[UPDATE]
the problem persist in .8
Offline

Peter Chamberlain

Blackmagic Design

  • Posts: 13875
  • Joined: Wed Aug 22, 2012 7:08 am

Re: Lab Curves not precise RESOLVE 14 [BUG] UPDATE (Also in

PostThu Aug 24, 2017 9:14 pm

Jean Paul, your issue with LAB doesn't seem to be significantly affecting other users, many who like LAB and use it daily in Resolve, so we will monitor reports before prioritizing review of the current process.
DaVinci Resolve Product Manager
Offline

Jean Paul Sneider

  • Posts: 264
  • Joined: Sat Jul 01, 2017 5:46 pm

Re: Lab Curves not precise RESOLVE 14 [BUG] UPDATE (Also in

PostFri Aug 25, 2017 8:16 am

Ok, I imagined. What probably a lot of people would appreciate is a Lab pick. Thanks for all your work.

Return to DaVinci Resolve

Who is online

Users browsing this forum: akyhne, Bing [Bot], panos_mts, Rui Branquinho, tomnic and 185 guests