How do I render a "richer" Z depth?

[zqw]

Are there some attributes somewhere that allow me to broaden the grey scale that Maya uses to render the Z Depth channel? My issue is... I've got a scene that I want to render in layers then do Z depth sorting later in compositing. The default settings do not give me a z depth channel with a large enough grey scale range to do much with it.

Any ideas?

Thanks, -Greg

[user_deleted]

If you mean the aliased edges, the answer is no, there is no easy way to anti-alias the z-depth channel. Unfortunately.
Cheers

[zqw]

What I need is more grey scale range in the depth channel. In other words, large objects in the scene are only identified in the depth channel with one grey scale values. What I need is an entire range of values that represent that object in z space.

I thought that by changing the clipping planes of my renderable camera, that that would also move the endpoints of the grey scale range, but no luck.

Is there a way to do this?

[user_deleted]

Oh, i see what you mean.
I have to disappoint you. I have no clue.
Good luck

[carmine-1]

fake it with a grayscale render with no lights, all surfaces with a pure white shader (illumination up all the way to white), black fog with clipping from the closest object to the furthest object. use that render as a fake z-depth. If i remember correctly, i made the "z-depth" with hardware render in the past when i needed a z-depth for compositing. much faster than waiting on a software rendered z-depth.

[gmask]

If you render 16 bit images will get twice the depth in your z-depth channel. However this does not eliminate the unaliased edges and also requires that you use a compositing sytem that handles 16bit images.

I would suggest that if you have a few objects in the forground that the jagged depth is a particular problem with that you render a seperate alpha pass to comp that element. A simple example would be if you had a table and for some reason you wanted to later comp an apple coming up through the table you would get a hard edge on the apple where it intersects witht he table. This would look better if you rendered the table and render the animation of the apple like you would to comp them with z-depth. Then you would setup a scene with the table and apple assign a flat black shader to the table and a flat white shader to the apple and use the luminance of this sequence as the alpha channel for the apple when comping it with the table thus resulting in a smooth intersection with the table.

Adrian

[zqw]

Jason at A|W gave me a good idea...

Texture the entire scene with a black and white ramp that is projected so that the objects closer to the camera are white and the objects far away are black. This way, you get a full range of gray scale values in addition to a high quality, anti-aliased render (unlike depth channels in RLA or IFF).

-Greg

[gmask]

Texture the entire scene with a black and white ramp that is projected so that the
objects closer to the camera are white and the objects far away are black. This way,
you get a full range of gray scale values in
addition to a high quality, anti-aliased render (unlike depth channels in RLA or IFF).

This technique is also good for creating certain displacement effects like a face pushing through a sheet.. although now you could do that with softbodies.

My question about using this for z-depth is do you have to convert the image to a specifc format to use it instead of the z-depth or can you trade one out for the other and it just works???

Also you'd loose access the texture id's and other z-depth tags if you needed them.