If you pass the -cropoutput argument, then no, it won't. If you use -cropoutput it outputs an image that is the size of the tile, or the crop, being rendered, with no black padding around it. If you want an image that is the full size of the image, i.e. a black image with a small area rendered within it, then don't use -cropoutput.
"Do crop region" in the GUI changes the area that is rendered, like a mask, but not necessarily the dimensions of the image file.
-tilex and -tiley change the area that is rendered, like a mask, but not necessarily the dimensions of the image file.
If "Do crop region" is enabled in the GUI, -tilex and -tilex are fractions of that already-smaller area, instead of being fractions of the whole image. This is because you might want to split a crop region further into tiles.
-cropoutput changes how that rendered area is put into the file. If "Do crop region" and/or -tilex/y were used, then the rendered area (the crop region or the tile) will be smaller than the full image. -cropoutput controls whether you want to just output a small image file with the dimensions of the rendered area, or output a full-size image file with black pixels in the areas that are outside the rendered area.
http://planetside.co.uk/wiki/index.php?title=Linux_Command_Line_ReferenceQuote
-tilex <minx> <maxx>
Limit the rendered region to a fraction of the image or crop region to be rendered. minx and maxx should be numbers between 0 and 1, where minx < maxx. For example, "-tilex 0 0.5" renders only the left half of the image or the left half of whatever crop region was already set.
-tiley <miny> <maxy>
Limit the rendered region to a fraction of the image or crop region to be rendered. miny and maxy should be numbers between 0 and 1, where miny < maxy. For example, "-tiley 0 0.5" renders only the bottom half of the image or the bottom half of whatever crop region was already set.
-cropoutput
Crop the output image(s) to the dimensions of the rendered region (defined by "crop region" and/or -tilex, -tiley), i.e. do not pad to the full image dimensions.
Matt