So here is a soldier in prone position:
http://s1.directuplo...09/tmseoir6.jpg (307 kB)
Let's take a look at the soldiers alpha layer:
And now here is a grass alpha layer I quickly sketched in photoshop:
http://s1.directuplo...09/jp5ucqp4.jpg (210 kB)
Using pixelshaders and combining global and local texture coordinates, we could multiply the alpha layer of the soldier with the grass layer, resulting in this:
http://s1.directuplo...09/qvojajgf.jpg (155 kB)
http://s1.directuplo...09/uaklkkkf.jpg (266 kB)
Now let's place our little fellow on some terrain!
Looking through the scope, without the multiplied alpha...
http://s1.directuplo...09/wjyay4ch.jpg (565 kB)
... and here it is with the new alpha:
http://s1.directuplo...09/iyqh868j.jpg (564 kB)
Here's what it looks like without scope...
http://s7.directuplo...09/3hd3afbd.jpg (164 kB)
... and here with the describben technique:
http://s14.directupl...09/doxrq6q7.jpg (164 kB)
Of course I simplified everything, so it is understandable to hopefully everyone, but I know this is possible since the day pixel shaders were invented. The performance impact should be minimal, considering no additional geometry has to be rendered and it can be done entierly on the GPU. I don't know if something like this is planned, so I thought I'd just share my thoughts on it.
Thanks for all the comments, I made a feedback entry on the tracker, here's the link:
(thx for the tip coulum)
Edited by Elcoo, 10 March 2013 - 01:12.