As I keep pointing out, there is a lot of phrasing that will happen naturally. However, we often need to indicate when to swell.
Conventional wisdom is that in this case (Ex.1), they will go up one dynamic. I have found that more often than not, all sections of the orchestra go up almost two levels. The lack of a target dynamic may also lead to a question from the orchestra so I find it is best to put one in to make it obvious.
Now we have added the mp in the middle (Ex. 2), but guess what? They still have a habit of overshooting.
What I have found is best for a light, unforced swell is to use the word poco. I hate to admit it, but sometimes even then it may be too much and we just ask them to think about the crescendo.
Note that I did not put a dynamic at the end of the diminuendo in the examples. It is a given that they will come back to around the start dynamic. I say around because players are not computers, and to some extent their behavior depends on context and training. More often than not they will actually dip under the start dynamic if they rest in the next bar. This happens EVEN if you put a dynamic. For example I want to go from mp to mf and back. Even if I put mp at the end, they will more than likely end more quietly than they began. I have also found that putting p, pp, ppp or niente makes no difference. I have conducted scores where one phrase has a diminuendo down to pp, the next p and another ppp or niente, but they all sounded the same to me! My theory is that players naturally taper off the ends of phrases; basically, they will come down to the level you want, but then add their own little taper, which means they are now lower than the dynamic you have notated. It sounds confusing, but is really not an issue, as it sounds very natural, the players know what to do, trust them.
Swells are a great place to over-notate; it is so easy to copy and paste. It also makes you think you have put a lot of thought into it, and helps your computer play the music back better. But we are not writing for what you think, or what your computer thinks; we are writing for an orchestra of amazingly talented people who only see and have to think about one part. Look at this example.
In reality, after the first two bars you only need the hairpins. By default they will repeat the last instructions. One could also just use the word sim. after the first two bars and skip the hairpins as well. However I have had trouble with getting players to continue the swells with that notation. I have no idea why, as they should be able to do it, but I often have to remind them. This is just one of those things that makes no sense, but I can tell you from experience it is a fact.
While we always strive to nail the dynamics, when your reference is a computer mockup, getting every one of them to sound exactly as the composer wants it is not going to happen. It is very common to massage the dynamics after the first run through. I am always conscious of this when I notate. My goal is to make it as easy as possible for the players to make the changes in the part. If you have gone and pasted everything out like Example 4, the players have to change a lot more ink if the dynamics change. If you do it like Example 5, they only have to change two bars. I use this concept for articulations and slurs too.
Bar 1/2- The hairpins should never touch the bar line. Leave a little space.
Bar 3/4- This looks messy and implies a later diminuendo.
Poco does not replace a dynamic. It should go on the hairpin up (Ex.3) as that is what it is referring to.
The poco makes no sense. It is not a ‘little’ crescendo, it is a one dynamic crescendo, and you have already specified that with the mp indication.