Iliah Borg
Forum Pro
Yes, broadly speaking that's it.
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Xrite Color Profiles ?
- Thread starter Lbs26
- Start date
Jack Hogan
Veteran Member
Good post Glenn. Two pet peeves:
1) Scene linear does not' look' dark, it only appears dark if the output medium is not properly color managed (i.e if it tries to apply a gamma curve where none is needed) , otherwise it will look 'properly' bright. It may on the other hand look dull but see 2) below.
2) Human vision expects a linear input. How could it not? Light behaves pretty well linearly right up until it hits the eyes. From then on the HVS behaves like it always does, non-linearly, based on a linear input.
The 'curve' before the input, may be necessary mainly because today's cameras capture a larger DR than the output medium, so it's an easy way to squeeze the former into the latter. Absent that it looks 'dull'. Though it carries its own problems, such as chromaticity shifts. One of DcamProf/Lumariver's better attributes is that it tries its best to compensate for such.
Jack
Last edited:
ggbutcher
Senior Member
- Messages
- 1,842
- Solutions
- 4
- Reaction score
- 1,487
But, isn't "perceptual" about what the brain does with the linear input?
TheGrammarFairy
Leading Member
Absolutely. You may love the results or you may hate the results but either way you will have learned just a little bit more about how your particular camera works and how color science is used on a professional level.
I predict you will be surprised at what a blue hour profile does to your photos but I won’t spoil the fun.
The times when a calibration shot of the Color Checker has been a life saver for me have involved shooting at big downtown festivals where there is a lot of cooking smoke in the air or where you have giant banners bouncing colored light everywhere creating a color cast. There have also been times when atmospherics create strange color responses. The problem usually affects skin tones the most, and a shot of your color checker is cheap insurance.
I predict you will be surprised at what a blue hour profile does to your photos but I won’t spoil the fun.
The times when a calibration shot of the Color Checker has been a life saver for me have involved shooting at big downtown festivals where there is a lot of cooking smoke in the air or where you have giant banners bouncing colored light everywhere creating a color cast. There have also been times when atmospherics create strange color responses. The problem usually affects skin tones the most, and a shot of your color checker is cheap insurance.
Last edited:
Lbs26
Well-known member
- Messages
- 109
- Reaction score
- 37
I ended up creating Daylight profile using Adobe DNG profile editor with the Xrite target,the results are very similar to Dcamprof/Lumariver...more eye pleasing...a touch of magenta and warmness to the overall tones,very subtle difference but i like it a lot more,and the red now is not Red/magenta it is exactly how i see it in the real world.
To my taste and my eyes Dcamprof/Lumariver and Adobe DGN are better.
To my taste and my eyes Dcamprof/Lumariver and Adobe DGN are better.
Jack Hogan
Veteran Member
Right. Linear input from the scene or, alternatively, linear output from an output medium, say a monitor. Ideally the output of the monitor is identical to what's coming from the scene; second best would be proportional to it. Meaning that the scene, the capture, the raw conversion, and the output render on the monitor are in the end ideally linearly related. We strive for the ideal but, alas, practice never achieves it (nor in many cases is it necessary to do so).
So if we had a 12 stop DR scene, a linear pipeline and a 12 stop CR monitor, would we need a tone curve for maximum fidelity? In an ideal world not.
On the other hand, a 12 stop scene out of an 8 stop CR monitor definitely needs some sort of tonal compression lest the output look 'dull'. See for instance linearly rendered Figure 13 before the tone curve here (vs Figure 15 after):
https://www.strollswithmydog.com/raw-file-conversion-steps/
Jack
Last edited:
Entropy512
Veteran Member
Just for reference:
As an experiment, I've taken a shot without a tone curve and converted it into a Rec.2020 HLG video, which then was fed to a Vizio P65-F1 - the results were, to say the least, impressive.
It's too bad we're a LONG way from having a consistent and well standardized way to deliver stills to HDR displays.
SigZero
Leading Member
Can you describe steps You've taken? I'd love to repeat that experiment but I'm really beginer in creating video files...
