[x264-devel] [PATCH]adaptive lowpass

[Gabriel Bouvigne]

Eric Viscito a écrit :

> I think your audio/video analogy is reasonable.  But isn't that audio
> lowpass filter applied to the raw samples themselves prior to any
> compression step, and not to some residual of a prediction step?  I think
> you would get more perceptual improvement by applying a filter before
> prediction and driven by analysis with a perceptual model.

There is no prediction step in audio, but I agree about the theorical 
benefits of running a perceptual model first.


> Would you vary the lowpass filter in audio on a very local time scale?  Or
> is it more that you filter certain signal types whenever they appear
> because either the information is irrelevant or cannot be adequately
> handled by the compressor?

No, in audio we are trying to keep the lowpass filter constant (in 
current Lame version, it's mostly a static lowpass according to target 
bitrate), as frequent changes of audio bandwidth is an annoying thing.
However, as already pointed, there is no prediction within audio.

What is happening in video is a bit different, if you consider this patch.
At first, your data is originally encoded as intra, with a lower Qp and 
an higher (ie less strong) lowpass (lowpass strength is lower for intra).
Then you encode the inter content, which is predicted from previous 
intra blocks. What your encoder will try encoding into the high freqs 
part is the "shape" changes, and it will also try to compensate against 
previous quantization noise. If you cut the HF part, you will mostly 
have in HF the inherited quantization noise from previous frames. Thus 
it's likely to not be the best HF content, but you still have HF 
content, which means that you should not see a sudden change of 
frequency coverage.

--
Gabriel
_______________________________________________
x264-devel mailing list
x264-devel at videolan.org
http://mailman.videolan.org/listinfo/x264-devel