Margus Voll wrote:Idea is just great.
I will see that in real life after we get hands on to that feature.
We need to refine how the HTML "Quote" works... apologies for mis-attributions.
What most of us who have been working with original Camera footage and not exclusively locked edits which have no access to any guidelines or charts or anything whatsoever (and probably fighting someone else's uncalibrated "corrections") are really commenting on is that while the approach is conceptually sound, it just doesn't work in practice. That is, giving an aimpoint over a range of exposures and conditions, based on a single point sample.
If anyone out there is imagining that it will make grading a picture (meaning feature film) as easy as pushing the "UP" button in the elevator at the office tower; the edge of the cliff beckons.
The physics of the situation are not linear by a long, long shot, and that's kind of what a single snapshot of a chart provides, and assumes. A controlled test, shot at home or any environment will provide a transform that is valid only for that setup: light source, exposure, (stock, in the olden days), lens,
lens setting... you name it -- one variable will change
everything ----> that is what charts or for. Every setup. What happens in the field has nothing to do with how the camera looked on a test stand.
The common misconception about navigating CIE colorspace is that it is a flat plane. It is not. Its actually a hill when you plot a Z-axis against luminance (you only ever see "White" in the middle, don't you -- so where is "Black"?). Consequently, tracking color over an exposure range on any practical physical media has more switchbacks and hairpins than Pikes Peak. You don't even have to twist the T-stop to affect this relationship, it will happen with ambient incident sources, and this is before the light gets through the lens; after that, you have the nonlinearities of the image-forming media which diverge in other, even more interesting ways. To a certain extent, ACES is trying to codify this, but we are definitely not in the same territory where simple arithmetic can cope. Second order differential matrix transforms,anyone? C'mon, it doesn't get any better than that.
jPo