rand module

Hash

`pt_hash_uint`(x[, normalize])	1D XQO-style hash for PyTorch.
`pt_hash2_uint`(x, y[, normalize])	2D XQO-style hash for PyTorch.
`pt_hash2_uv`(uv[, seed, tile_size])	2D normalized UV hash for PyTorch.
`pt_hash2_xy`(xy[, seed, tile_size])	2D denormalized XY hash for PyTorch.
`np_hash_uint`(x[, normalize])	1D XQO-style hash for NumPy.
`np_hash2_uint`(x, y[, normalize])	2D XQO-style hash for NumPy.

Scramble

`pt_noise2`(p)	2D value noise scramble.
`pt_noise3`(p)	3D value noise scramble.

tinytex.rand.pt_hash_uint(x, normalize=False)[source]

1D XQO-style hash for PyTorch.

Parameters:

x (torch.Tensor) – Input int tensor.
normalize (bool) – If True, returns float32 values in [0, 1].

Returns:

Hashed tensor (uint32 or float32).

Return type:

torch.Tensor

tinytex.rand.pt_hash2_uint(x, y, normalize=False)[source]

2D XQO-style hash for PyTorch.

Parameters:

x (torch.Tensor) – X coordinates (int tensor).
y (torch.Tensor) – Y coordinates (int tensor).
normalize (bool) – If True, returns float32 in [0, 1].

Returns:

Hashed values (uint32 or float32).

Return type:

torch.Tensor

tinytex.rand.pt_hash2_uv(uv, seed=0, tile_size=1023)[source]

2D normalized UV hash for PyTorch. Expects normalized coordinates. Produces uniform pseudo-random values.

Parameters:

uv (torch.Tensor) – Tensor of shape [2, H, W] with values in [0, 1].
seed (int) – Optional seed for scrambling.
tile_size (int) – Maximum tile size (default: 1023).

Returns:

Tensor of shape [H, W] with float32 values in [0, 1].

Return type:

torch.Tensor

tinytex.rand.pt_hash2_xy(xy, seed=0, tile_size=1023)[source]

2D denormalized XY hash for PyTorch. Expects denormalized integer coordinates. Produces uniform pseudo-random values.

Parameters:

xy (torch.Tensor) – Tensor of shape [2, H, W] with integer coordinates.
seed (int) – Optional seed for scrambling.
tile_size (int) – Maximum tile size (default: 1023).

Returns:

Tensor of shape [H, W] with float32 values in [0, 1].

Return type:

torch.Tensor

tinytex.rand.np_hash_uint(x, normalize=False)[source]

1D XQO-style hash for NumPy.

Parameters:

x (ndarray) – Input array (uint32-compatible).
normalize (bool) – If True, returns float32 values in [0, 1].

Returns:

Hashed values (uint32 or float32).

Return type:

ndarray

tinytex.rand.np_hash2_uint(x, y, normalize=False)[source]

2D XQO-style hash for NumPy.

Parameters:

x (ndarray) – X array (uint32-compatible).
y (ndarray) – Y array (uint32-compatible).
normalize – If True, returns float32 values in [0, 1].

Returns:

Hashed values (uint32 or float32).

Return type:

ndarray

tinytex.rand.pt_noise2(p)[source]

2D value noise scramble.

Parameters:: p (torch.Tensor) – Input tensor of shape [2, H, W].
Returns:: Scrambled tensor of shape [2, H, W], float32 in [0, 1].
Return type:: torch.Tensor

tinytex.rand.pt_noise3(p)[source]

3D value noise scramble.

Parameters:: p (torch.Tensor) – Input tensor of shape [3, H, W].
Returns:: Scrambled tensor of shape [3, H, W], float32 in [0, 1].
Return type:: torch.Tensor