I've been thinking about that for a long time and I hope you all can help me.First things first,I'm not a professional shake user,so maybe there's something that I'm doing wrong.
As you all know an iff image,or similar,can contain a z channel,but it's an aliased channel,with a very poor grey gamma,which results in something really useless for highend composites.
There are a few workarounds for that,like creating an external z,rendering a 3d scene with a depth shader,or maybe with fog or with a black to white ramp projected in Y or....there are a thousand ways to achieve a good depth tga or tiff file.
I've created a quick example for this:
here are the color files:
and here are the depth images:
Ok,here the composition I've done:
With a reorder node,I've added the red channel of the depth images to themselves as z,and with a copy node I've added the new z channel of the depth files to the color files.
Result:I have a color file with an antialiased and beautiful z channel!! am I right?
ok,so here's the problem,and the name of the problem is that:
why do I always have jagged edges???????
look at this tree:
I've created an over node and layered the left file with the floor.Obviously it's perfect,as always in shake....but look at that:
Here I compare the over node and the zcompose node.
as you can see,the composition with the zcompose node is absolutely aliased,as if the image were a tipical iff with z.
Then,why did I render the depth files if i cannot have a perfect z composition?
I though that everything would be great doing a separate z,but I cannot get rid of these artifacts.
So,let me finish with this:
Is there a way on earth to work with good resolution files,with a good resolution z,and compose them all with a good z compositing?????or having jagged edges and artifacts is the price to pay for using depth images?...mmmm,sorry,I don't belive that!!
what's the trick?