D3DShot VS python-mss

By using the Python package d3dshot, we can grab a screenshot of our RealFlight environment (we'll take just the part showing the downward-facing camera feed), and then send this image data (encoded using OpenCV) over UDP. On another computer we can have a script running with a UDP socket open and waiting to receive these messages. This script is representing the third-party peripheral, which in real life would be capable of obtaining the video footage on its own. Nonetheless, the peripheral now has its data to analyze. The important thing to note here is that this peripheral is self-contained. It is not part of the autopilot (it's written in Python, for one thing), and thus its hardware and software can be developed without any integration concerns, provided that it conforms to the autopilot's API. So while this "peripheral" currently exists on the same computer that is running the autopilot, this is by no means a constraint, and it will soon be moved to a separate piece of hardware. So what, you may ask, it the point of this peripheral? Well, that will just have to wait until next time. But I think it's pretty cool, so hopefully you'll come back to check it out. You're very patient, dear reader. That's what I appreciates about you. -Greg

VidGear provides an easy-to-use, highly extensible, Multi-Threaded + Asyncio Framework on top of many state-of-the-art specialized libraries like OpenCV, FFmpeg, ZeroMQ, picamera, starlette, streamlink, pafy, pyscreenshot, aiortc and python-mss at its backend, and enable us to flexibly exploit their internal parameters and methods, while silently delivering robust error-handling and real-time performance.

I'm using a python library called mss https://github.com/BoboTiG/python-mss

You can use PyGetWindow to get the info of a window based on its name, and then use your screenshot library of choice to get that part of the screen (personally I use mss because it’s fast and cross-platform).

mss can do this efficiently and cross-platform. What’s the use-case?

That was one of the main reasons I made my library pythonic-cv - I had solved the problem once (not super well), and realised I was going to have to do it again repeatedly, and thought that was a waste of time. While I was at it I added some additional nice features like iterating over a camera/video stream, using a context manager to handle cleaning it up/releasing it afterwards, and convenient processing and threaded recording loops. I've recently also added mss integration so you can use your screen as a video input in the same way as a camera or video file.

You may also want to consider using mss for taking screenshots efficiently, which allows specifying the screen region you want when taking the screenshot. It has OpenCV/numpy support.

That error shouldn't be happening, and is most likely some kind of error in mss. If you haven't already, I'd suggest trying to update your mss (e.g. python3 -m pip install --upgrade mss, and if that doesn't help then you should create an Issue on the GitHub page, mentioning your OS and what you're doing that's causing that error.

If you’re after speed I’d recommend mss for screenshots/recording, and tesserocr instead of pytesseract (note in particular the OpenCV option.

Your blog post claims to do image preprocessing with pyautogui, but I couldn’t find any image processing steps? Also, if you’re only using pyautogui to take screenshots, you might want to look at a library like mss, which is more efficient and also has support for multiple monitors :-)