[MEncoder-users] hqdn3d and banding

[Grozdan]

2009/8/4 Andrew Berg <bahamutzero8825 at gmail.com>:
> Grozdan wrote:
>> You'll either need to up
>> bitrate or use a much better denoiser than hqdn3d which mplayer
>> currently lacks.
> Upping the bitrate won't help since the banding is in the video that is
> to be encoded.
>
>>  However, you can use the new dabanding filter during
>> playback. -vf gradfun. You can also add gradfun after hqdn3d (-vf
>> hqdn3d,gradfun)
>>
> It must be really new because it's not in the build I have (r29355,
> built June 6). I've tried building MEncoder myself, but I always have
> problems. I'll have to wait for a new build.
> I'll definitely check it out when I can, though. Thanks.

If the banding is already in the video before encoding then all you
can do is use gradfun (play with its options a bit) and hope it'll get
rid of it. Yes, gradfun is very new, it was ported over by Loren
Merritt (x264 author & maintainer) and was committed a few weeks ago
so your build from June won't have it. Also, like Reimar said, don't
do too much chroma filtering with the denoiser.

>
>> but I'm not sure how much it'll help since encoding
>> tends to bring back banding artifacts
> I try to make transparent encodes; hopefully the bits needed won't go up
> much.
>> Also keep in mind that a denoiser not only kills noise but also detail
>> since there's no perfect math algo that can "know" what's noise and
>> what not. A tiny bit of noise can be an advantage and can act as
>> dither to randomize quantization errors.
> I find it's almost always a challenge to get the right settings,
> especially when there's a lot of noise.
> _______________________________________________
> MEncoder-users mailing list
> MEncoder-users at mplayerhq.hu
> https://lists.mplayerhq.hu/mailman/listinfo/mencoder-users
>
>