On Thu, 6 Apr 2006, Christian Ebert wrote:

> What about -H|--keep-hf ? Supposing I just want highest quality?

        I think it would be a good idea to define "highest quality" ;)  The
        same type of request/desire comes up a lot on a couple forums I lurk
        in.  It's often in the form of a statment that implies the existence
        of a magic option or tool that will give a perfect (restored/enhanced
        to be what the user *wants*) image even though the data's been
        compressed (lossily) between somewhere between 15 and 40 to 1

        Uncompressed 10bit 4:2:2 is the highest quality - but it's about
        200 megabits/second.   Next highest would be downsampled to 8bit
        4:2:0 at "only" 124 megabits/second.    That is as high as the
        quality gets.

        Oh, you want MPEG-2 :) :)   

        -H can actually lower the quality.  Remember - there is a hard limit 
        on the bitrate you can use and -H increases (often dramatically) the 
        number of bits used to encode an image.    IF the number of bits/sec 
        available is not sufficient then the encoder has NO choice but to 
        lower the quality (by increasing the amount of lossy compression 
        performed).    Using -H has, the times I've used it, raised the 
        average bitrate about 30%.  IF you're already at the maximum for
        a project (2hr movie on a DVD is limited to ~4.7Mb/s) then increasing
        the number of bits required will cause the quality to be dropped (by
        higher lossy compression) to make those bits available.

        If you're using double layer media and have sufficient space then -H
        might be usable but many projects are bitrate/size constrained and
        are already at those limits - no leeway or headroom left.

        Encoders in general have developed to the point where there aren't
        too many obviously "bad ones" around.    So that leads to the next
        paragraph...

        This may come as a surprise but the key to "quality"  lies LESS with
        the 'encoding' than it does with the pre-processing of the data _before_
        going to the encoder.  It's amazingly labor intense - going scene
        by scene to apply different filters depending on the content of the
        scene BUT it gives "higher quality" because the encoder is not spending
        bits on details/areas that the eye/brain will not see (or can be
        distracted away from).

        Compression is an art form - there is no magic "use this <feature> for 
        high quality".  

        Cheers,
        Steven Schultz



-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
_______________________________________________
Mjpeg-users mailing list
Mjpeg-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mjpeg-users

Reply via email to