VideoFrame.to_image() doesn't work properly if width/height is specified

Created by: animetosho

Overview

VideoFrame.to_image() permits width and height parameters, according to documentation, however, if specified, they cause issues.

Expected behavior

Either width/height is not supported in to_image, or it resizes the frame to specified dimensions.

Actual behavior

If the width/height is smaller than original dimensions, a runtime error (ValueError: not enough image data) is thrown. If it's larger, it resizes, but doesn't change the image's dimensions, giving messed up output.

Investigation

The to_image function uses the frame's original dimensions when constructing the image, as opposed to the after-reformatted dimensions.

(it may also be worth checking the other functions which pass the parameters onto reformat for similar issues)

Versions

• PyAV runtime: 8.1.0