4 ms·

Very cool!

• m0th87 17y ago
  Significantly optimized: import pydroid, operator INIT_COLOR = 473428 BELL_COLORS = set([12702690, 8950940, 9016477, 9017249, 9018536, 12505567, 12505568, 11715538, 10862803, 12965603, 12439776, 12834019, 14411502, 8953000, 5401724]) CANVAS_WIDTH = 751 CANVAS_HEIGHT = 501 CHECK_WIDTH = CANVAS_WIDTH / 2 - 50 def find_init_color(screen, xr, yr): for x in xr: for y in yr: if screen.get_color(x, y) == INIT_COLOR: return (x, y) def init(): while True: screen = pydroid.bitmap.capture_screen() first = find_init_color(screen, xrange(0, screen.width), xrange(0, screen.height)) last = find_init_color(screen, xrange(screen.width - 1, -1, -1), xrange(screen.height - 1, -1, -1)) if first is not None and last is not None and last[0] - first[0] >= CANVAS_WIDTH and last[1] - first[1] >= CANVAS_HEIGHT: return first, last def triangle(base, width, height, top_left, bottom_right): for i in xrange(0, height): y = base[1] - i diff = i * width / height if y > bottom_right[1]: continue if y < top_left[1]: break for x in xrange(base[0] - width + diff, base[0] + width - diff): if x > top_left[0] and x < bottom_right[0]: yield (x, y) def check(first, last, prev): screen = pydroid.bitmap.capture_screen() if prev is not None: for y in xrange(prev[1] + 20, prev[1] - 20, -1): for x in xrange(prev[0] - 20, prev[0] + 20): if screen.get_color(x, y) in BELL_COLORS: pydroid.mouse.move(x, y) return (x, y) for point in triangle(prev, CHECK_WIDTH, 150, first, last): if screen.get_color(point[0], point[1]) in BELL_COLORS: pydroid.mouse.move(point[0], point[1]) return point for y in xrange(last[1] - 100, first[1], -1): for x in xrange(first[0] + 1, last[0]): if screen.get_color(x, y) in BELL_COLORS: pydroid.mouse.move(x, y) return (x, y) def main(): print 'Searching for Winterbells...' first, last = init() prev = None print 'Found at %s, %s. Running.' % (first, last) while True: prev = check(first, last, prev) if __name__ == '__main__': main() It still only scores in the thousands because one of the bell colors sometimes maps to the rabbit, but it's at the point where it scores better than a newbie. Instead of looking from the bottom up for bell pixels, it tries to look in the following regions in order: 1) The area near where it last looked. 2) A triangular area above where it last looked, which is a heuristic guess of where the rabbit would be able to jump to. If those fail, then it looks from the bottom up for a bell. Okay that was fun, but it made me realize I really need a day job.
  • meese_ 17y ago
    Cool :) Just a minor nitpick: you may want to try the library method bmp.find_color() instead of your find_init_color() (http://msanders.github.com/autopy/documentation/api-reference/bitmap.html#bmp.find_color http://msanders.github.com/autopy/documentation/api-referenc...), it should be significantly faster. Also, instead of iterating through the width & height manually, autopy offers this shortcut: for x, y in bitmap: do_something(bitmap, x, y)