[MPlayer-users] on trying out codec options

[Yang Yang]

after frantically trying  to find a good set of options for lavc to
encode my video, I finally did it with the help of many guys here.

but having got the results, I still dont' know how these options work,
and I believe I wont' unless I devote much effort to understanding the
algorithm, and the tricky relations between different parts of the
algorithm, ----- anyway, it would be a great feat to figure out some
___analytical__ rules  for choosing options for different
sources/rerquired quality/speed, and I'm sure that's worth a place in an
academic journal. 

that being out of my reach, I think the only choice for a scientific
method to solve the "encoder option problem" is to exhaustively search
the option space, like many people have done before. previous work by
rguyom still left many options of lavc and xvid, and there is not a
systematic __framework__, or standard benchmark/database, I guess people
here could do it. I'm going to do it for my own purpose at least .


big problem is PSNR, besides that, is there any other metric which is
amenable to automatic data-processing? PSNR is probably not in agreement
with human subjective perception in many cases ( for example, I would
image the case of little flowers on a field ) 


Y