filmic-rgb provided, what I think is, a useful insight into the dynamic range of the developed images … unfortunately that is not available when using sigmoid.
Is there another way of measuring the final data range?
I use a grayscale step image I created from a series of ISO 100 Expodisc exposures on my A7III. The lightest patch is where the green (and red/blue after white balance scaling) channel is just barely clipped and each patch below that has 1 stop less exposure. I combined center crops from white balanced 16 bit linear conversions and numbered the patches with reference to the camera’s full dynamic range potential according to the photon transfer curve from photonstophotos.net.
When you use the waveform scope it’s incredibly simple to see exactly what is happening in terms of dynamic range; e.g., here’s what the it looks like with no exposure changes or sigmoid (scope is showing it in sRGB gamma):
Here it is with sigmoid defaults:
and here you can see Sigmoid (with ACES 100 nit like preset) making use of 100% of the camera’s full dynamic range when exposure is scaled so that the camera’s “film-like” zone 5 is corrected to 50% luminance. The darkest patch is RGB 1 and the brightest is RGB 254. Perfect full tonal scale exposure according to Ansel Adams : )
1 Like
As I interpreted it you could assign a value in the data from your sensor to be considered middle gray more or less using exposure and then decide how to map the data on either side determining how much you would expand or compress it but I’m not sure that is a true reflection of the actual dynamic range of your image… I might not really be getting what you meant by “insight” or over thinking it… so that might not be what you meant…
This is an interesting series where the author defined the true DNR of their camera… I think it was an older Nikon if I recall…
I think Jandren made a comment that the data when using sigmoid goes from 0 to inf with every pixel having a valid rgb data… and is not tied to the the constructs of relative white and black as used in filmic… there was some good discussion I just don’t remember at the moment where it was…
I guess I did not fully explain what I had in mind.
With filmic-rgb the dynamic range of the image is directly displayed on the white (and black) relative exposure sliders. Adding the 2 values together allows us to see just how much overall compression we are achieving with the image.
1 Like
Ah yes well I don’t think the sigmoid works exactly like that so its not 100% analogous…
Its a fair question I guess how to compare those concepts…
Jandren’s remarks
- filmic depends heavily on the definition of a scene black and white point. The sigmoid curve is defined from 0 to inf, so there is no such concept as a scene black and white, just relative exposure to middle grey. It would thus be hard to make it work inside of filmic without redoing/removing both some functionality and UI elements.
The goal of pre-artistic processing is to retain as much of the scene as possible. Then sigmoid-like curves compress that by rolling-off range extremities infinitely so that values approach target black and white points. The closer we are to the extremes, the less detail we retain. But that is generally true of our vision. Due to pixel discreteness, we round down and up to 0 and 255, or thereabouts, when we export. So, in a sense, we do have black and white points, but the export defines them.
1 Like
I didn’t finish my write-up. Filmic-like curves take a different approach. While the sigmoid roll-off is elegant mathematically, one may want to tame extremities and give them more definition. That is why black and white points exist in the filmic module: to provide guidance on which points up to details are still important.
So, do you envisage that sigmoid will still evolve (maybe in the same way that filmic has changed).
Although I do like the simplicity of sigmoid, I always feel that the lack of ability to tweak its outcome is frustrating.
It will for sure evolve as more people use it 
As for displaying the dynamic range of the picture. You will technically be better of doing that at a earlier stage of the pipeline. So maybe we need something like a RAW histogram? Or define the extreme points of the tone equalizer as black and white point instead of exposure plus contrast. It would also be possible to add a histogram with the curve in the sigmoid module.
The question I have thought is if this helps you with the image editing or if you are just curious about the numbers?
1 Like
That’s exactly my experience. I’m currently trying to re-learn my workflow, editing many of the play raws. Mostly the sigmoid module is just activated at the end of the pipeline. Dynamic range has been defined earlier with exposure and tone equalizer. If I need to touch something in sigmoid, it’s the skew slider to find a balanced position for the image at hand.
I don’t feel like I’m missing something actually.
That said, I don’t have a ton of experience with filmic to unlearn.
1 Like
Good for you! It is beneficial to unravel workflows to learn more about and refine one’s processing. Looking forward to your Play Raw second takes. 
I actually find it pretty good in terms of range of adjustments and it works very well to just tweak a bit with the tone eq so I don’t feel like its missing too much… Statements like this can really be advanced with an example for people to work and comment on…
You might have noticed some of them already or perhaps not, as they are not of outstanding value. In fact, I chose to not bother the people with boring edits. So I’ve only submitted them, when I feel like something has been accomplished that hasn’t been captured…
I agree. It would be great if @davidvj gave us a tour of his thought or workflow process, with examples, that led to his questions.
I may have, but I forgot. 
Some people have done a filmic take and then a sigmoid one for kicks or comparison, but I wasn’t sure if you have.
Interesting idea even if I don’t yet know what would be the purpose. Checking white and black points? Dynamic range? But how would that help while processing so late in the pipeline.
I agree, I think the raw histogram is good to show the DNR of the incoming data. I think the OP was talking about the processed image using the mindset the if I set rel white and black to +4 and -4 then my processed image has a DNR of 8 and how do I get to something similar with sigmoid… at least I think that is what is being asked
As I see it … assuming that I use the exposure compensation in the camera to reduce the sensor overflow, I am in fact compressing the entire scene DR on the sensor. So the raw camera data that is being fed to dt needs to be ‘unpacked’ to correctly match the scene dynamics.
I am not sure that a histogram would be helpful unless it provided some sort of scale to assist in that unpacking.
This is a rather (very) low contrast image that I initially developed with filmic … I felt the results were fine but I decided to then switch and see if sigmoid could do any better.
So, with the 2 versions side-by-side I worked sigmoid until I felt that it could be just marginally better. It would have been nice if I could have said “sigmoid just provided a better range” …
In working sigmoid I had the advantage of using fillmic as some sort of target, so I simply thrashed around until there was some sort of minimal improvement … not really a desirable working process at all.
For me, a measurable result is helpful and being able to quantify the extraction process in terms of a DR range value might just be what I am looking for.
1 Like
Lovely photo…on a tangent but still as part of the tonemapping and managing the snow. I think here its gone to grey… Not sure if filmic or sigmoid does a better job but in an effort to get detail I often see snow like this and it just looks dirty gray. This would be a great playraw to let people have a go at the snow with both tools and their workflow to see the results…
If you go to the link @jandren provided for the curve testing. You can see how the curve shape changes when you adjust the sliders. From the shape you can understand what is happening to DNR and contrast. You can also plot filmic side by side to compare…