Do you do any intermediate encoding and/or resolution changes? What's the resolution and codec of the original? And into what codec is the final file encoded, and what's the relative size of the final file compared to the original?
Generally speaking, it's best to preserve the dimensions during editing, and do any intermediate encodes using only lossless codecs (meaning there is no loss of image quality). Huffyuv (
http://neuron2.net/www.math.berkeley...g/huffyuv.html) is very fast and well-suited for that task. Finally at the very last step, you can resize (if you want) and encode lossily for maximum compression (though you'll always have to make a decision where to cut off the compression-to-quality balance: it's a tradeoff).