Hey there! Let's fix up your screen recorder to get that crisp HD quality and the semi-transparent cursor circle you want. Here's how to tackle both issues:

1. Achieving HD Screen Recording #

The main reason your recordings aren't high-def right now is because you're intentionally scaling down the screen size and using a low frame rate. Let's adjust that:

• Use your actual screen resolution: Instead of hardcoding a smaller size like (960,540), grab your full display dimensions with pyautogui.size(). This ensures you capture every pixel of your screen.
• Boost the frame rate: 12 FPS is too low for smooth video—bump it up to 25 or 30 (standard for most screen recordings).
• Remove the resize step: You're shrinking the captured frame before saving it, which directly reduces quality. Ditch that cv2.resize() line entirely.
• Optional: Use a better codec: While XVID works, if your OpenCV setup supports it, using *'X264' will give you better compression with higher-quality output.

2. Making the Cursor Circle Transparent #

The issue with cv2.circle() here is that your frame is a 3-channel RGB/BGR image—there's no alpha channel to handle transparency. To fix this, we'll draw the circle on a separate transparent overlay and blend it with the original frame:

• Create a 4-channel (RGBA) overlay that matches your screen size.
• Draw the circle on this overlay with an alpha value (0-255, where lower = more transparent).
• Convert the original frame to RGBA, then use cv2.addWeighted() to merge the overlay and frame together.

Modified Full Code #

import cv2
import numpy as np
import pyautogui
import datetime
import win32api

# Get current timestamp for filename
date = datetime.datetime.now()

# Use full screen resolution instead of scaling down
SCREEN_SIZE = pyautogui.size()
# Increase frame rate for smoother video
framerate = 25

# Define codec - X264 offers better quality/compression if available; fallback to XVID
try:
    fourcc = cv2.VideoWriter_fourcc(*'X264')
except:
    fourcc = cv2.VideoWriter_fourcc(*'XVID')

filename = f'E:/project/videos/rec_{date.year}{date.month}{date.day}{date.hour}{date.minute}{date.second}.avi'
# Initialize VideoWriter with full screen size
out = cv2.VideoWriter(filename, fourcc, framerate, SCREEN_SIZE)

while True:
    # Capture full screen screenshot
    img = pyautogui.screenshot()
    frame = np.array(img)
    # Convert from BGR (pyautogui default) to RGB for OpenCV
    frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
    
    # Get cursor position
    _xs, _ys = win32api.GetCursorPos()
    
    # Create a transparent overlay for the cursor circle
    overlay = np.zeros((SCREEN_SIZE[1], SCREEN_SIZE[0], 4), dtype=np.uint8)
    # Draw circle on overlay: (0,255,255) is yellow, alpha=80 (adjust for transparency)
    cv2.circle(overlay, (_xs, _ys), 20, (0, 255, 255, 80), -1)
    
    # Convert original frame to RGBA to match overlay
    frame_rgba = cv2.cvtColor(frame, cv2.COLOR_RGB2RGBA)
    # Blend overlay with frame: adjust alpha weights if needed
    frame_with_cursor = cv2.addWeighted(frame_rgba, 1.0, overlay, 0.7, 0)
    
    # Convert back to RGB for VideoWriter (since most codecs don't support RGBA)
    frame_final = cv2.cvtColor(frame_with_cursor, cv2.COLOR_RGBA2RGB)
    
    # Write the frame to video
    out.write(frame_final)
    
    # Show the preview window
    cv2.imshow('screenshot', frame_final)
    
    # Exit on pressing 'q'
    if cv2.waitKey(1) == ord("q"):
        break

# Cleanup
cv2.destroyAllWindows()
out.release()

Quick Notes: #

• For cursor transparency, tweak the alpha value in (0,255,255, 80)—lower numbers make the circle more see-through.
• If the X264 codec isn't available, the code falls back to XVID automatically, so no compatibility issues.
• The preview window will show your full screen (you can resize it manually if needed, but the recorded video will stay full resolution).

内容的提问来源于stack exchange，提问作者arvind8