Trouble with rendering depth

Hi,

I'm trying to create a render to texture of my scenes depth from an orthographic camera above the scene.

I think the camera is setup correctly and I can generate a texture but its not as I expect, it appears banded like the depth is repeating. I was expecting a more gradiented image where the pixels near the camera were black and those further away are white. Am I doing something wrong? using the wrong shader? ("Hidden/Camera-DepthNormalTexture")
I'm also using a custom camera projection matrix to get the camera to fit the geometry exactly.

 

Is the floating point depth being encoded as 8 bit rg or rgb? That would cause the banding but would add color to more channels. I see that this texture is argb32 and not a rfloat, the only way to get accurate depth would be to encode the float across multiple channels.