I was wondering if I always need to use the maximum value of the options from the elements of "Add Modifier". For example, do I need to use the view 6 in Subdivision surface? I know it depends of what we are trying to achieve, but we all want the same. Make something look the most believable. So do you always use the maximum value Blender offers?
My advice would be the exact opposite:
Use the minimum amount of subdivision that will give you acceptable results.
How much that is depends on the complexity of the shape and the resolution of the images you want to generate.
Needlessly increasing the amount of subdivisions comes at a price: RAM (memory) usage and computer processing. A higher number of vertices will impact your computer's performance. Keep in mind that for every level of subdivision you are increasing the number of vertices exponentially, so it's quite easy to end up with geometry that exceeds your computing capacity.
For more details read Render crash using subsurf modifiers.
3d realism requires taking clever shortcuts and fooling the viewer. Use high level of detail only on those areas that are featured prominently on the final image(s), objects that are going to be far away need less detail than those up close.
With subsurf, you don't want to apply it before you move on. This is meant to stay throughout the project (and even in final renders) to keep your base mesh simple enough to edit. Here is a great example of using the subsurf modifier all over the place:
You can look at nearly every production-quality file and see just about everything with subsurf modifiers still on a simple mesh. Here is another production file (a benchmark for cosmos laundromat) for you to look at: