Struct Nl4dOptions 

pub struct Nl4dOptions {
    pub motion: MotionSearch,
    pub sigma: Option<f32>,
    pub sigma_scale: f32,
    pub thsad_scale: f32,
    pub refine: u32,
    pub spatial_radius: u32,
    pub lambda_ht: Option<f32>,
    pub lambda_ht_scale: f32,
    pub c_min: f32,
    pub mismatch_scale: f32,
    pub confidence_variance: bool,
    pub windowed_noise_estimation: bool,
}

Settings for Algorithm::Nl4d.

nl4d runs the HQ front end only for its machinery, the frame ring, the motion field, and the noise estimate. Nothing weights or averages patches the NLM way, so the NLM knobs are absent here and the fields below are the whole surface.

The temporal radius comes from DenoiserOptions::mode, which has to be Temporal { .. }. Motion tracking is always on, because the grouping kernel reads the motion field and confidence scores it produces.

lambda_ht has a per-plane default. None resolves through nl4d_default_lambda_ht once the plane being denoised is known. lambda_ht_scale then multiplies whichever value that resolves to.

Every other default comes from [Nl4dParams::default].

Fields

motion: MotionSearch

How motion between frames is tracked.

sigma: Option<f32>

A fixed noise standard deviation in [0, 1] units, replacing the automatic per-frame estimate.

None, the default, measures the noise in each pushed frame and smooths it over time.

sigma_scale: f32

A multiplier applied to the measured noise level before anything reads it. Defaults to 1.0.

This does nothing when sigma pins the noise level, because the estimator never runs in that case.

thsad_scale: f32

A multiplier on the per-block mismatch threshold, which sets how much extra SAD a block tolerates before its confidence starts to fall. Defaults to 1.0.

Higher values tolerate larger mismatches.

refine: u32

Half-width of the refine window searched around each neighbour frame’s motion-predicted position, in 1..=4. Defaults to 2.

spatial_radius: u32

Half-width of the spatial candidate window searched in the centre frame, in 1..=16. Defaults to 9.

lambda_ht: Option<f32>

Hard-threshold multiplier on the propagated coefficient sigma. Higher removes more noise and more fine detail.

None resolves through nl4d_default_lambda_ht, which returns a different value for luma than for chroma.

lambda_ht_scale: f32

A multiplier applied to the resolved lambda_ht. Defaults to 1.0.

It scales an explicit lambda_ht and the calibrated per-plane default alike, so one value moves both planes together. Has to be finite and in [0.1, 10.0].

c_min: f32

The confidence floor below which a whole neighbour block is skipped rather than scored, in [0, 1). Defaults to 0.05. Only affects how much compute a submit spends, never which candidates are admitted once they are scored.

mismatch_scale: f32

A multiplier on the mismatch variance a poorly matched temporal member carries into the hard threshold. Defaults to 1.0.

The variance grows with the square of this. The mechanism saturates well before the top of its accepted range, see [crate::nl4d::Nl4dParams::mismatch_scale].

confidence_variance: bool

Whether a temporal member’s mismatch variance reaches the hard-threshold shrinkage at all. Defaults to true. See [crate::nl4d::Nl4dParams::confidence_variance].

windowed_noise_estimation: bool

Estimates noise fresh from each frame’s own window instead of smoothing it across the whole stream’s history. Defaults to false, matching every calibrated preset.

av-denoise-vs turns this on unconditionally, because a VapourSynth filter has to return the same pixels for a frame no matter what order frames were requested in, and history-dependent estimation breaks that guarantee under random access. See HqParams::windowed_noise_estimation.

Trait Implementations

impl Clone for Nl4dOptions

fn clone(&self) -> Nl4dOptions

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for Nl4dOptions

impl Debug for Nl4dOptions

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl Default for Nl4dOptions

fn default() -> Nl4dOptions

Returns the “default value” for a type.

impl PartialEq for Nl4dOptions

fn eq(&self, other: &Nl4dOptions) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for Nl4dOptions