You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have some input data that depends on how the user sizes my GUI window so I can't guarantee things required by ffmpeg like frame sizes divisible by 2. To get around that I add a filter used by ffmpeg to scale the frame to the nearest 16 pixels. This is similar to what imageio-ffmpeg does to make frames consistent with the "macro block size".
When a filter is provided it seems that imageio/pyav always produces a video of 480x640 pixels instead of a size based on the input frames. From what I can tell the issue in the pyav plugin is that it provides the filters the video stream before it is full initialized during the frame writing. Is there a way around this?
But only in cases where my frame was already divisible by 2 (and possibly only when the filter didn't change the frame size). I think it is because the frame influences the stream size (width/height) but the stream influences how the filter behaves (chicken/egg problem).
I have some input data that depends on how the user sizes my GUI window so I can't guarantee things required by ffmpeg like frame sizes divisible by 2. To get around that I add a filter used by ffmpeg to scale the frame to the nearest 16 pixels. This is similar to what
imageio-ffmpeg
does to make frames consistent with the "macro block size".When a filter is provided it seems that imageio/pyav always produces a video of 480x640 pixels instead of a size based on the input frames. From what I can tell the issue in the pyav plugin is that it provides the filters the video stream before it is full initialized during the frame writing. Is there a way around this?
The text was updated successfully, but these errors were encountered: