Class 38 — Monday, November 22

Green screening

It’s simply smashing — For visitors on the Lawn — If they watch their steps

---

Look both ways

• Back or ahead

---

Agenda

• Problem solving

• Image manipulation

---

Downloads

• Module mash.py

• Tester smashing.py

---

Module mash.py

• Supports the creation of a new image that is a mash up of two existing images — bg and fg. The new image has bg as its background. The pixels in fg that are dissimilar to the back screen color are copied on top of the background.

---

Primary function

• combine( bg, fg, sx, sy, back_screen )

• Returns mash up of background image bg and foreground image fg, where the pixels in fg that are dissimilar to back_screen are copied on top of the background.

• The mash up uses location (sx, sy) as the spot in the background to start the overlay of fg.

def combine( bg, fg, sx, sy, back_screen ) :

  bw, bh = bg.size # size of background image

  fw, fh = fg.size # size of foreground image

  copy = bg.copy() # get a copy of background image

  new_image = copy.convert( 'RGB' ) # to be the basis of new image

  for fx in range( 0, fw ) : # consider every (fx, fy)

  for fy in range( 0, fh ): # possibility of the new image

  fspot = ( fx, fy ) # name spot (fx, fy )

  fpixel = fg.getpixel( fspot ) # get fspot pixel on fg image

  nx = sx + fx # determine the shifted nspot on new image

  ny = sy + fy # to paint fpixel

  nspot = ( nx, ny )

  # check whether fpixel should be painted on new image at nspot

  if ( colorable( nspot, new_image, fpixel, back_screen ) ) :

  new_image.putpixel( nspot, fpixel )

  return new_image # return the mash up

---

Support constants

SIMILARITY_THRESHOLD = 85 # minimum distance to be dissimilar

GREEN = ( 0, 255, 0 ) # default back screen color

---

Support functions

• distance( p1, p2 )

• Returns the the sum of RGB differences between colors p1 and p2. That is, the sum of the p1 and p2 red-level difference, green-level difference, and blue-level difference.

---

• similar( p1, p2 )

• Returns True if the distance between pixels p1 and p2 is less than similarity threshold; otherwise, returns False.

---

• is_background( p, back_screen = GREEN )

• Returns True if p is similar to the background color back_screen; otherwise, returns False.

---

• inbounds( spot, img )

• Returns True if spot lies on img; otherwise, returns False. That is, it returns whether the x-value and y-value for spot are nonnegative, the spot x-value is less than img width, and the spot y-value is less than img height.

• Note, the dimensions w and h of image img can be gotten using img.size.

---

• colorable( spot, img, p, back_screen = GREEN )

• Returns True if both spot is a location on img (i.e., inbounds) and color p is not similar to the background color back_screen; otherwise, returns False.

 

---