I have the following function, that creates a large number of dict entries that describe cubes:

def fill_dict_cubes(
    length: np.ndarray,
    width: np.ndarray,
    height: np.ndarray,
) -> np.ndarray:
    numObjects, numFrames = length.shape
    cubePrimitives = np.empty((numObjects, numCycles), dtype=object)
    for objIdx in range(numObjects):
        for frameIdx in range(numFrames):
            cubePrimitives[objIdx, frameIdx] = CubePrimitive(
                length=length[objIdx, frameIdx],
                width=width[objIdx, frameIdx],
                height=height[objIdx, frameIdx],
            ).attributes

    return cubePrimitives

I need to use the dict structure, it is predefined by an external API. The input data length, width and height are 2D numpy arrays. The output is an array that contains the dicts.
Since I have a large number of cubes and also many frames, filling all these dicts using nested for-loops takes quite some time. Unfortunately I could not find so far a nice way to make this faster by vectorizing / multiprocessing / parallelization / etc.

Has anybody a clever idea how the creating of these dicts could be made faster? Btw, I am using python 3.10, but updating to 3.13 would be no issue, if new features are required.