Br, Pawel.
Entropy512
Veteran Member
ffmpeg -loop 1 -r 24 -t 30 -i DSC02708_03.tif -vf pad=1438:2160,zscale=tin=linear:t=arib-std-b67:m=bt2020nc,format=yuv420p10le,pad=3840:2160:1200:0:black -c:v libx265 -preset medium -crf 26 -x265-params "colorprim=bt2020:atc-sei=18:colormatrix=bt2020nc" antelope.mp4
The TIF input was a 16-bit linear Rec2020 image
This requires ffmpeg to be compiled with the zscale filter (many binary releases of ffmpeg are not) - the pad parameters are, if I recall correctly, designed for a 3:2 landscape input... but maybe it was portrait????
Edit: A more recent attempt was:
ffmpeg -loop 1 -r 24 -t 30 -i DSC02236.tif -vf pad=3840:2160:302:0:black,zscale=tin=linear:t=arib-std-b67:npl=800:m=bt2020nc,format=yuv420p10le -c:v libx265 -preset medium -crf 26 -x265-params "colorprim=bt2020:atc-sei=18:colormatrix=bt2020nc" naturalbridge.mp4
If you don't set npl to something higher than 100, zscale will map a value of 65535 in the input TIF to 100 nits and the whole thing will be dark
--
Context is key. If I have quoted someone else's post when replying, please do not reply to something I say without reading text that I have quoted, and understanding the reason the quote function exists.
Last edited:
Just reset, then select linear profile and PV2. "Zero" of some adjustments is set with non-zero value. Brightness 50, Blacks 5, Contrast 50 and Curve Medium Cobtrast makes ACR pipeline true linear.
Iliah Borg
Forum Pro
You got it exactly wrong.
Here is our artice that explains the issue https://www.rawdigger.com/howtouse/overriding-raw-converter-default-adjustments-settings
Last edited:
Great article, Iliah. But you've missed the main point. ACR with Adobe Standard profile applies default tone curve. If DCP-profile doesn't have ProfileToneCurve tag, ACR applies the default one. Default tone curve pushes midtones exactly +1EV . Plus BaselineExposue. That is why you'v got 47% instead of 18%. Difference is 1.38EV, very close to +1EV after default tone curve and +0.35EV baseline exposure. By setting sliders to 0 you've got 18% gray of patch M by chance. The process wasn't linear. Later you measure values of another pathes of Q13, but you've used synthetic dng file with embedded dcp-profile. That syntetic dng image could have embedded dcp-profile with explicit linear tone curve. That is why you've got that RGB values. If you measured the values of the real image from the top of your article, you would get different values of non-linear process.
Iliah Borg
Forum Pro
Try your setting instead without modifying the dcp to make the embedded curve (0,0) (1,1) and prove the resulting numbers for G with gamma 1 form a linear progression, increasing 2x per stop.
"
To embed a linear curve into an existing profile, I use a very fast, reliable, and simple method:
> exiftool -ProfileToneCurve="0 0 1 1" -o new.dcp old.dcp
"
Are you suggesting that a user does the above (as in my comment to the article)? If so, IMHO you should have explained it explicitly, starting with what is "linear profile" in your "Just reset, then select linear profile and PV2. "Zero" of some adjustments is set with non-zero value. Brightness 50, Blacks 5, Contrast 50 and Curve Medium Cobtrast makes ACR pipeline true linear" and how to make one.
Last edited:
Iliah, i've made some research to explain you more clearly what I mean.
I downloaded raw image of studio shot on Nikon 4DS with ColorChecker24 and made some measurements and calculations. You can examine that measurements by yourself.
Here some results:
1. Max DE with your settings for linear process (profile Adobe Standard, Process Version 2, Blacks 0, Brightness 0, Contrast 0, Curve Linear) is 11,06.
2. Max DE with Adobe Linear profile and default (non-zero) settings (profile Adobe Linear, Process Version 2, Blacks +5, Brightness +50, Contrast +25, Curve Medium Contrast) is 6.61. Colors of gray patches exactly the same as with previous settings. But color patches are more close to reference values.
