Can a Profile Express a Rendering Intent?

Home Forums General Discussion Can a Profile Express a Rendering Intent?

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
  • #36055

    • Offline

    I’ve been reviewing Charles Poynton’s document Perceptual Uniformity, Image State, and Picture Rendering. (PU_IS_PR.pdf) and wondering about how to use ICC tech to simulate a rendering intent.

    To summarize the paper, the history of imaging includes a legacy of a transfer function that includes an implicit contrast/color bump at the point of display to account for the typical performance / surround (viewing environment) which is coded as an approximately 1.2 power relation at the display to enhance perceived color and contrast given the departure of viewing environments from assumptions about production environments. According to the history Poyton describes, the features and dimensions of video predominate assumptions about end-to-end response for legacy images.  BT.1886 was created to allow modern systems the freedom to account for tradeoffs in black-level vs. power response in legacy CRT reference monitors.

    In seems that in an ICC workflow, we can shoot for a linear light end–to-end response, but we can’t use this to overcome the legacy of content which are coded with CRT television end-to-end assumptions of viewing rendering intent baked into the source, even in computer graphics, because computer display technology was derived from television equipment. If a reader carefully surveys the literature referenced across Poyton’s papers, he should find that “rendering intent” (to be clear I am not talking about ICC conversion Rendering Intent, I am talking about a psycho perceptual feature of television that’s part of video engineering) is a matter that gets pushed around into different subsystems of video and computing, and that keeping the facts straight is convoluted.

    Think what you may about Poyten’s opinions, but he convincingly argues that assumptions about viewing conditions are coded into the end-to-end response of imaging systems elsewhere that the receiver. This leads me to a question about how can to arrange an ICC-oriented  production system where the display expresses one face to the conversion engine and another to to the viewer; IOW how to code local rendering intent into a display profile — or maybe this is just a bad idea for reasons I need help seeing.

    For example, when I was first getting to know DisplayCal, I discovered that Apple’s build-in display profiles produce a pop-ier rendering on the display face than the sRGB xfer function in the profile. It looks very intentional on Apple’s part. Since reading Poynton, and examining my own system alignment, I can see Apple has made a wise choice as a lot of imaging looks better (subjectively) to the viewer who is uninformed about alignment details.

    This is a tricky subject, because an experienced technician will be able to see the signs of intentional psycho-perceptual distortion (a-linearity). In audio, it’s widely observed that as purity of reference systems reaches a zenith, their dynamics depart the most enjoyable reproduction.

    The key in a high-fidelity system is knowing how tolerate or even add distortion, and why you want to place it in one stage of the end-to-end response rather than another.

    Consider DisplayCal support for BT.1886 display alignment: what function does this serve? The ICC engine will dutifully map the image state into the calibration of the the display, so how is that different from using any other alignment? It looks like the point of 1886 is to give the technician the capability to simulate a specific era/generational of studio reference kit, but I see no way to actually apply it in practice.

    And my ignorance is key here. For example, I can imagine the use of Device Links as an approach to handling rendering intent, I just happen to have zero experience with the topic, so I’m asking for clues. Am I off-base? Am I fundamentally misunderstanding these concerns?

    It seems that HDR, esp DolbyVision is taking this into a whole new realm, where rendering intent can be handled with arbitrary degrees of freedom on a frame-by-frame basis.

    I’m ruminating without a clear train of though.

    Any responses / critique / questions appreciated


    Roger Breton
    • Offline

    I wish I could throw in my two cents, Wire… I met Mr Poynton a long time ago during a CIC meeting in Arizona. His book was on sale on the table, along with a slew of others… You’re getting into an area of color management that I could only relate to my Summer Short course at the Munsell Institute of Color Science… I’m sure Mark Fairchild would have an opinion on the matter because they have done a lot of work for Hollywood many years ago…


    • Offline

    A few years ago I remember there was a lot of controversy about mothers being birth raped in labour

    This meant like having internal exams done against will and being forced into stirrups etc.

    What would you do if this was your birth? Would you tolerate any of it?

Viewing 3 posts - 1 through 3 (of 3 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS