sorry, I'm not familiar with this feature, but it highly needed reminding it is there to replace the old internal resolution.
the new updated documentation is good and I think I need to understand it a little bit more.
the spinner has ultra high maximum value, so is there cases that makes the render need higher value than 8 (50or100or1000) which is introduced in documentation as best value? or when will the render need super low value (1or2or3)? I mean if that too low or too high value is unusable or able to hurt the render, so imho it shouldn't be exposed or touchable by user, doesn't it?
it would be nice if there is a fix best value for most common scenes or is it what 8 means in documentation? :)
or.. maybe in future, if it is possible :D corona will have ability to detect if the render has bright contrast color and has jagged edges then internally highlight clamping will set to the best needed value or when the render is fine and no jagged edges then highlight clamping will turned off internally. but of course a checkbox to turn on/off this is exposed in devel/debug rollout.
or maybe I understanding it in a bad wrong way.
cheers :D
If you have high RGB values from very strong lights and you want them to be visible in the 0-255 8bit or 0-1 RGB color space, you need high highlight compression values. So let's say you are shooting a corona animation on Venus, then the sunlight will produce 500 real RGB values. If you don't want them blown out, but completly included in the visible color space, then you gotta use 500 Highlight compression, so 500 real RGB = 255 RGB and 400 real RGB something in the 200 range.
There is no perfect value, all of this stuff is 100% subjective and depends how much you want to be blown out to white and how much you want to exist in the visible color space. Low values are needed if you don't want to compress stuff and let them be white.
If all your image color is already between 0-255 (or 0 and 1 not in 8bit), then highlight compression won't do anything, since there is nothing to compress 0-255
The thing with jagged edges is a whole other topic. High contrast areas sample slower. If you have a high contrast light behind a dark silluette, then you are taking real RGB values of 50 and average them against real RGB values of 0.01. As you see, the influence the "normal" lit silluette barely influences the 200 times higher intensity of the light, thus the edge will be mostly dominated by the bright Staircase effect everyone hates. Either we need a magnitude more samples to average these stairs out, or we blur it with image filters.
Why highlight compression clears this out, is by bringing the ultra bright pixels down in intensity and only those not high values are now visible and the jagged edges are seemingly gone. In reality you just brought those values down, so they don't create this high energy inbalance.
« Last Edit: 2016-06-01, 14:43:26 by SairesArt »
Logged
I'm 🐥 not 🥝, pls don't eat me ( ; ; )
