I am majoring in Computer Science and me and my friend are doing our thesis on Color Theory (in particular, automatic color scheme generation).

We have created some models that create palettes from given colors based on color wheel combinations. Another model is based on data taken from Adobe Kuler color palettes and ratings, where the scheme is created in such a way to achieve the highest rating given by the model.

The thing is that our advisor wants us to find a way to measure the quality of the work done, i.e. how do we know that the palettes proposed by the model are actually aesthetically pleasing? I know that this question has the subjectivity component to it, but can you suggest any ideas on how to measure the effectiveness (whether the color schemes make sense aesthetically) of our methods? Thank you!

1 Answer 1


A record breaking downvote magnet: You should program a formula for aesthetical nicety.

A serious answer: You can launch an empirical test. People should tell how pleasant they see your combinations vs some other combinations which satisfy some common demands but not those which you claim to be superior.

This kind of test is difficult. You can conveniently reach only a limited set of people and very likely they have some common feature which is easy to pull in the frontline when one wants to sunken your achievements.

Another possiblity is to scan say 50000 commercial webpages and find the profits of the owners of those which match with your criterias superior to the others. Of course you must develop some software to read effectively generally available business statistics sites.

The 3rd possiblity is to show that your system is in accordance with some already existing scientifically acceptable pleasantness research. I would start with this.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.