Dancing noise

[senseiam]

Samples:
sample1 (http://rapidshare.com/files/280702183/hellr-005.mkv.html)
sample2 (http://www.megaupload.com/?d=QQZ94SJ8)

Would someone mind helping me devise a script that could minimize the dancing noise that is present throughout?

Since I am very new at this all I have got is:
MCTemporalDenoise(settings="low",chroma=true, stabilize=true, sharp=false)

I know there is no magic bullet to denoising while keeping all of the details but I am just looking for a middle ground.

I would like to preserve as much detail as possible while at the same time getting rid of that horrible snow/noise that is present in the above samples.

Any help would be tremendously appreciated. :)

[LaTo]

I would like to preserve as much detail as possible while at the same time getting rid of that horrible snow/noise that is present in the above samples.

One of the most important thing with MCTemporalDenoise is to tweak the "sigma" parameter.

For a video, too high sigma will destroy much detail and too low sigma will do a bad denoising.

You should set the sigma value at the better compromise between denoising and preserving detail.

MCTemporalDenoise(settings="low",sigma=XX,chroma=true, stabilize=true, sharp=false)

High sigma value for heavy noise.
Low sigma value for low noise.
Default for settings="low" is 4.

PS: You can also use the "post" parameter to tweak your result

[senseiam]

One of the most important thing with MCTemporalDenoise is to tweak the "sigma" parameter.

For a video, too high sigma will destroy much detail and too low sigma will do a bad denoising.

You should set the sigma value at the better compromise between denoising and preserving detail.



High sigma value for heavy noise.
Low sigma value for low noise.
Default for settings="low" is 4.

PS: You can also use the "post" parameter to tweak your result

Ok, thanks for the advice, going to try it out and see what I can come up with. :)

[BigDid]

Hi,

Sample1 being around crf-21, I suppose it is an encoded clip not a source-clip. So any tips relative to this sample will have to be confirmed with a source-clip.

Just a remark that the beginning of sample1 could be well denoised with "low" settings of MCTemporalDenoise (I have used sigma=3 from fft3dgpu to get rid of it, quick and dirty), but the noise at the end of the clip is really stronger; you may need to use "medium" or "high" settings of MCTemporalDenoise to get rid of it (I had to rise fft3dgpu sigma to 6 to remove it).

As indicated in the filter thread, higher settings means more details preserved but long to very long processing time.

Did

PS: even if same parameter-name, "sigma" strength in fft3dgpu and MCTemporalDenoise are different, so will be the sigma numbers used for each respective filter. The comparison with fft3dgpu is just to highlight the gap of denoising strength needed between the beginning and the end of sample1-clip.

[senseiam]

Hi,

Sample1 being around crf-21, I suppose it is an encoded clip not a source-clip. So any tips relative to this sample will have to be confirmed with a source-clip.

Just a remark that the beginning of sample1 could be well denoised with "low" settings of MCTemporalDenoise (I have used sigma=3 from fft3dgpu to get rid of it, quick and dirty), but the noise at the end of the clip is really stronger; you may need to use "medium" or "high" settings of MCTemporalDenoise to get rid of it (I had to rise fft3dgpu sigma to 6 to remove it).

As indicated in the filter thread, higher settings means more details preserved but long to very long processing time.

Did

PS: even if same parameter-name, "sigma" strength in fft3dgpu and MCTemporalDenoise are different, so will be the sigma numbers used for each respective filter. The comparison with fft3dgpu is just to highlight the gap of denoising strength needed between the beginning and the end of sample1-clip.

Just to clarify, both samples are from the source, all I did was extract the .h264 from the .m2ts, followed that up by putting the .h264 into an mkv for easy splitting. No encoding done. Just a poor transfer to Blu ray.

Thanks for the advice, very much appreciated. I haven't really been able to test anything yet thought as I have been trying to finish another encode.

:thanks: