[MEncoder-users] Happy with encoding results but wish better understanding

Grozdan neutrino8 at gmail.com
Wed Apr 21 15:11:14 CEST 2010


On Wed, Apr 21, 2010 at 3:09 PM, Grozdan <neutrino8 at gmail.com> wrote:
> On Wed, Apr 21, 2010 at 2:49 PM, Sue Morton <zytb8phzgv at snkmail.com> wrote:
>>
>> Hello RC,
>>
>>> vhq is mbd=1, so a conflict with mbd=2 in there.
>>>
>>> You would get the most help from adding "trell".  You might be able
>>> to squeeze some more quality out of it using vlelim=-4:vcelim=7,
>>> and perhaps nr=.
>>>
>>> I would think you'd be happy with yadif=0 and drop mcdeint,framestep=2,
>>> but if performance is no issue, no reason not to keep it.
>>>
>>> Additionally, you'll get a big quality improvement/size reduction if you
>>> crop the black borders and noise around the edges, but then it will
>>> display at least a bit differently on your player, so be a bit careful.
>>> If you go that way, make sure you use final dimentions that are
>>> multiples of 16 for progressive, and 32 for the height when the video
>>> will be interlaced.
>>
>> Thank you for the feedback, very much appreciated!  I will look into your
>> suggestions, in detail, as soon as I have time (probably not until the
>> weekend).
>>
>> I do already have 'trell' specified, it's right after the mbd=2 parameter --
>> so I've got your suggestion in place already :-)
>>
>> What does 'trell' do vs. leaving it out, such that it 'helps'?  And what is
>> the 'help' it gives (less motion artifacts in video, helps sync video with
>> audio, etc.)?  I've done a lot of reading, and I'll go back to read the man
>> pages again, but I'm not really understanding what these parameters 'do' --
>> a pretty learning curve is still ahead of me I think :-)
>
> trell is short for Trellis Quantization which helps both increase
> quality and save bits
> http://en.wikipedia.org/wiki/Trellis_quantization
>
> Also, you may want to use cmp=3 and subcmp=3 which will select a
> different and better (SATD Hadamard) transform comparison function for
> full and subpixels, further increasing quality.
> http://en.wikipedia.org/wiki/SATD

Small correction, SATD is 2, not 3 which is DCT. So it'll be cmp=2:subcmp=2

>
>>
>> I'll post back when I've had a chance to look into your suggestions.  In the
>> meantime, if you (or anyone) could explain a bit what any of these
>> parameters are supposed to do, it would be of help to me.
>>
>> Thank you again.
>>
>> Sue
>> _______________________________________________
>> MEncoder-users mailing list
>> MEncoder-users at mplayerhq.hu
>> https://lists.mplayerhq.hu/mailman/listinfo/mencoder-users
>>
>


More information about the MEncoder-users mailing list