So I have been trying profile prism for a while now and my results are ok, but i think there is a way to vastly improve the results with a simple idea.
The idea would be to iteratively improve the profile generation. This would be achieved by adding in a field "source ICC profile"
the steps would be
0. Set the ICC profile to "raw"
1. Print the test target using current profile (initially raw)
2. Scan target
3. generate new profile based on scan AND existing profile
4. goto 1