<img height="1" width="1" src="https://www.facebook.com/tr?id=414634002484912&amp;ev=PageView%20&amp;noscript=1">
Donate
Media in the Cloud: Ontology Guide
Donate

Dammit, Gamut, I Love You! | 8:30 AM on Wednesday

October 23, 2014

The moderator of this session, Arjun Ramamurthy of 20th Century Fox, tried to wake us with Rocky Horror Picture Show joke. Since this was a Fox production, we're sure he had proper permissions. 

Kenichiro Masaoka presented the work of NHK as they worked to approach Rec. 2020 in their UHDTV production chain. The title of the paper is: Color management for wide-color-gamut UHDTV production. New prisms, optimized color filters and in particular, color management was discussed from the work of their paper. All this in preparation of reproducing colors according to standards that are yet to be developed. He left nothing out, from graphs that showed problems with the primary point of lasers, the instance of developing a calibrated 2020 color monitor and the complications of rules and mapping for interchange from and to Rec. 709. Non-linear cyan was shown with moving pictures are the lightness is changed. As usual, there is no way to describe the detail in words that took so many amazing slides to show. We are promised the papers on Friday.

Among others, he should what constant lightness and hue while clipping chroma will do as a conversion standard (blown out begins to describe some areas, while others are fine or richer) compared to a color scale algorithm that they are proposing.

A question after the presentation brought a conversation about observer metamerics (or metameric failures to be more exact), which walks straight into the work of the next presenter. 

Quality Assessment Framework for Color Conversions and Perception is the paper of Francois Helt, a long-time observer and developer of restoration, archiving and transfer equipment, and now CTO with the new company Highlands Technologies. Highland Technologies is building unique quality assurance equipment for the cinema auditorium, and doubtless will get into other fields. 

Quantities and Qualities and their consequences began the discussion, including quotes from Rutherford and Aristotle. Perception viability for the audience goes back to the basic math, to eliminate the loss of basic information as transfers are made. Objective Quality Measures was the previous paper that this presentation builds on, adding distortions into the previous measures...vectors as perception models evolve.

Francois points out that the basic CIE models were derived from few (read, 17) pairs of eyes, all male, and the MacAdam ellipses – work representing the ability to judge differences in colors – was derived from the database of one pair of eyes. Following those points his graphs show the losses of various conversions, including ACES. Several slides are devoted to losses due to entropy through a transfer function. Boring, he admits, but true, and without having done these calculations we don't know what we are losing – or what we gain by going through the trouble of starting with higher bit depths.

New buzzword (perhaps only for this author) introduced: Barten model – a complicated model of the contrast sensitivity function as described by Peter Barten; Contrast Sensitivity of the Human Eye and Its Effects on Image Quality 

Lars Borg of Adobe describes his talk as being a more day-to-day description instead of the previous speakers' more far reaching and philosophical discussions. He wants to share color science information with non-color scientists who have to deal with wide-gamuts today.

His slides show a number of simple ways to 'see' this. At the end of this presentation I'll be able to write the Color Matching for Dummies. Grey response=good. Color saturation=all over the map, especially if you follow the standards. Elephants hiding in the corners that are only showing up as the color gamuts get broader...though a problem since 2008.

Buzzword Compliance note: "Dynamic Clip metadata"

High Dynamic Range Intermediate is the topic for Gary Demos, who is caught in traffic. Alas.

Worth the wait. First point...we are still living in a phosphor world. Amounts of emmission function instead of color matching functions will be more appropriate as we spread the primaries or even add primaries. A proposal to mimick the previous use of LAD, where the blacks can float, when film used 18% grey during production and 10% grey at display.

More precision, and he proposes 16bit half float, one gets new problems – one example being quick bright to dark changes bringing color after images.

How to get displays in different environments with different parameters and capabilities and settings (backlighting activated?), why not start with a very wide box, 2020 with a wide Cyan specification for example. Unsaid is that computing power in the monitor is finally capable of these types of internal monitoring.

Several slides are shown using different scenes with different renderings. [Note to auditorium staff: make the room darker for those of us with older eyes that don't have perfect contrast adjustment.] Gary invites us to comment on how the pictures from the powerpoint show differences on our various computers.

Buzzword takaway: Aesthetic Rendering (as different from Computer Generated Rendering). He makes the point that we are in new territory with new discoveries daily as we approach and use HDR.

Tag(s): UHDTV

Charles Flynn

Related Posts