Gamma vs Temperature

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

Benjamin de Menil

  • Posts: 330
  • Joined: Wed Feb 04, 2015 9:04 pm

Gamma vs Temperature

PostMon May 15, 2017 3:09 pm

If working on a non raw file - ProRes in this case - how does changing the temperature in resolve (on color wheel tab) differ from using the gamma color wheel to heat up or cool down?

I can get close to the same results using either. Obviously the gamma wheel has more options for which direction to shift things. But on the line between red/orange and blue/light blue, gamma and temp seem quite similar. How are people deciding when to use one vs the other?

Thanks
Offline
User avatar

JPOwens

  • Posts: 1511
  • Joined: Fri Apr 12, 2013 8:04 pm
  • Location: Victoria, British Columbia, Canada

Re: Gamma vs Temperature

PostMon May 15, 2017 6:21 pm

Benjamin de Menil wrote:how does changing the temperature in resolve (on color wheel tab) differ from using the gamma color wheel to heat up or cool down?


Might be better to use the term "mids" when describing an action with the middle wheel - technically you are not adjusting gamma -- there is a multiplier, yes, but it should not be confused with a color temperature shift.

Without a specific metadata reference, there is no way to determine what the scene-source color temperature was at the time of image-gathering. This was discussed not that long ago before there was a "Kelvin" slider in Resolve. What you actually have is a "mired-shift" control.

https://en.wikipedia.org/wiki/Mired

What you are doing is essentially adding a gel filter to your post-production matte box.

Where this goes off the rails is when you have clipping or crushing at either end of the overall image -- whites may be clipped unevenly to some (not found in nature, but chain-sawed at the top) value -- but they are still R=G=B as far as the electronics are concerned. A sky, for example that veers off toward cyan right before it is truncated. Black, not so much -- you expect 0=0=0 if the lens is capped, right?

So trying to offset a color temperature error day/tungsten/fluorescent can be approached using a mid-hue adjustment in a lot of cases. Where it gets weirder is in the mixed-light or cross-light circumstance, and this also happens with illumination that is not full-spectrum (stadiums and city street lamps).

But this is not gamma, which is a contrast relationship, and more relevant to display.

jPo, CSI
Offline

Benjamin de Menil

  • Posts: 330
  • Joined: Wed Feb 04, 2015 9:04 pm

Re: Gamma vs Temperature

PostMon May 15, 2017 7:40 pm

So practically speaking, how does Resolve's temperature setting differ from using the low mid high wheels? Does the temp setting apply changes in color evenly across the luminance range? Does it change color in some fundamentally different way from the low mid high wheels?

does your "going off the rails" warning apply specifically to the temperature param because it will shift the extremes away from pure white and pure black? And wouldn't this be counterbalanced by applying a saturation bell curve that desaturates the extremes of the luminance range? Or does changing temp change the clip's values for pure white and pure black in a fundamental way that desaturating wouldn't fix (might even make it worse)?
Offline
User avatar

Cary Knoop

  • Posts: 1437
  • Joined: Sun Mar 12, 2017 6:35 pm
  • Location: Newark, CA USA

Re: Gamma vs Temperature

PostTue May 16, 2017 3:51 am

Benjamin de Menil wrote:So practically speaking, how does Resolve's temperature setting differ from using the low mid high wheels? Does the temp setting apply changes in color evenly across the luminance range? Does it change color in some fundamentally different way from the low mid high wheels?

Rather than learning it in the abstract I think the best way to get a feel for what each control does is to practice with a color chart, make an adjustment and see how the change impacts the waveform and vectorscope and of course the colors on your monitor.
Offline
User avatar

JPOwens

  • Posts: 1511
  • Joined: Fri Apr 12, 2013 8:04 pm
  • Location: Victoria, British Columbia, Canada

Re: Gamma vs Temperature

PostTue May 16, 2017 4:16 am

Benjamin de Menil wrote:So practically speaking, how does Resolve's temperature setting differ from using the low mid high wheels? Does the temp setting apply changes in color evenly across the luminance range? Does it change color in some fundamentally different way from the low mid high wheels?


