How to convert an array of JMh or Jab values to XYZ?

[abnormally-distributed]

I have arrays of JMh & corresponding Jab values that I want to convert to XYZ.

What I am doing is going from an sRGB cube LUT --- > Linear RGB --- > [Insert a colorspace here, in this case, CIECAM16/02] ---> Adjust some values to make manual LUT adjustments by hand (apply curves to specific combinations of J, M, h, etc) -- > XYZ --> Linear RGB --> sRGB.

In the case of working with something like Lab it's simple to do those steps to convert RGB->Lab->Lch->Lab->XYZ->RGB or whatever, but the only XYZ functions related to CIECAM seem to be only capable of taking a specification object and spitting back out the xyz values used to create it.

I am using Python via the Reticulate package in RStudio, so I don't know if there's some odd behavior masking a function capable of doing this or what but from what I can tell it seems like there's no Jab02 or Jab16 conversion to XYZ

[KelSolaar]

Hi @abnormally-distributed!

I'm trying to understand the issue here, is the problem that R does not let you create the CAM specifications?

We are using some convenience code to do that in the Automatic Colour Conversion Graph:

def CIECAM02_to_JMh_CIECAM02(CAM_Specification_CIECAM02):
    """
    Converts from *CIECAM02* specification to *CIECAM02* :math:`JMh`
    correlates.

    Parameters
    ----------
    CAM_Specification_CIECAM02 : CAM_Specification_CIECAM02
        *CIECAM02* colour appearance model specification.

    Returns
    -------
    ndarray
        *CIECAM02* :math:`JMh` correlates.

    Examples
    --------
    >>> specification = CAM_Specification_CIECAM02(J=41.731091132513917,
    ...                                        M=0.108842175669226,
    ...                                        h=219.048432658311780)
    >>> CIECAM02_to_JMh_CIECAM02(specification)  # doctest: +ELLIPSIS
    array([  4.1731091...e+01,   1.0884217...e-01,   2.1904843...e+02])
    """

    return tstack([
        CAM_Specification_CIECAM02.J,
        CAM_Specification_CIECAM02.M,
        CAM_Specification_CIECAM02.h,
    ])


def JMh_CIECAM02_to_CIECAM02(JMh):
    """
    Converts from *CIECAM02* :math:`JMh` correlates to *CIECAM02*
    specification.

    Parameters
    ----------
    JMh : array_like
         *CIECAM02* :math:`JMh` correlates.

    Returns
    -------
    CAM_Specification_CIECAM02
        *CIECAM02* colour appearance model specification.

    Examples
    --------
    >>> import numpy as np
    >>> JMh = np.array([4.17310911e+01, 1.08842176e-01, 2.19048433e+02])
    >>> JMh_CIECAM02_to_CIECAM02(JMh)  # doctest: +ELLIPSIS
    CAM_Specification_CIECAM02(J=41.7310911..., C=None, h=219.0484329..., \
s=None, Q=None, M=0.1088421..., H=None, HC=None)
    """

    J, M, h = tsplit(JMh)

    return CAM_Specification_CIECAM02(J=J, M=M, h=h)


def CAM16_to_JMh_CAM16(CAM_Specification_CAM16):
    """
    Converts from *CAM16* specification to *CAM16* :math:`JMh` correlates.

    Parameters
    ----------
    CAM_Specification_CAM16 : CAM_Specification_CAM16
        *CAM16* colour appearance model specification.

    Returns
    -------
    ndarray
        *CAM16* :math:`JMh` correlates.

    Examples
    --------
    >>> specification = CAM_Specification_CAM16(J=41.731207905126638,
    ...                                     M=0.107436772335905,
    ...                                     h=217.067959767393010)
    >>> CAM16_to_JMh_CAM16(specification)  # doctest: +ELLIPSIS
    array([  4.1731207...e+01,   1.0743677...e-01,   2.1706796...e+02])
    """

    return tstack([
        CAM_Specification_CAM16.J,
        CAM_Specification_CAM16.M,
        CAM_Specification_CAM16.h,
    ])


def JMh_CAM16_to_CAM16(JMh):
    """
    Converts from *CAM6* :math:`JMh` correlates to *CAM6* specification.

    Parameters
    ----------
    JMh : array_like
         *CAM6* :math:`JMh` correlates.

    Returns
    -------
    CAM6_Specification
        *CAM6* colour appearance model specification.

    Examples
    --------
    >>> import numpy as np
    >>> JMh = np.array([4.17312079e+01, 1.07436772e-01, 2.17067960e+02])
    >>> JMh_CAM16_to_CAM16(JMh)  # doctest: +ELLIPSIS
    CAM_Specification_CAM16(J=41.7312079..., C=None, h=217.06796..., s=None, \
Q=None, M=0.1074367..., H=None, HC=None)
    """

    J, M, h = tsplit(JMh)

    return CAM_Specification_CAM16(J=J, M=M, h=h)

Would that help?

[abnormally-distributed]

What I mean is, you know how the Lab_to_XYZ function takes an array of CIELab values and returns XYZ, and you can feed any Lab values you want to it? The CAM16_to_XYZ only takes the specification and returns the XYZ values originally used to create it. However, I am applying curves and the like to the JMh or Jab (which I am working with depends on what I am doing) as part of creating some custom color LUTs and am unable to do convert my modified values back to XYZ.

Likewise, I also discovered that RLab lacks an RLab_to_XYZ function entirely. Is there a way to do that conversion? Figured I'd mention it here instead of opening up another thread.

[KelSolaar]

The last definition in the code I quoted above should do the conversion, then you can pass the specification to colour.CAM16_to_XYZ:

def JMh_CAM16_to_CAM16(JMh):
    J, M, h = tsplit(JMh)

    return CAM_Specification_CAM16(J=J, M=M, h=h)

As for RLab_to_XYZ, yes this is known issue! It is on @MichaelMauderer's plate, some day!