cap.read returns black image after adjusting focus

Hello,

I have this program to get the best focus for my image:

import cv2 as cv
import time

cap = cv.VideoCapture(0, cv.CAP_DSHOW)

cap.set(cv.CAP_PROP_FRAME_WIDTH, 3264)
cap.set(cv.CAP_PROP_FRAME_HEIGHT, 2448)
cap.set(cv.CAP_PROP_BRIGHTNESS, -12)
cap.set(cv.CAP_PROP_CONTRAST, 50)
cap.set(cv.CAP_PROP_SHARPNESS, 100)
cap.set(cv.CAP_PROP_GAMMA, 100)
cap.set(cv.CAP_PROP_WHITE_BALANCE_BLUE_U, 6500)
cap.set(cv.CAP_PROP_BACKLIGHT, 0)
cap.set(cv.CAP_PROP_EXPOSURE, -5)
cap.set(cv.CAP_PROP_AUTOFOCUS, 0)
cap.set(cv.CAP_PROP_FOCUS, 240)

time.sleep(0.3)

focus_found = 0

for z in range(240, 280, 2):
 #Setting Focus
 cap.set(cv.CAP_PROP_FOCUS, z)
 time.sleep(0.3)

 #Read Data from Camera and convert to Grayscale
 ret, data = cap.read()
 gray = cv.cvtColor(orig_data, cv.COLOR_BGR2GRAY)

 ...
 #Code to determine the focus. Setting variable focus_found ether to 0 or 1
 ...

 #Break if focus  is found
 if(focus_found):
  break
  

This works now for 2-5 steps for "z", but then i only get a complete black image from cap.read() even though "ret" is returning True. Do you know what i'm missing?

Start with the reported Python reproducer, especially VideoCapture(0, cv.CAP_DSHOW), the CAP_PROP_FOCUS updates, and cap.read(). Reproduce the focus loop while inspecting ret and the returned frame after each setting; done requires determining why later reads are black despite ret being True and documenting a reproducible fix or confirmed backend limitation.