I have a video with dimension (13, 222, 356, 3) and I want to convert this into a video file. I tried the below code snippet. I request u guys to please tell or suggest what can be done to make it work.
import numpy as np
import cv2
size = 720*16//9, 720
duration = 2
fps = 2
out = cv2.VideoWriter('output.mp4', cv2.VideoWriter_fourcc(*'mp4v'), fps, (size[1], size[0]), False)
for i in range(aj.shape[2]):
data = aj[:,:,i,:]
out.write(data)
out.release()
Your video dimension is (13, 222, 356, 3)?
I assume that the shape of the NumPy array that contains your video frames is (13, 222, 356, 3).
According to the common convention for storing video frames in NumPy array:
13 is the number of video frames.222 is the number of rows (frame height).356 is the number of columns (frame width).3 is the number of color channels (in OpenCV the channels ordering is BGR).This is the convention that makes sense (13 color channels doesn't make sense)
size is not 720*16//9, 720, but 356, 222.cv2.VideoWriter applies isColor=false, but your video has 3 color channels, so don't set it to false - keep the default value - true.i in range(aj.shape[2]) is wrong - use i in range(aj.shape[0]) for iterating your 13 frames.data = aj[:,:,i,:] should be data = aj[i,:,:,:].Here is a code sample (the sample builds synthetic frames for testing):
import numpy as np
import cv2
# Create aj for testing, dimension (13, 222, 356, 3)
aj = np.zeros((13, 222, 356, 3), np.uint8)
for i in range(aj.shape[0]):
cv2.putText(aj[i, :, :], str(i+1), (aj.shape[2]//2-50*len(str(i+1)), aj.shape[1]//2+50), cv2.FONT_HERSHEY_DUPLEX, 5, (255, 30, 30), 10) # Blue number
# size = 720 * 16 // 9, 720
# duration = 2
fps = 2
out = cv2.VideoWriter('output.mp4', cv2.VideoWriter_fourcc(*'mp4v'), fps, (aj.shape[2], aj.shape[1]))
for i in range(aj.shape[0]):
data = aj[i, :, :, :]
out.write(data)
out.release()
Note:
Color channels ordering of OpenCV is BGR (blue is first).
Many Python packages uses RGB convention (red is first).
In case you see that red and blue are swapped, use RGB to BGR conversion:
data = cv2.cvtColor(data, cv2.COLOR_RGB2BGR)
