Hello. Let’s say we have a standart 8 bit SDR monitor and a properly exposed photo of some low contrast scene with dynamic range lower than what our monitor can potentially show. What tone curve will approximate human vision best? Let’s ignore that the camera and lens are not perfect, a human eye only sees a fraction of an image at a time, and other things like that.
Thanks
@sammartins28 Welcome to the forum.
Before we get flooded by responses, what is your experience with RT, colour management, post-processing? What OS are you using? What are your processing goals? Could you provide an example raw image? The more we know about you, the easier it is to figure out the motivation behind your question and to give you a proper answer.
I’m using RawTherapee since version 5.7 and I’m using Linux. This is just an experiment, I understand it won’t look as good as a high contrast curve. For an example image, this one looks suitable - Urdenbacher Kämpe in snow
The Human Visual System expects light to behave as it always does on its way to the retina, linearly. Therefore in theory the best our capture and display systems can do is behave perfectly linearly, including any curve in between.
The HVS will then proceed to process those linearly acquired photons as it always does: non linearly. But that’s a different question altogether.
In practice we have to contend with the fact that current capture and display systems have a limited range, and that’s where Tone Reproduction Operators (e.g. curves) come into play to squeeze a potentially larger range into the smaller one, invariably messing things up. YMMV
Jack
3 Likes
I don’t think that you can do a completely isolated experiment: if you’re looking at the monitor, the displayed image will be surrounded by your room (bright or dark, but not the natural surroundings of the original scene), and the brain’s processing will be different than it would have been on the scene.
1 Like
I suppose that human vision brightness is non linear in regard to the luminance. This provides a subjective impression from the scene.
When squeezing the luminance of the scene through the capture and display system and viewing conditions, human impression will be quite different.
I think it is there where tone curves and other processing like CAT adaptation intervene to try to reproduce the original subjective impression. Those processing cannot be linear.
If you stay linear, impossible to reproduce the impression of a scene.
I should add that being realistic is not always desirable. If you look at films, artistic intent is fundamental to provide a specific mood.
The answer can only be obtained through subjective experiment.
And welcome @sammartins28 in the rabbit hole of color management.
1 Like
Based on your short answer, filmic may interest you; particularly the module in darktable. It is one of the most popular topics in the past 3 years. In RawTherapee, there is a set of match curve presets.
Filmic is primarily a kind of dynamic compression perhaps with some correcting tone curve. Is it also implementing some part of color adaptation model?
The OP is interested in human vision and not dynamic compression.
Thank you for your answers. I want to clarify that I’m more interested in contrast rather than in color, and I want to use a scene that is already flat, so that we don’t have to compress the dynamic range
They eye is only a small part of seeing. Even when staring at one thing in a scene your brain is dynamically interpreting and patching up what is recorded by the eye. Contrast is perpetually adapted by the brain and the fact that you can’t consciously fix your pupil to a size.
So a photo is quite different from seeing. But we also see a photograph (on paper, alu or device) how we see a photograph is set by conventions of photography itself. We interpret what looks right based on our experience of other images.
So i the end it’s quite a difficult question to answer 
1 Like
I’m not familiar enough with the ins and outs of RawTherapee’s pipeline, but if the original scene is not very bright you can probably duplicate exact original scene brightness ratios by keeping everything linear up to the point where the output color space gamma is applied, which is the inverse of what the OS+monitor gamma.
As long as the shadows are not too dark you’ll have a good facsimile of the original scene.
1 Like
Two stimulating thought experiments tangentially related to this question:
-
Shortly after sunrise you are sitting on the bed of your alpine chalet, watching the beautiful scenery come into life through the only window in your otherwise darkish bedroom. The window is north facing, 65" on the diagonal with a 16:9 aspect ratio. Now you cover the window perfectly with a state of the art Monitor of the same size.
