Stefan de Konink wrote: > I wonder if a combined range coder for multiple streams could give a > better compression than one for each stream individually. The range coder is not using an adaptive probability model, so no, this is not likely to help.