[MEncoder-users] Noise reduction and interlacing: (hq)denoise3d vs. nr

[Pierre Catello]

Hi Levente,

I'm not an expert but I think that denoise3d hqdn3d are likely
dangerous for interlaced content as they work both temporally and
spatially. The spatial filter will be harmful unless you separate the
fields before applying the filter and then weave them to
reinterlace... Something like :

-vf tfield=0,hqdn3d,tinterlace

or slower but better: -vf yadif=1,mcdeint,hqdn3d,tinterlace

Regarding the internal NR filter from Lavc, I would also be happy to
hear from the experts here in which way it is better. From what I have
read, it is supposed to preserve fine details and avoid "ghosting"
effects that can often plague temporal filters (when set too strong).

>From my understanding, a noise reduction filter working internally in
the codec have the opportunity use the motion vectors computed by the
codec and thus doing motion compensated noise filtering, but here I'm
speculating and I haven't looked at the source of the nr filter from
neither lavc nor x264.

Can anyone confirm or infirm this ?


2007/6/20, Levente Novák <lnovak at dragon.unideb.hu>:
> What are the differences between denoise3d (or hqdn3d) and the nr lavc
> denoise filter? Beside that the latter is faster, are both equally
> appropriate for denoising interlaced content? I am asking this as it is
> perfectly possible to put hqdn3d between il=d:d and il=i:i (so this is
> interlaced-safe), but nr is a lavc option and it is not clear to me
> whether it does or does not harm interlaced content.
>
> Levente
>
>
> _______________________________________________
> MEncoder-users mailing list
> MEncoder-users at mplayerhq.hu
> http://lists.mplayerhq.hu/mailman/listinfo/mencoder-users
>