3. I've made the most simple profile Adobe ColorMatrix with only four image-related tags ColorMatrix1, ColorMatrix2, DefaultBlackRender and ProfileToneCurve. Only 22 numbers. Settings: Process Version 2, Blacks +5, Brightness +50, Contrast +25, Curve Medium Contrast. Max DE 6.52, very close to Adobe Linear. Just interesting.
Settings 1 from your article (Adobe Standard, Process Version 2, Blacks 0, Brightness 0, Contrast 0, Curve Linear) make process very close to linear, but not exactly linear. ACR applies default tone curve, but another settings applies some sort of "reverse" curve. Not very precise, but close enough.
Settings 2 with linear profile and non-zeroes values (profile Adobe Linear, Process Version 2, Blacks +5, Brightness +50, Contrast +25, Curve Medium Contrast) makes true linear image processing pipeline.
I have color chart that was measured with spectrophotometer. So I know exact Lab values of it patches. Settings 2 with custom profile allow me to get patches Lab values very close to measured values. I use it in reproduction process.
Last edited:
Iliah Borg
Forum Pro
...
That's not my settings, as shown on Figure 13 on our https://www.rawdigger.com/howtouse/overriding-raw-converter-default-adjustments-settings, we account for baseline exposure:
Figure 13. PV 2 (2010), zeroed-out defaults, 0.25 BaselineExposure subtracted
We are discussing here profiles created by X-Rite ColorChecker Camera Calibration tool. Fact is, the option to create linear profiles was added to X-Rite profiling engine SDK only May 11 2021, after much pressure from our side. Did it make to the end-user product? AFAIK it's a no. That triggered my response. If you read our page, modification to make a DCP profile linear is mentioned in my comments, as I already said.
I have many different cameras of different brands, a studio with stable lights, and I check things based on my own shots, taken in tightly controlled environment, against the target spectral response measurements. On a side note, Dan Margulis is famous not for being blind. With un-modified DCP zeroing-out the settings in ACR results in normal saturation and normal brightness, no acid colours any more.
I don't see this discussion as a useful one.
--
http://www.libraw.org/
Last edited:
ggbutcher
Senior Member
- Messages
- 1,842
- Solutions
- 4
- Reaction score
- 1,487
Frankly, I don't see how making anything other than a linear camera profile is useful. The profile is supposed to represent the camera's data, tone and color, for transforming into other gamuts and tonalities. Doing anything else in a camera profile to me is just a stupid pet trick to do processing work in a place not intended....
That's not my settings, as shown on Figure 13 on our https://www.rawdigger.com/howtouse/overriding-raw-converter-default-adjustments-settings, we account for baseline exposure:
Figure 13. PV 2 (2010), zeroed-out defaults, 0.25 BaselineExposure subtracted
We are discussing here profiles created by X-Rite ColorChecker Camera Calibration tool. Fact is, the option to create linear profiles was added to X-Rite profiling engine SDK only May 11 2021, after much pressure from our side. Did it make to the end-user product? AFAIK it's a no. That triggered my response. If you read our page, modification to make a DCP profile linear is mentioned in my comments, as I already said.
I think it is if there are things to be communicated about 'linear'. Thanks for engaging in it.
Iliah Borg
Forum Pro
Combining profiles and looks is a bad idea, you (and Kodak) are right.
Jack Hogan
Veteran Member
Those are fairly bad. With a linear rendition you should be getting about half that
Loading…
www.strollswithmydog.com
Last edited:
Entropy512
Veteran Member
Adobe's insistence on doing so appears to be the likely root cause of nearly all Sony color complaints...
Jack Hogan
Veteran Member
From the raw file for the Nikon 4DS above, not necessarily representative but SMI 88 (pretty good) as follows:
Left deltaE2000, right deltaE76
This is what can be expected from a 'good' sensor and a linear matrix profile.
Jack
Last edited:
Similar threads
Keyboard shortcuts
- f
- Forum