WIP : Made a screen recorder and automated tutorial for use of waypoints

[risehr]

1. Made a screenrecorder.py file
2. Made a automated tutorial for use of waypoints(to be completed)

Here is a google doc file explaining the details:
https://docs.google.com/document/d/14nNaPpIvkWXEl3MwDzD_4BItpKhOaphvUSuP_rw5rxg/edit?usp=sharing

[ReimarBauer]

please move tutorials/requirements.txt to requirements.d/tutorials.txt

This file should only have later the new packages or different versions you need to use not the full mamba list of packages which are already installed.

We have already the requirements.d to collect lists for different purposes

[ReimarBauer]

next time when you want to merge changes from the main repository please do this by a seperated merge request. e.g. name it "merge develop to my branch"
It is than easier for us to look at your changes ;)

[ReimarBauer]

I guess that this is another mss, with an identical name as our project, or?

[risehr]

Yeah. This mss is (multi screen shot) Its used for grabbing images while recording. Since, it is a cross platform compatible, I used this. (And not pillow)

[ReimarBauer]

this is a constant on your system, is this needed for others?

[risehr]

I actually created this yml env file purposefully to help in the creating of a new environment for anybody wanting to test my code. It's not needed later. (I guess I should remove this now.)

[ReimarBauer]

I'm not sure if the additional prefix tut_ before picures is needed. If you have no other directory in mind then it could be just pictures, or images?

tutorials/tut pictures/zoom.PNG

[risehr]

please move tutorials/requirements.txt to requirements.d/tutorials.txt

This file should only have later the new packages or different versions you need to use not the full mamba list of packages which are already installed.

We have already the requirements.d to collect lists for different purposes

Okay. I will do it that way.

[risehr]

next time when you want to merge changes from the main repository please do this by a seperated merge request. e.g. name it "merge develop to my branch"
It is than easier for us to look at your changes ;)

So sorry for this. I actually had no idea this 'merge develop into GSOC21-HrithikKumarVerma' will be this big in changes. I will take care the next time.

[risehr]

I'm not sure if the additional prefix tut_ before picures is needed. If you have no other directory in mind then it could be just pictures, or images?

tutorials/tut pictures/zoom.PNG

Yeah, it sounds good. 'Pictures' or simply 'Images' should do the work. Also since the tut_pictures here can give the idea of pictures of the tutorials to someone even if its not so..! Hence, I also think this name is not appropriate.

[Marilyth]

I have a few comments regarding your FPS problem you mentioned last week.

[Marilyth]

Your loop in here does not contain everything you have in your capture loop.
The FPS you get here are always higher than your computer is able to produce, since the capture loop also writes to the mp4 file and is thus slower.

[Marilyth]

Regardless of your FPS calculation, the time to record and write and image fluctuates and creates inconsistent results.
To achieve somewhat consistent FPS during your recording, it would make sense to keep track of the captured frames vs expected frames.
You can then duplicate old frames or wait additional time to exactly match your desired FPS.
Here I have a suggestion on how to do that:

Suggested change

	def capture(self):
	"""
	Captures the frames of the screen at the rate of fps frames/second and writes into the
	video writer object with the defined fourcc, codec and colour format.
	"""
	with mss.mss() as sct:
	bbox = {"top": 0, "left": 0, "width": self.width, "height": self.height}
	self.start_rec_time = time.time()
	print(f"Starting to record with FPS value {self.fps} ...")
	while self.record:
	img = sct.grab(bbox)
	img_np = np.array(img)
	img_final = cv2.cvtColor(img_np, cv2.COLOR_RGBA2RGB)
	cv2.imshow(self.window_name, img_final)
	self.recorded_video.write(img_final)
	# Exits the screen capturing when user press 'q'
	if cv2.waitKey(1) & 0xFF == ord('q'):
	break
	def capture(self):
	"""
	Captures the frames of the screen at the rate of fps frames/second and writes into the
	video writer object with the defined fourcc, codec and colour format.
	"""
	with mss.mss() as sct:
	bbox = {"top": 0, "left": 0, "width": self.width, "height": self.height}
	self.start_rec_time = time.time()
	frame_time_ms = 1000 / self.fps
	frames = 0
	print(f"Starting to record with FPS value {self.fps} ...")
	while self.record:
	img = sct.grab(bbox)
	img_np = np.array(img)
	img_final = cv2.cvtColor(img_np, cv2.COLOR_RGBA2RGB)
	cv2.imshow(self.window_name, img_final)
	self.recorded_video.write(img_final)
	frames += 1
	expected_frames = ((time.time() - self.start_rec_time) * 1000) / frame_time_ms
	surplus = frames - expected_frames
	# Duplicate previous frame to meet FPS quota. Consider lowering the FPS instead!
	if int(surplus) <= -1:
	frames -= int(surplus)
	for i in range(int(-surplus)):
	self.recorded_video.write(img_final)
	# Exits the screen capturing when user press 'q'
	if cv2.waitKey(max(int(surplus * frame_time_ms), 1)) & 0xFF == ord('q'):
	break

[risehr]

Hey @Marilyth The code you suggested is much better than what I have tried myself. I have tried this thing before but not exactly like you. So, for much of the parts its working fine. The recorded video is not playing too fast as of now after implementing this code. Earlier when I was adding the repeated frames it was kind of slow and therefore, I discarded the idea. But the changes you have done in the waitkey() is nice and its recording and playing much better now. Thanks.

[ReimarBauer]

superseeded by PR #1067