I appreciate these detailed analyses, and thank you for the url and pdf. Lately I have been using Sigmoid more, because - as you point out - a bright sky is often presented more pleasing up front in Sigmoid. I wasn't aware of the precise functionality of the skew-slider either, good to know that it flattens/bends the shoulder only.
@@adabbleinphotography8721 It actually does impact both I think more than that curve simulation shows...if you take a test image that has a gray ramp and you range select the color picker so that the histogram is restricted to the ramp and you have it set to waveform... you will see a sigmoidal curve of that ramp..Now you can adjust contrast and skew and you will see skew positive puts more of the values towards max so expands the shoulder but it lifts the blacks... Going negative is the reverse. The transition to white is more gradual and the one to black is more sharp... its a nice little experiment to see how the curve works when showing it within the functioning DT pipeline
see my reply below and how you could visualize...I don't think that is actually true....positive skew will soften the transtion to black and steeper to white and the reverse...
@@emrg777 With "gray ramp" - do you mean a continuous greyscale? Unfortunately I don't have such an image. Tried with a grey sky with a bit of fall off - the result is a belt that moves up and down but doesn't chance width and height when the skew slider is moved - until it reaches the top limitor - then it is "eaten" away from the top to prevent out of gamut. Maybe Nicolas can show your suggested experiment in his controlled environment?
A key point when comparing sigmoid and filmic of which you did a nice job is the shift in the behaviour of that model from v5 which does not have the gamut handcuffs as they have been deemed to version 6 and 7 which do. This is where the salmon and magenta and even blue in the hightlights comes from or is exaggerated when you use the relative white to tone map the highlights. If you were to use v5 with no color preservation and safe shoulders I think you would have not had to do any of the reconstruction that you did... Its one of the main reasons that I don't use v6 or v7 when I use filmic. I like to manage gamut myself and its far more controlled at least for me to just use the gamut compression in the color calibraiton module... I also like the control that you have with latitude and the saturation slider in v5 and that is part of v6 too but v7 was simplified .... just making this point as that feature of the image is not a filmic trait but a trait in certain versions of filmic.....
You are right that each version brought its new and modified features. To tell the truth, I can't be bothered to go back to v5 but now you mention it, I should give it a whirl!
@@adabbleinphotography8721 Well you know I guess really its great to have all those options....for me 6 and 7 offer no real improvements. In 6 there was the tweak to manage gamut and add the safe shoulders so that you could get a similar look to sigmoid and then 7 worked to simplify things and so you still have the gamut handcuffs of v6 and then changes to the function of the latitude defining the region for the midtone saturation which was replaced by a slider that tweaked highlight saturation and the color preservation modes were removed so really it is about the same in v7 as v6 but with some attempts to simplify things.... for me i don't like those weird highlight things caused by these gamut control that you can correct of course and luckily they became less prevalent once the inpaint highlight reconstruction was added. V6 with clip highlights as the HR mode was often not good at all. Also fighting those salmon colours is not for me. I would rather have the rich yellows and oranges and then tame those to suit my taste than fight to achieve a nice deep orange if I wanted it... Sigmoid does simplify some of this but it really compresses highlights and so you do need to work a bit harder to restore that highlight detail which can become very flat when things are always set to fade to white....it is normally quite decent out of the gate but I find less ability to tweak it than you can with filmic wrt tone and I can use the primaries module with filmic if that is needed so for me I guess when I do use one of the tone mappers it is more often still filmic... Sigmoid is great in that with the default you really can just boldly work with exposure when you begin editing and the image doesn't break. It might be too light when you add too much exposure but because sigmoid will always fade to white it doesn't break the image and so you can sort of push and pull exposure more easily with sigmoid I think that you can with filmic which you would often have to go back and re adjust if you make a significant exposure change....anyway the thing I do really like is that we have all the options left open to us and the new ones did not replace but rather augmented what tools we have to work with...thank you as always for your great video's you do a wonderful job of explaining and demonstrating the technical aspects for those curious about them....
@@emrg777 Thanks Todd I really appreciate the details you went into. I appreciate that the more recent versions (6 and 7) of filmic have simplified the UI and give us a more robust curve… with tradeoffs. I did explore the color preservation options of v5 but didn’t get a result as satisfying as Sigmoid, which has other issues, like the flatness of highlights. I will now delve back into the older versions of filmic in preparation of the last part of the series. The pinkish/magenta highlights that remain after filmic need to be dealt with before filmic and the recent inpainting option really helps… and it is quicker than the laplacian method! The good old tone equaliser is helpful too.
![[ENG] Darktable 5.0 What is new ?](http://i.ytimg.com/vi/-gNm230e5pA/mqdefault.jpg)
![[ENG] Darktable 5.0 What is new ?](/img/tr.png)
It is the best practical user guide for these modules. I'm looking forward to the next episode.
Thanks a lot!
Loving this series 👍
Great!
thank you for filling the gap left in the documentation of DT. I did not understand skew untill now.
Thank you!
I appreciate these detailed analyses, and thank you for the url and pdf. Lately I have been using Sigmoid more, because - as you point out - a bright sky is often presented more pleasing up front in Sigmoid. I wasn't aware of the precise functionality of the skew-slider either, good to know that it flattens/bends the shoulder only.
Yes, there is little information out there about how it works.
@@adabbleinphotography8721 It actually does impact both I think more than that curve simulation shows...if you take a test image that has a gray ramp and you range select the color picker so that the histogram is restricted to the ramp and you have it set to waveform... you will see a sigmoidal curve of that ramp..Now you can adjust contrast and skew and you will see skew positive puts more of the values towards max so expands the shoulder but it lifts the blacks... Going negative is the reverse. The transition to white is more gradual and the one to black is more sharp... its a nice little experiment to see how the curve works when showing it within the functioning DT pipeline
see my reply below and how you could visualize...I don't think that is actually true....positive skew will soften the transtion to black and steeper to white and the reverse...
@@emrg777 With "gray ramp" - do you mean a continuous greyscale? Unfortunately I don't have such an image. Tried with a grey sky with a bit of fall off - the result is a belt that moves up and down but doesn't chance width and height when the skew slider is moved - until it reaches the top limitor - then it is "eaten" away from the top to prevent out of gamut. Maybe Nicolas can show your suggested experiment in his controlled environment?
A key point when comparing sigmoid and filmic of which you did a nice job is the shift in the behaviour of that model from v5 which does not have the gamut handcuffs as they have been deemed to version 6 and 7 which do. This is where the salmon and magenta and even blue in the hightlights comes from or is exaggerated when you use the relative white to tone map the highlights. If you were to use v5 with no color preservation and safe shoulders I think you would have not had to do any of the reconstruction that you did... Its one of the main reasons that I don't use v6 or v7 when I use filmic. I like to manage gamut myself and its far more controlled at least for me to just use the gamut compression in the color calibraiton module... I also like the control that you have with latitude and the saturation slider in v5 and that is part of v6 too but v7 was simplified .... just making this point as that feature of the image is not a filmic trait but a trait in certain versions of filmic.....
You are right that each version brought its new and modified features. To tell the truth, I can't be bothered to go back to v5 but now you mention it, I should give it a whirl!
@@adabbleinphotography8721 Well you know I guess really its great to have all those options....for me 6 and 7 offer no real improvements. In 6 there was the tweak to manage gamut and add the safe shoulders so that you could get a similar look to sigmoid and then 7 worked to simplify things and so you still have the gamut handcuffs of v6 and then changes to the function of the latitude defining the region for the midtone saturation which was replaced by a slider that tweaked highlight saturation and the color preservation modes were removed so really it is about the same in v7 as v6 but with some attempts to simplify things.... for me i don't like those weird highlight things caused by these gamut control that you can correct of course and luckily they became less prevalent once the inpaint highlight reconstruction was added. V6 with clip highlights as the HR mode was often not good at all. Also fighting those salmon colours is not for me. I would rather have the rich yellows and oranges and then tame those to suit my taste than fight to achieve a nice deep orange if I wanted it... Sigmoid does simplify some of this but it really compresses highlights and so you do need to work a bit harder to restore that highlight detail which can become very flat when things are always set to fade to white....it is normally quite decent out of the gate but I find less ability to tweak it than you can with filmic wrt tone and I can use the primaries module with filmic if that is needed so for me I guess when I do use one of the tone mappers it is more often still filmic... Sigmoid is great in that with the default you really can just boldly work with exposure when you begin editing and the image doesn't break. It might be too light when you add too much exposure but because sigmoid will always fade to white it doesn't break the image and so you can sort of push and pull exposure more easily with sigmoid I think that you can with filmic which you would often have to go back and re adjust if you make a significant exposure change....anyway the thing I do really like is that we have all the options left open to us and the new ones did not replace but rather augmented what tools we have to work with...thank you as always for your great video's you do a wonderful job of explaining and demonstrating the technical aspects for those curious about them....
@@emrg777 Thanks Todd I really appreciate the details you went into. I appreciate that the more recent versions (6 and 7) of filmic have simplified the UI and give us a more robust curve… with tradeoffs.
I did explore the color preservation options of v5 but didn’t get a result as satisfying as Sigmoid, which has other issues, like the flatness of highlights. I will now delve back into the older versions of filmic in preparation of the last part of the series.
The pinkish/magenta highlights that remain after filmic need to be dealt with before filmic and the recent inpainting option really helps… and it is quicker than the laplacian method! The good old tone equaliser is helpful too.