so how is that being finished precisely?

With devoted {hardware} paths particularly designed to optimize this widespread process, and never out there so that you can customise programmatically past turning depth testing and writes on/off, or altering the comparability used.

may I as an alternative select to write down the depth of furthest fragments (e.g. “Greater” depth comparability), and draw shade as closest as common.

No. There are usually not actually two separate writes occurring right here, only one. Either the depth take a look at passes, and each color and depth are written (if color and depth writes are enabled, respectively), or the depth take a look at fails, and neither are written.

So when you can change the path of the depth take a look at to less-than or greater-than, the identical take a look at is all the time used for the entire fragment, each color and depth parts.

You can management what depth to output within the fragment shader by writing a worth to the particular DEPTH semantic, however doing so disables some optimizations associated to “early z rejection”. The {hardware} pipeline can now not decide whether or not the fragment will likely be occluded till the fragment shader has been run, so it has to run the complete shader even for fragments which is able to by no means be seen.

You may obtain one thing much like this with a number of passes, rendering your content material as soon as with a “Less” take a look at and saving the ensuing depths to 1 buffer, then rendering your content material a second time with a “Greater” take a look at, and studying the buffer you beforehand saved to do no matter computations you need that depend on figuring out each the best and least depth values at some extent.

You can also wish to look into the strategy of “depth peeling” which is utilized in some order-independent transparency approaches.