About Denoisers

[bxyhxyh]

I have 3 questions.

1. What is the advantages of temporal, spatial and spatio-temporal denoisers?

2. What is the side effects of temporal, spatial and spatio-temporal denoisers?

Temporal - Ghosting and ...?
Spatial - Banding and ...?
Spatio-Temporal - ...?

3. Good Temporal, Spatial and spatio-temporal denoisers?

[x265]

Spatial Noise Reduction
A spatial noise reduction filter uses a more complex algorithm than blur/low-pass, attempting
to find where on a frame noise is and isn’t, and only trying to remove the noise. In general,
spatial noise reduction will try to preserve sharp edges, and remove or reduce noise from
flatter areas with a lot of seemingly random fine detail, like grain/gain. The challenge is they
can’t always discriminate between static textures, like cloth, and actual noise.

Temporal Noise Reduction
For real grain/gain removal, you need a temporal filter. That compares a pixel with its
spatial neighbors in the current frame and its temporal neighbors in adjacent frames. Since
grain/gain is random on every frame, the filter can pick out flecks of detail that change from
frame-to-frame (noise) and those that don’t (detail), and thus be much more aggressive about
eliminating the former. And since that random noise is the biggest compression killer (fine
detail without any motion vectors to take advantage of), this can pay off very nicely.

[feisty2]

I think temporal denoisers are usually better than spatial denoisers for most sources
my personally favorite temporal denoiser is MDegrain (in MVTools2)
because unlike lots of other denoisers,it makes no artifacts even if use a very strong parameter (unless you use a crazily high thSAD,then there'll be ghosting) and if you use strong parameters in other denoisers on some very dirty sources, your clips will usually become like oil painted which is very bad
but MDegrain doesn't have this kind of issues

[x265]

MDegrain is and excellant filter, but it oversmoothes detail's in dark areas.

[feisty2]

MDegrain is and excellant filter, but it oversmoothes detail's in dark areas.

then you can create a gradient mask to protect dark areas ,and all the denoisers will kill some details it's a common denosing problem, not a particular mdegrain problem

[x265]

Masktools2 documentation is confusing. I did not understand anything by reading it.

[feisty2]

something like this maybe help

a=last
g_mask=mt_lut("x 128 >= 255 x 16 <= 0 x 16 - 255 * 112 / ? ?")
denoise=smdegrain ()
mt_merge(a,denoise,g_mask)

the effect is do denoise unconditionally if Y value > 128
gradually decrease the strength of denosing if Y value < 128
and do nothing if Y value < 16

[x265]

Is there a good guide, one which does not require any prerequisites for learning this filter(masktools2)?

[feisty2]

there's one :) but not in english