Output interpolated image frames are of different size then input frames?

I tried to generate slow motion videos without using ffmpeg, i.e using opencv-python by running this command: python eval.py data/input.mp4 --checkpoint=data/SuperSloMo.ckpt --output=data/output.mkv --scale=3 Here the input video had, say, 175 frames, then using the following command, I tried to get individual output frames of the output video: ffmpeg -i /path/to/output.mkv /path/to/out_frames/%06d.png .

But if original image frames have size of (175,315) then output of slomo which are interpolated frames all has size of (174,314). My doubt is exactly where the size of the image is reduced and how we can ensure we get correct interpolated frames with the same size? Can someone please mention in detail the steps I should follow for the same? This will be of great help!