-
It’s way past bedtime in the same room, it’s pitch black except for the red power led of the monitor. The led looks to you like a normal in-gamut red, meaning that it looks to be the same shade of red as this morning before the TV was moved to cover the window. You silently curse me: you have been sitting here in the dark for an hour without being able to fall asleep because I asked you over a glass of your favorite beverage what color temperature and CAT should be used to represent said red led ‘properly’ during raw conversion.
I don’t know about that. It sounds like he wants to bring a flat image to a contrast that is natural to the eye. filmic can be a part of that. We 
s on the forum tend to over-answer, so I pivoted mine.
To strictly answer this question, there are 2 of them:
- the group of curves that can be found in Ciecam02/16
- in “Exposure” the “perceptual” mode of the tone-curve that uses Ciecam
More generally, the problem of rendering an image (color, contrast…) that should at best approach the human eye is obtained with CAMs (color appearance models).
But, before that all the process will contribute to a good management of color and contrast.
I enclose here a summary on the importance of the RGB / XYZ / Lab / Ciecam modules.
https://rawpedia.rawtherapee.com/Toolchain_Pipeline#Colorimetry
Some modules have a particular importance on this subject:
-
the 2 modules " wavelet levels " and " local adjustments wavelet " allow to act on the contrasts without any curves. Of course they may seem complex…but the learning curve is worth it.
-
The “denoise” modules, particularly the “Local Adjustment” module, take into account the criterion of human vision: noise is perceived more on the flat surfaces than in the structures. The tools must therefore allow to differentiate the acttion between flat surfaces and structures.
-
Retinex as its name indicates allows either actions on strong contrasts (main) or actions close to Dodge and Burn (Local Adjustments)
https://rawpedia.rawtherapee.com/Comparison_of_the_3_Rawtherapee_noise_reduction_tools
But other tools allow a different treatment of colors and contrasts and (some) are present in the development version and should be usable as soon as 5.9 is released.
Ciecam02/16 which is a color appearance “CAM” model, combining powerful color adaptations (CAT02 or CAT16) and tools to manage objects such as:
- a color will be perceived differently on a light or dark background, the darker the background, the more it will be necessary to reinforce the color
- an object appears brighter and more contrasted in direct light than in shadow.
- as the luminance increases, dark colors appear darker and bright colors appear brighter.
- colored objects appear lighter than achromatic objects with the same luminance. The most saturated colors appear the brightest
https://rawpedia.rawtherapee.com/CIECAM02
A Color Appearance Model (CAM) is a mathematical model that seeks to describe the perceptual aspects of human color vision i.e. the viewing conditions under which the appearance of a color is different to the corresponding physical measure of the source stimulus (RGB, XYZ are not CAMs, Lab is a CAM with limited possibilities).
These CAMs are used in several modules :
-
Ciecam02/16
-
Log encoding + Ciecam16 (“Local adjustments” RT spot local or Full image)
-
Warm-cool (“Local adjustments” RT spot local or Full image)
Log encoding (I was inspired by log TM in ART) is close to “filmic”. Both are based on the work of ACES.
Log encoding allows you to “compress” and “adapt” lights without the use of tones-curves. The approach is attractive, because it can in some cases compensate for a lack of Ciecam for HDR images.
Ciecam16 completes it, both for chromatic adaptation, but also to take into account the “scene conditions” and “viewing conditions”. It may seem obvious that these conditions are different on a smartphone outdoors, in a projection room, in printing, etc… Likewise the time of shooting, the latitude, the climate will have an important impact…Ciecam is there for that.
But beware, “Log encoding” like other “Tone mapping” modules also compresses the very high lights… Those that we tried to recover for example by “Color propagation”… And in this case it will remove or even accentuate the bad rendering of the tones. It is therefore necessary to provide a recovery device (here either “excluding spots” or “Recovery based on luminance masks”).
jacques
2 Likes
Sorry for such a late reply, but thank you all again very much, I’ll definitely check the tools you’ve mentioned