Actually I used 2 dimensions. Doing what you did won't skew the image properly.
You have to alter the x dimensions accordingly. Here is what I used in my script, though it's probably embarrassing math-wise.
PHP Code:
this.mapsize=screenheight*.75;
this.mappos={(screenwidth-screenheight*.75)/2,(screenheight-screenheight*.75)/2};
with (findimg(200)) {
dimension=2;
polygon={
thiso.mappos[0]+25,thiso.mappos[1],
thiso.mappos[0]+thiso.mapsize-25,thiso.mappos[1],
thiso.mappos[0]+thiso.mapsize+150,thiso.mappos[1]+thiso.mapsize,
thiso.mappos[0]-150,thiso.mappos[1]+thiso.mapsize
};
image="map.png";
layer=4;
mode=1;
alpha=.9;
}
EDIT: Oh... as far as player detection... I don't advise doing what I did. I suggest learning some trig and scripting a fully 3D polygon and learn how to 'place' something on it. It would be much more versitile.
When doing stuff like putting the player's horizontal position on the map, I use the basic formula (player.x/gmap.width)*mapwidth. What I did for this map is determine the mapwidth depending on the player's position vertically. I use a base mapwidth of the widest part of the map(in this case, the bottom) and subtract ((player.y/gmap.height)*mapheight)*(widestmapwidth-skinniestmapwidth).
If that makes sense. Either way, I don't advise doing it that way. Also, it seems the script doesn't work exactly the way I wanted. The position seems off, but I think it's because of the way Graal maps the image onto the polygon(as you can see in the screenshot, the bottom right is stretched horizontally a lot more than the top-right).
I wanted to do this script using eulerrotation... but it doesn't seem to work and I can't find anyone who knows how to use it.