2d loess question

Hi,

We have been trying to use loess on 2D data (basically a matrix) in the 
following way:

x <- 1:256
y <- 1:256
z <- data # input from data
z.loess = loess(z ~ x + y)

We get a 256 x 1 vector of fitted values with a 256 x 256 array of 
residuals,
but not a 256 x 256 array of fitted values.

Why would this be?  I think we are using loess incorrectly but can't 
figure out what is wrong.  I have looked at past messages on this 
mailing list and searched the web, without any more insight.  Any help 
would be much appreciated.

Thanks,
Saurav