Greetings,
About differences in the bunching up business: It is only relevant when stimulus timing is different across subjects. Either for uncontrollable constraints on stimulus events such as when trial duration is determined by a subject's response, or because one decides to randomize the stimulus presentation orders within runs for one subject and across subjects.
It is OK to scale your regressors as long as the scale factor is the same for all subjects in a study. But I just don't recommend doing this.
cheers,
-z