Levelings are morphological filters that are used to simplify and segment images. Depending on the scale, different amounts of detail disappear while the contours of the remaining objects remain sharp. In this work, we will consider two colour morphological methods to perform the dilation and erosion operations necessary for leveling. The first one computes a direct supremum based on the spectral decomposition of symmetric matrices, which offers some favourable properties, such as transitivity. Based on this, the second method generates a colour ordering of the input colours. We compare the levelings resulting from these two methods experimentally using natural colour images.