What I derived is:

functions can cause Row by row processing and also that would result in a table scan as for each row the function will have to be executed. The analogy which can be taken here is the use of aggregate function in the where clause.

Questions:

In case of table valued functions how does the processing take place? .... I read that joins can be executed by FunctionReturningTable(parameter) ... In this case the execution plan generated for the SP will include that of Function ... but when the function is in the where clause how does the database engine create a execution plan?

how and why in the above example returning a table reduce the logical reads?