Search code examples
pythonpython-3.xopencvtkinter

Flickering video in opencv-tkinter integration


I am trying to build a GUI using tkinter in Python 3.6.4 64-bit on Windows 8 by integrating opencv components into the program. I can get the video to play, but there is a significant flicker going on. Namely, a screen the same color as the native tkinter background shows up briefly several times a second. I've tested several cameras with similar results, and double-checked that the cameras work properly via native video playback software. Here is my code:

from tkinter import *
from PIL import Image, ImageTk
import cv2
import threading

cap = cv2.VideoCapture(0)

root = Tk()
def videoLoop():
    global root
    global cap
    vidLabel = None
    while True:
        ret, frame = cap.read()
        frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
        frame = Image.fromarray(frame)
        frame = ImageTk.PhotoImage(frame)
        if vidLabel: vidLabel.configure(image=frame)
        else:
            vidLabel = Label(root, image=frame, anchor=NW)
            vidLabel.pack(expand=YES, fill=BOTH)

videoThread = threading.Thread(target=videoLoop, args=())
videoThread.start()
root.mainloop()

Could anyone suggest what I might be doing wrong? I did hear that Tkinter doesn't always play well with threads, but that's about all I can think of. In response to a comment suggesting the flicker is caused by label updating, I added some code that still reads from video and updates the label within the loop, but uses an image loaded outside of the loop to update the label. The flicker then goes away, although (to my understanding) the efficiency of the loop and updating of the label hasn't changed. Here is the changed videoLoop function (with flicker gone): def videoLoop(): global root vidLabel = None cap = cv2.VideoCapture(0)

    ret, frame = cap.read()
    frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
    frame = Image.fromarray(frame)
    frame = ImageTk.PhotoImage(frame)
    
    while True:
        ret, lame = cap.read()
        lame = cv2.cvtColor(lame, cv2.COLOR_BGR2RGB)
        lame = Image.fromarray(lame)
        lame = ImageTk.PhotoImage(lame)

        if vidLabel:
            vidLabel.configure(image=None)
            vidLabel.configure(image=frame)
        else:
            vidLabel = Label(root, image=frame, anchor=NW)
            vidLabel.pack(expand=YES, fill=BOTH)

Solution

  • Solution: Make sure the image configure call precedes storing the image in the label image attribute. I've managed to solve this issue, however I don't fully understand WHY it works as I am a python/tkinter novice. I will post the solution for now and will update the answer when I manage to find a proper explanation to this behavior. My best guess is that the storing of the image in the label attribute is what actually causes it to update on the screen, while the configure method just declares that there will be an image attached, which causes the loop to have to go through another iteration before getting to the update image update statement. The below code works fine without flicker:

    from tkinter import *
    from PIL import Image, ImageTk
    import cv2
    import threading
    
    cap = cv2.VideoCapture(0)
    
    root = Tk()
    def videoLoop():
        global root
        global cap
        vidLabel = Label(root, anchor=NW)
        vidLabel.pack(expand=YES, fill=BOTH)
        while True:
            ret, frame = cap.read()
            frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
            frame = Image.fromarray(frame)
            frame = ImageTk.PhotoImage(frame)
            vidLabel.configure(image=frame)
            vidLabel.image = frame
    
    videoThread = threading.Thread(target=videoLoop, args=())
    videoThread.start()
    root.mainloop()