diff --git a/axi/device.py b/axi/device.py
index 87db605..f8caa84 100644
--- a/axi/device.py
+++ b/axi/device.py
@@ -27,7 +27,7 @@ PEN_DOWN_DELAY = 0
 
 ACCELERATION = 8
 MAX_VELOCITY = 3
-CORNER_FACTOR = 0.001
+CORNER_FACTOR = 0.005
 
 VID_PID = '04D8:FD92'
 
diff --git a/examples/topo.py b/examples/topo.py
index 792251c..085969d 100644
--- a/examples/topo.py
+++ b/examples/topo.py
@@ -6,29 +6,102 @@ from PIL import Image
 Image.MAX_IMAGE_PIXELS = 1000000000
 
 import axi
+import math
 import numpy as np
 import sys
 
-WIDTH = 12
+LNG1 = -125
+LNG2 = -100
+LAT1 = 49
+LAT2 = 31
+
+WIDTH = 13
 HEIGHT = 8.5
-LANDSCAPE = False
-ROWS = 36
+LANDSCAPE = True
+ROWS = LAT1 - LAT2
 
 if not LANDSCAPE:
     WIDTH, HEIGHT = HEIGHT, WIDTH
 
 def remove_flats(path):
+    # return [list(path)]
     paths = []
-    for k, g in groupby(path, lambda p: p[1]):
-        if k > 0:
+    for k, g in groupby(path, lambda p: p[1] > 0):
+        if k:
             paths.append(list(g))
     return paths
 
+def crop(im):
+    w, h = im.size
+    lng1 = LNG1 + 180
+    lng2 = LNG2 + 180
+    lat1 = 90 - LAT1
+    lat2 = 90 - LAT2
+    pix_per_lng = int(w / 360)
+    pix_per_lat = int(h / 180)
+    x1 = lng1 * pix_per_lng
+    x2 = lng2 * pix_per_lng
+    y1 = lat1 * pix_per_lat
+    y2 = lat2 * pix_per_lat
+    return im.crop((x1, y1, x2, y2))
+
+def circle(cx, cy, r, n):
+    points = []
+    for i in range(n + 1):
+        a = 2 * math.pi * i / n
+        x = cx + math.cos(a) * r
+        y = cy + math.sin(a) * r
+        points.append((x, y))
+    return points
+
+def lat_label(text, y):
+    d = axi.Drawing(axi.text(text, axi.FUTURAL))
+    d = d.scale_to_fit_height(0.1)
+    d = d.move(WIDTH + 1 / 8, y, 0, 1)
+    # d.paths.append(circle(12.125 + d.width + 1 / 16, y - d.height, 1 / 48, 36))
+    d = d.join_paths(0.01)
+    d = d.simplify_paths(0.001)
+    paths = d.paths
+    # paths.append([(WIDTH, y), (WIDTH + 1 / 16, y)])
+    return paths
+
+def lng_label(text, x):
+    d = axi.Drawing(axi.text(text, axi.FUTURAL))
+    d = d.scale_to_fit_height(0.1)
+    d = d.move(x, HEIGHT + 0.125, 0.5, 1)
+    # d.paths.append(circle(x + d.width / 2 + 1 / 16, 8.5 + 0.125 - d.height, 1 / 48, 36))
+    d = d.join_paths(0.01)
+    d = d.simplify_paths(0.001)
+    paths = d.paths
+    paths.append([(x, HEIGHT - 1 / 8), (x, HEIGHT - 1 / 16)])
+    return paths
+
+def vertical_stack(ds, spacing=0):
+    result = axi.Drawing()
+    y = 0
+    for d in ds:
+        d = d.origin().translate(-d.width / 2, y)
+        result.add(d)
+        y += d.height + spacing
+    return result
+
+def title():
+    d = axi.Drawing(axi.text('Topography of the Western United States', axi.FUTURAM))
+    d = d.scale_to_fit_height(0.25)
+    d = d.join_paths(0.01)
+    d = d.simplify_paths(0.001)
+    return d
+
 def main():
     paths = []
     im = Image.open(sys.argv[1])
+    im = im.convert('L')
+    im = crop(im)
+    # im.save('crop.png')
+    print im.size
     w, h = im.size
-    data = np.asarray(im) / 255
+    data = np.asarray(im)
+    data = data / np.amax(data)
     # data = data ** 0.5
     lines_per_row = int(h / ROWS)
     for j in range(0, ROWS, 1):
@@ -37,19 +110,41 @@ def main():
         d = data[y0:y1]
         for q in range(0, 101, 25):
             print j, q
-            values = np.percentile(d, q, axis=0) * 0.9
+            values = np.percentile(d, q, axis=0) * 1.2
             path = enumerate(values)
             for path in remove_flats(path):
                 x = np.array([p[0] for p in path]) * WIDTH / w
                 y = (j - np.array([p[1] for p in path])) * HEIGHT / ROWS
                 path = zip(x, y)
-                path = axi.simplify_paths([path], 0.001)[0]
+                path = axi.simplify_paths([path], 0.005)[0]
                 paths.append(path)
+        lat = LAT1 + (LAT2 - LAT1) * j / (ROWS)
+        paths.extend(lat_label('%g' % lat, j * HEIGHT / ROWS))
+    for lng in range(LNG1, LNG2 + 1):
+        x = (lng - LNG1) / (LNG2 - LNG1) * WIDTH
+        paths.extend(lng_label('%g' % abs(lng), x))
     d = axi.Drawing(paths)
+    print len(d.paths)
+    print 'joining paths'
+    d = d.join_paths(0.01)
+    print len(d.paths)
+    print 'sorting paths'
+    d = d.sort_paths()
+    print 'joining paths'
+    d = d.join_paths(0.01)
+    print len(d.paths)
+
+    d = vertical_stack([title(), d], 0.25)
+
+    # d = d.rotate(180)
+    d = d.rotate_and_scale_to_fit(12, 8.5, step=90)
+
     im = d.render(
         scale=109 * 1, line_width=0.3/25.4,
-        show_axi_bounds=False, use_axi_bounds=False)
+        )#show_axi_bounds=False, use_axi_bounds=False)
     im.write_to_png('out.png')
+    # d = d.rotate_and_scale_to_fit(12, 8.5, step=90)
+    d.dump('out.axi')
 
 if __name__ == '__main__':
     main()