Crate steel_math 

all the math of steel

Re-exports

pub use crate::angle::DEG_TO_RAD;

pub use crate::angle::DEG_TO_RAD_F64;

pub use crate::angle::DEGREE_90;

pub use crate::angle::DEGREE_180;

pub use crate::angle::DEGREE_270;

pub use crate::angle::DEGREE_360;

pub use crate::angle::RAD_TO_DEG;

pub use crate::angle::RAD_TO_DEG_F64;

pub use crate::angle::convert_to_rotation_segment;

pub use crate::angle::wrap_degrees;

Modules

angle

Shared angle constants. Common angle constants.

float_consts

IEEE 754 binary32 layout constants

noise_math 🔒

Math utilities used by vanilla world generation noise.

rounding 🔒

Java-compatible floating-point rounding

trig

Port of Mojang net.minecraft.util.Mth’s float trig lookup tables.

vector

Vanilla Vec3 axis rotations. Rotation helpers for DVec3 using Steel trig lookup table.

Constants

GRADIENT

Gradient vectors shared between Perlin and simplex noise (from vanilla SimplexNoise.GRADIENT).

GRADIENT_F32

See GRADIENT for details. This is a f32 version of the gradient table for use in f32 noise functions

Functions

bias_towards_extreme

Bias a noise value towards extremes (-1 or 1) using a sine curve.

clamp

Clamp a value to the range [min, max].

clamp_i32

Clamp a value to the range [min, max] (i32 version).

clamped_lerp

Clamped linear interpolation.

clamped_lerp_simd

Clamped lerp for N lanes.

corner_noise_3d

Compute corner noise contribution for a simplex vertex.

cube

Cube a value.

dot

Dot product of gradient vector and offset vector.

fast_floor

Floor function that matches Java behavior.

fast_floor_simd

SIMD implementation of fast_floor.

fast_lfloor

Long floor function matching Java behavior.

grad_dot

Calculate the f32 dot product used by vanilla Perlin noise.

grad_dot_simd

f32 N-lane gradient dot product.

inverse_lerp

Inverse linear interpolation (find the factor t such that lerp(t, a, b) == value).

lerp

Linear interpolation.

lerp2

Bilinear interpolation.

lerp3

Trilinear interpolation.

lerp2_simd

SIMD bilinear interpolation.

lerp3_simd

Trilinear interpolation for N lanes. see lerp3.

lerp_simd

SIMD linear interpolation.

map

Map a value from one range to another (unclamped).

map_clamped

Map a value from one range to another with clamped lerp.

round_f32

Rounds ties toward positive infinity, unlike f32::round, which rounds them away from zero. Uses the JDK’s IEEE 754 bit-shift implementation of Math.round(float), including saturation and NaN handling, without rounding up values just below a half-integer

smoothstep

Smoothstep - quintic Hermite interpolation (NOT cubic!)

smoothstep_derivative

Smoothstep derivative for noise with derivatives.

smoothstep_simd

Smoothstep for N lanes: 6x^5 - 15x^4 + 10x^3. Per-lane identical to smoothstep.

square

Square a value.

wrap

Wrap a coordinate to prevent precision loss at large values.

wrap_simd

Wrap N coordinates to prevent precision loss (N-lane SIMD version of wrap).