Postby markfm » 12 Jul 2004 03:06
If you step back and think about it, I believe that you need a complete transcode (stating what compression you want used in the changed-resolution output) already, inherent in changing the resolution.
The way video CODECs work, generally, is by taking the base video format (e.g., 640x480), then performing mathematical transforms on the values in sub-arrays (e.g., 8x8, 16x16, or 64x64 blocks of pixels). They often do the compression in the frequency domain on these blocks, using various methods to achieve compression (throwing out higher frequency elements, or performing various predictions of how things will change in the blocks from frame-to-frame).
The compressed video is tightly coupled to the original base format -- the video is represented in terms of values put out for the NxN pixel chunks of data, only meaningful if dealing with the original frame size, not at the individual "raw" pixel level. If you want to change the video size, you need to bring the video back to baseband (decompress it), do what you want to alter the resolution (methods to change the scale by factor of 0.75, 0.5, whatever), then recompress it.
I'm not a VLC developer, could be totally wrong, but the above is generally what you do with standard fourier transform-based compression (such as MPEG and JPEG).