I'm trying to use opencv (cv2) to stream a webcam feed into a pygame surface object. The problem is the colors aren't displaying correctly. I think it is the type casting, but I'm having trouble understanding the pygame surface documentation to know what it expects.
This code demonstrates what I'm talking about
import pygame
from pygame.locals import *
import cv2
import numpy
color=False#True#False
camera_index = 0
camera=cv2.VideoCapture(camera_index)
camera.set(3,640)
camera.set(4,480)
#This shows an image the way it should be
cv2.namedWindow("w1",cv2.CV_WINDOW_AUTOSIZE)
retval,frame=camera.read()
if not color:
frame=cv2.cvtColor(frame,cv2.COLOR_BGR2GRAY)
cv2.flip(frame,1,frame)#mirror the image
cv2.imshow("w1",frame)
#This shows an image weirdly...
screen_width, screen_height = 640, 480
screen=pygame.display.set_mode((screen_width,screen_height))
def getCamFrame(color,camera):
retval,frame=camera.read()
if not color:
frame=cv2.cvtColor(frame,cv2.COLOR_BGR2GRAY)
frame=numpy.rot90(frame)
frame=pygame.surfarray.make_surface(frame) #I think the color error lies in this line?
return frame
def blitCamFrame(frame,screen):
screen.blit(frame,(0,0))
return screen
screen.fill(0) #set pygame screen to black
frame=getCamFrame(color,camera)
screen=blitCamFrame(frame,screen)
pygame.display.flip()
running=True
while running:
for event in pygame.event.get(): #process events since last loop cycle
if event.type == KEYDOWN:
running=False
pygame.quit()
cv2.destroyAllWindows()
The ultimate goal I have is to create a small photo booth application for a DIY wedding next year. I'm new to programming, but I've managed to cobble this together. I was also trying to accomplish this with VideoCapture, which outputs a PIL, which I also couldn't get to work with the surface object. I want to use a pygame surface so I can animate and overlay count-down text, borders, etc.
Update: The issue was that the cv2 function camera.read() returns a BGR image, but the pygame.surfarray expects a RGB image. This is fixed with the line
frame=cv2.cvtColor(frame,cv2.COLOR_BGR2RGB)
Also, when converting to grayscale, the following code works:
frame=cv2.cvtColor(frame,cv2.COLOR_BGR2GRAY)
frame=cv2.cvtColor(frame,cv2.COLOR_GRAY2RGB)
So, the function getCamFrame should now be
def getCamFrame(color,camera):
retval,frame=camera.read()
frame=cv2.cvtColor(frame,cv2.COLOR_BGR2RGB)
if not color:
frame=cv2.cvtColor(frame,cv2.COLOR_BGR2GRAY)
frame=cv2.cvtColor(frame,cv2.COLOR_GRAY2RGB)
frame=numpy.rot90(frame)
frame=pygame.surfarray.make_surface(frame)
return frame