TL,DR: This is a call for help to improve de-interlacing in VideoIO.
Thanks to @ianshmean, VideoIO has received a large number of significant improvements lately. These changes make VideoIO a viable tool for reading and writing videos. There is however a few remaining issues to deal with. One of them is interlacing. Here is a MWE:
import FileIO, VideoIO
file = download("https://s3.eu-central-1.amazonaws.com/vision-group-file-sharing/Data%20backup%20and%20storage/Yakir/interlacing.mkv")
f = VideoIO.openvideo(file)
img = VideoIO.read(f)
FileIO.save("interlacing.jpg", img)
While a closeup of the extracted image looks like this:
ffmpeg manages to remedy the artifact with its yadif filter:
ffmpeg -i interlacing.mkv -vframes 1 -vf yadif interlacing.jpg
Can someone please take a look at this ffmpeg filtering example and help us convert it to Julia, similarly to how VideoIO’s examples are? Alternatively, suggest a simpler way to implement filters?
The relevant VideoIO issue is here.
Thanks in advance!
As I mentioned on slack, I likely won’t have time to take this on. So I’ll just say that I learned from