Like this:

https://en.wikipedia.org/wiki/Planckian_locus

When you apply a mired shift to an image, essentially you're waving the RGB gain around with a ten-foot pole, but not in a linear way. When you look at the CIE plot, what complicates this even further is that the traditional x-y plot is 2-dimensional and you don't see the Y-weighting projected along the orthogonal axis that presents the horsehoe. In a peerspective view, you see that the actual 3-D x-y-Y body is quite non-uniform:

http://www.gamutvision.com/docs/gamutvi ... plays.html

Something that you should notice when looking at the regular representation of the 1931CIE-RGB plot is:
Where's the black? Where's the grey scale? What you are seeing is the surface of a solid body. What is going on internally is way more complex than a mid-tone adjustment. This is why it is mathematically dangerous to implement a "color temperature" slider without knowing what the white point reference actually is. If you have a clipped signal where R=G=B, all squared off at the top... you have no real indication of the true image values in the original scene. Blue Dress/Gold Dress?

In that instance, you are just brewing your own cup of tea. If you are looking for mathematical rigour, you're not going to find it; both the slider and the midtone adjustment are purely cosmetic. If that sounds like "faking" a correction... that's pretty much the whole game. Haven't had a complaint yet from the viewing audience. :roll:

jPo, CSI
Offline

Benjamin de Menil

  • Posts: 330
  • Joined: Wed Feb 04, 2015 9:04 pm

Re: Gamma vs Temperature

PostWed May 17, 2017 6:35 pm

Thanks JP.

The info on those links is quite dense, and so maybe am I. I understand that the temperature is has a quite different effect from the color wheels. But beyond that, I'm lost.

Does someone have a real world example of how he or she would use the two controls for different results?
Offline
User avatar

Micha Clazing

  • Posts: 240
  • Joined: Sat Apr 01, 2017 3:26 pm

Re: Gamma vs Temperature

PostWed May 17, 2017 7:02 pm

JPOwens wrote:you have no real indication of the true image values in the original scene. Blue Dress/Gold Dress?

What's funny is that the first thing I did to that image is auto levels, which revealed it to be a blue dress. It's amazing that whatever camera that was used to shoot that even saved it in that sorry state to begin with, because it's so obviously wrong. There's a lot of complicated science behind illuminants, tristimuli and chromatic adaptation, but at the end of the day it's quite easy to get into the right ballpark with just the lift and gain wheels. If white is chopped off I just bring it back up with the highlight wheel to take that nasty colour shift out of it (as always, just shift in the opposite direction to your white balance adjustment to get the complementary colour).

Sorry for veering slightly off topic.
Offline
User avatar

Marc Wielage

  • Posts: 10901
  • Joined: Fri Oct 18, 2013 2:46 am
  • Location: Hollywood, USA

Re: Gamma vs Temperature

PostThu May 18, 2017 2:47 am

Benjamin de Menil wrote:Does someone have a real world example of how he or she would use the two controls for different results?

I find that the "temperature" and "tint" controls don't do a lot more than you can already do within the Primary controls (sometimes with a little curve adjustment). If you have actual Raw files, then the temperature controls in the Camera palette do a bit more than you can do in Primary.

I haven't seen your specific issue, but it is a challenge to take something with a baked-in "wrong color temperature" issue and salvage it. I often say, "we can make it better, but the image is never going to be great when it's been damaged this much."
marc wielage, csi • VP/color & workflow • chroma | hollywood
Offline
User avatar

JPOwens

  • Posts: 1511
  • Joined: Fri Apr 12, 2013 8:04 pm
  • Location: Victoria, British Columbia, Canada

Re: Gamma vs Temperature

PostThu May 18, 2017 5:39 pm

Marc Wielage wrote:I find that the "temperature" and "tint" controls don't do a lot more than you can already do within the Primary controls


And that's the "real world" -- we're stuck with the original numbers, and where there are gaps, its up to us to reconcile the difference. Temp and tint are Photoshop approaches that work where there are not significant issues and help out with managing many adjustments into a single knob.

Best short introduction in my experience was the few minutes that Bruce Graham spent with a "cone of color" on a chalkboard, Day One of daVinci Academy. He drew a sort of dunce cap with an origin at "0" and three legs (R, G, B) extending from it and describing a circle that represented the possible sum value of the maximum saturated hues that could result - a blunt representation of a colorspace. He elaborated that colorists get images with values that fall in that "cone" however truncated and twisted they might be, and that our job was to "straighten them out." At that point in technology, it was about balls and rings and nothing else. We were just getting secondaries, kilovectors and so on. The notion of color temperature was related to stock. Camera negative formulations, not RAW-encoded metadata. Every time a telecine was rolled into a new scene, it was another adventure. If this was a dailies situation, you had 6 controls to make it look acceptable to the production, pleasing to the cinematographer, useful to editorial, and not hoop the finishing colorist. The expectation was to deliver one hour of logged, sync-audio rushes every three hours.

You got good at assessing value placements, fixing them with whatever tools you had at hand, or you just didn't progress.

jPo, CSI
Offline
User avatar

Marc Wielage

  • Posts: 10901
  • Joined: Fri Oct 18, 2013 2:46 am
  • Location: Hollywood, USA

Re: Gamma vs Temperature

PostFri May 19, 2017 2:00 am

JPOwens wrote:Every time a telecine was rolled into a new scene, it was another adventure. If this was a dailies situation, you had 6 controls to make it look acceptable to the production, pleasing to the cinematographer, useful to editorial, and not hoop the finishing colorist. The expectation was to deliver one hour of logged, sync-audio rushes every three hours.

Doh, the dailies crew at Technicolor had to do 1 hour of logged & synced dailies in 2 hours. Tough to do when you're dragging 3 or 4 VTRs along simultaneously. Those were the days... :o
marc wielage, csi • VP/color & workflow • chroma | hollywood
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: Gamma vs Temperature

PostFri May 19, 2017 4:05 pm

for those controls to work correctly (mathematically) 3-4 things should be known by the software:

which color temperature is currently the image at
gamut of the image
tonal range of the image
and Kelvin differential you want to apply

then you can do a pretty tight math.

but at the state you are, I think resolve does not know 3 out of 4 of those values and it simply assume that you're in a "fixed" color space and that the starting point is also "fixed" (i.e rec709/5500k)

so, if u use the [temp/tint] you're just cooling/warming an image, not really changing the Kelvins.
If I have to get closer to a mathematical approach, roughly, if you know in which color space/gamut you are, you can transform that in linear light and use only the gain controls (luma mix = 0) and simulate the [temperature/tint] controls: that is closer to a mathematical approach without numbers...
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
User avatar

JPOwens

  • Posts: 1511
  • Joined: Fri Apr 12, 2013 8:04 pm
  • Location: Victoria, British Columbia, Canada

Re: Gamma vs Temperature

PostFri May 19, 2017 6:33 pm

waltervolpatto wrote:so, if u use the [temp/tint] you're just cooling/warming an image, not really changing the Kelvins.


That's what I'm talking about with the reference to "Mired shift" -- micro reciprocal degree
https://en.wikipedia.org/wiki/Mired

"In photography, mireds are used to indicate the color temperature shift provided by a filter or gel for a given film and light source. For instance, to use a tungsten light (3200 K) in natural light (say, 5700 K) without introducing a color cast, one would need a corrective filter or gel providing a mired shift of -137 microKelvins.
This corresponds to a color temperature blue (CTB) filter"...

Which you can match to a Tiffen or Rosco or whatever filter.

jPo, CSI
Offline
User avatar

Marc Wielage

  • Posts: 10901
  • Joined: Fri Oct 18, 2013 2:46 am
  • Location: Hollywood, USA

Re: Gamma vs Temperature

PostSun May 21, 2017 12:48 am

JPOwens wrote:\That's what I'm talking about with the reference to "Mired shift" -- micro reciprocal degree
https://en.wikipedia.org/wiki/Mired

I've been stuck in the mire before.
marc wielage, csi • VP/color & workflow • chroma | hollywood
Offline
User avatar

Tasio Liberakis

  • Posts: 124
  • Joined: Thu Feb 25, 2016 12:25 pm

Re: Gamma vs Temperature

PostMon May 22, 2017 6:09 am

Theoretically, tint/temperature should effect all pixels of the image evenly. Changing color temperature is a shift in color, best illustrated by the transmission diagram of a lee filter.

http://www.leefilters.com/lighting/colo ... s.html#204

Additionally, I would suppose that the color cast, works like in this mired calculator.

http://www.leefilters.com/lighting/mire ... lator.html

If I enter 3200 on the top slider and 5600 on the bottom slider the result is a cyan/ light blue, presumably if I shift colors in this direction, I would get the same result as the temperature slider, which moves in an orange/teal? axis, while the tint would move on a red/green axis perpendicular to the latter. Please correct me where I have it wrong.
W10 pro 22H2|X399 Taichi|TR4 1920x|64GB Hyper X@2133|
3060 12GB/Nvidia 531.41
Nvme system&scratch disks

Return to DaVinci Resolve

Who is online

Users browsing this forum: Baidu [Spider], Greg Agiannidis, panos_mts and 191 guests