cutcutcodec.core.io.framecaster

This module, implemented in C, offers functions to normalize frames.

Functions

from_rgb	Convert a frame from the RGB space to float32 frame in the RGB space.
from_yuv	Convert a frame from the YUV space to float32 frame in the Y'PbPr space.
to_rgb	Convert a frame from the RGB float32 space to uint8 frame in rgb24.
to_yuv	Convert a frame from the Y'PbPr float32 space to uint16 frame in the YUV tv space.

Details

cutcutcodec.core.io.framecaster.from_rgb()

Convert a frame from the RGB space to float32 frame in the RGB space.

• add 1 leading channel to grayscale frame (h x w -> h x w x 1)

• cast into float32

• convert limited range to full range (based on UIT-R)

Parameters

framenp.ndarray

The float32, uint8 or uint16 video frame of shape (height, width, [channels]).

is_tvbool

If True, consider the input as a limited range coding.

threadsint, default=0

Number of threads used.

Returns

normalizednp.ndarray[np.float32, np.float32, np.float32]

The normalized frame with [r, g, b] in [0, 1]**3.

Notes

• It is optimized for C contiguous array.

• This function makes a safe copy of the input frame, no modification inplace.

cutcutcodec.core.io.framecaster.from_yuv()

Convert a frame from the YUV space to float32 frame in the Y’PbPr space.

• add 1 leading channel to grayscale frame (h x w -> h x w x 1)

• cast into float32

• convert limited range to full range (based on UIT-R)

Parameters

framenp.ndarray

The float32, uint8 or uint16 video frame of shape (height, width, [channels]), with optional channel dimension in {1, 3, 4}.

is_tvbool

If True, consider the input as a limited range coding.

threadsint, default=0

Number of threads used.

Returns

normalizednp.ndarray[np.float32, np.float32, np.float32]

The normalized frame with [y, u, v] in [0, 1] x [-1/2, 1/2]**2.

Notes

• It is optimized for C contiguous array.

• This function makes a safe copy of the input frame, no modification inplace.

cutcutcodec.core.io.framecaster.to_rgb()

Convert a frame from the RGB float32 space to uint8 frame in rgb24.

Parameters

framenp.ndarray[np.float32, np.float32, np.float32]

The float32 video frame of shape (height, width, channels).

to_tvbool, default=False

If True, convert the frame in limited range, default is full range.

threadsint, default=0

Number of threads used.

Returns

outnp.ndarray[np.uint8]

The YUV frame in limited range.

Notes

• It is optimized for C contiguous array.

• This function makes a safe copy of the input frame, no modification inplace.

cutcutcodec.core.io.framecaster.to_yuv()

Convert a frame from the Y’PbPr float32 space to uint16 frame in the YUV tv space.

Parameters

framenp.ndarray[np.float32, np.float32, np.float32]

The float32 video frame of shape (height, width, channels).

to_tvbool, default=True

If False, convert the frame in full range, default is limited range.

threadsint, default=0

Number of threads used.

Returns

outnp.ndarray[np.uint16]

The YUV frame in limited range.

Notes

• It is optimized for C contiguous array.

• This function makes a safe copy of the input frame, no modification inplace.