From the mailing list (see there for names, etc):

“I spent some time today trying to get a webcam feed into my psychopy proj, inside my visual.window. The solution involved using the opencv module, capturing the image, converting that to PIL, and then feeding the PIL into a SimpleImageStim and looping and win.flipping. Also, to avoid looking like an Avatar in my case, you will have to change the default decoder used in PIL fromstring to utilize BGR instead of RGB in the decoding. I thought I would save some time for people in the future who might be interested in using a webcam feed for their psychopy project. All you need to do is import the opencv module into psychopy (importing modules was well documented by psychopy online) and integrate something like this into your psychopy script.”

from __future__ import print_function

from psychopy import visual, event, core
import Image, time, pylab, cv, numpy

mywin = visual.Window(allowGUI=False, monitor='testMonitor', units='norm',colorSpace='rgb',color=[-1,-1,-1], fullscr=True)
mywin.setMouseVisible(False)

capture = cv.CaptureFromCAM(0)
img = cv.QueryFrame(capture)
pi = Image.fromstring("RGB", cv.GetSize(img), img.tostring(), "raw", "BGR", 0, 1)
print(pi.size)
myStim = visual.GratingStim(win=mywin, tex=pi, pos=[0,0.5], size = [0.6,0.6], opacity = 1.0, units = 'norm')
myStim.setAutoDraw(True)

while True:
img = cv.QueryFrame(capture)
pi = Image.fromstring("RGB", cv.GetSize(img), img.tostring(), "raw", "BGR", 0, 1)
myStim.setTex(pi)
mywin.flip()
theKey = event.getKeys()
if len(theKey) != 0:
break