A few possible improvements to our project could be:

• Adding a scripting component to identify and correct the direction of the letters. 

•  It is possible to run the shader in a few layers using multipass and this way add depth to the image and improve the overall look.

• LiDAR API for AR foundation & Unity supports classification of objects. Identification of humans and other complex entities will be very easy to implement. This can be used to customize the shader on these objects and also improve the look