Originally Posted by Light Illusion
Again, it is a difference between complex calibration requirements, and displays that are near perfectly linear, where actually a simple 3x matrix would be the best solution, rather than a 3D LUT.
LS is aimed at displays where simple matrix calibrations would not work - which is the majority of poor consumer display as it happens.
To me, this doesn't explain the fact that the Argyll algorithm produces superior results to LS, on the same display
, with a limited number of patches. Stress tests are a good indication of performance, and using a limited number of patches seems to be a good stress test.
All the data posted in this thread shows that LS starts to produce acceptable results only when the number of patches is increased to a high number. In the limit, if you were to use millions of patches, any piece of software could produce amazing results. It's what you can do with a limited number of patches that counts, as this really reveals how efficient and intelligent the algorithms are.