I'll have a stab at this but not really sure if this is the correct answer. wall:invisible (or wall subtype invisible) is not the same as the -visibility option. I think -visibility just controls whether the symbol can be seen in the 2d outputs. However I think of wall:invisible as equivalent to "wall does not exist" as that is the effect (generally) that it has in all outputs and scrap joining algorithm (passage interiors). It is mainly useful for indicating that an area is actually inside of a passage when making complex multi-scrap joins at junctions, so that the pdf's and lox models are rendered properly. But maybe it is a bug and they should behave the same? I have always used wall:invisible and usually get the result I desire for pdf, lox and most of the time for kml, but for complex arrangements it takes a bit of fiddling. I wonder how the join statement treats walls with -visibility off? Check that you do not have any -outline in/out or wall -reverse on/off type anomalies in your scrap. Bruce -----Original Message----- From: Footleg A curious thing I just discovered. One of my scraps was not showing the walls properly in the lox model. The cause was one wall line being set to invisible using the syntax: line wall:invisible If I changed this to: line wall -visibility off Now the lox model looks fine. I would have expected both these ways of hiding a wall line in the PDF would be processed the same way (both look the same in the PDF). But clearly something different is happening in the lox model. Footleg