I'm not very skilled in C# (to my standards) but I was wondering if its possible to adjust the landmark detection in Unity.
Right now I am building upon the sample script that provides the specific landmarks/trackers but some trackers especially the ones at the corner of the mouth, don't move properly especially when lips squeezed together to form an O shape.
I saw a guy at an intel sdk video suggesting that the provided scripts are not that great.
So is it possible to get better results by custom scripting the trackers/landmark detection?
Yes, the supplied TrackingAction, in its basic unaltered state, does struggle with detection of some of the smaller facial landmarks. I halve always found that for making an 'O' shape with the mouth, the 'Landmark Lower Lip Left' or 'Landmark Lower Lip Right' are best. These can detect a strong expression made with the mouth, yet - unlike landmarks such as Chin - are not so easily triggered that you find yourself constantly generating the 'O' expression by accident.
You can also amplify how much movement a TrackingAction generates by opening the script in the Unity script editor and making a simple edit to insert a multiplication sum. I one wrote a step by step guide on how to do so.
I recommend creating a special version of TrackingAction to do this so that you do not affect the speed of any other objects that are using TrackingAction. I also have a guide for this.