I don't see the attachment.

I'll bet you're working with a table valued function, probably a multi-statement function instead of an inline. Right?

If so, the optimizer doesn't and can't know what to do with that when it's desiging the execution plan, so it creates a plan for a one row table. That sure sounds like what you're seeing. It's a major reason why the multi-statement UDF is a construct to avoid when you're dealing with sets of data larger than 10's of rows.