In 99% of the cases it’s not possible to derive height information from a regular picture. Even though they look like grayscale pictures, heightmaps are just a completely different concept where each pixel encodes a height (the lighter, the higher), rather than a reflected light intensity (as in a regular picture)
In a very limited number of cases you may get lucky and be able to create an “interesting” heightmap from a bitmap picture, but mostly you will be disappointed.
2.5D is doable manually, if you vectorize the input image and then create toolpaths of different depths for different parts of the image, but that is quite tedious and only practical for very simple pictures. I did something like this in this example.
For real 3D, your best bet is to search for a 3D model (STL) and then convert that to a heightmap using @fenrus tools, and import than in CC Pro to create 3D toolpaths.