Sorry, I’ll start by apologizing because I tend to over think things but I think you’ll understand why this is not a simple answer.

The Background
The musician can crescendo and decrescendo.
The recording engineer can make musical parts louder or softer in the mix.

There can be an important difference particularly when it comes to synthesized music. Because sampled-based instruments often require different velocity values to recall their increasingly harder strike samples, simply raising the Volume slider does not change the audio sample with the harder strike result. This is particularly true on percussive family instruments. So on hammered-struck-or-plucked instruments a crescendo can mean increasing the velocity of events over time, by structuring the events... not just raising the volume of the track.

Changing the velocity of a musical phrase can result in different articulations/different samples being selected.
Changing the volume of a musical part can result in the currently recorded articulations being heard louder.

This can be a subtle or a profound difference.

Crescendo for percussion family instruments typically requires you to play harder over a series of notes - but it is clearly a “musical” thing.
Fading the volume up is a mixing gesture of the engineer that makes the playback volume louder...
Crescendo for self-oscillating family instruments do not necessarily require a series of notes to change volume over time.
Here fading the volume is a musical gesture than can be made more realistic if the tone/timbre also changes as the loudness changes.

You’ve heard thousands of records where the engineer “fades” the music out at the end. This is not accomplished by the musicians... they are not really changing how hard they are playing in most instances, the fade out is strictly volume fading down. In plain language you might describe this as a decrescendo, but musically it isn’t. Musically it’s science fiction... it didn’t really happen. The musicians continued playing as normal (the same velocity), it’s the volume that changed.

CRESCENDO (Reference Manual Page 89) works on increasing/decreasing Velocity
CREATE CONTINUOUS DATA (Reference Manual 92) works on Volume of sustained notes

Volume and Velocity
VOLUME changes can be overdubbed to the track using CONTROL CHANGE (cc) messages. You can use a Knob or a Foot Controller to adjust the Volume in real-time with the Track in “Overdub Rec”... once the data has been overdubbed, you can undo the most recent, record, you can edit, change, delete, just your overdubbed data without disturbing the previously recorded note data. The MOXF Sequencer has a VIEW FILTER which will allow you to view just the Volume/Expression messages you recorded. You can edit that data using the Event list (EDIT) or via the various Sequencer [JOB]s.

VELOCITY changes can be edited (adjusted) using Sequencer JOBs (or by replaying the data)... VELOCITY determines how much energy was used to turn the Note-On event On. Example: A snare drum roll that has two dozen events all at the same Velocity...if you increase the Volume will get louder over time. But if you alter the Velocity because there are five different snare drum samples between soft velocity and hard velocity, the roll will seem a bit more realistic as the timbre, Length, etc., of the drum stroke is changing, as well.

There is a JOB that allows you to “Modify Velocity” across a precise Measure/Beat Range as necessary.
There are two Control Change messages that affect output level ... cc007 main Channel Volume, and cc011 musical Expression.
cc007 sets the main Volume for the Midi Channel... say you set the Fader at 96... that becomes the cc007 value.
cc011 works within the maximum set by cc007. Such that say you are controlling output level with an FC7 Foot Pedal Controller... heel-down would be 0, and toe-down would never exceed the maximum set by cc007... so toe-down would be the equivalent of the maximum set by the channel Fader (96).

If you want to just adjust the VOLUME in the Mix:
Place the Track you want to adjust in OVERDUB Record... use either the Knob (cc007) or the FC7 (cc011) to Record you changes in real time.
If you like your Overdub keep it. If you do not, you can always *undo* the most recent recording (so you can undo the Overdub and try again).
If it is close and you manually want to edit what you’ve overdubbed.
Select the Track > press [EDIT] > [F2] VIEW FILTER > Set it to view just the events you want to see.
(See the Manuals for details on Event editing... each Measure:Beat:Clock that has an event can be shown...edited
STORE any changes you make.

SEQUENCER [JOB]s
In addition to the CRESCENDO Job, you’ll find the MODIFY VELOCITY (useful for making straight changes to recorded Velocity) - both work on Velocity - where Crescendo is more about shaping the *change* in Velocity values over time. If you want to use MODIFY VELOCITY (Reference Manual Page 88) you have a great explanation in the manual on how to work with this Sequencer JOB as an offline Edit. Again you can ‘trial and error’ it.

Because you have an UNDO feature this makes it painless to try out different ‘fades’ using the Crescendo JOB... you can set over how many Measures and Beats the fade up/down occurs... from what value to what value... and you can even weight/shape the movement ... very powerful.

CREATE CONTINUOUS DATA (Ref Manual Page 92) allows to generate cc messages offline with great precision. You can set the measure/beat range, the cc message you want to create, the density of the change, you can even weight the Curve, etc. You mention you want to enter Pan messages as well (cc010) - these can also be entered in real-time using the Knobs or can be entered offline by using this Create Continuous Data JOB. PAN position allows the illusion of Stereo to exist... when sitting in the sweet spot area in front of a pair of Stereo speakers, adding electrical resistance to one side causes the signal to appear more in the other. Panning something right is accomplished by adding resistance to the left channel. Because we have ears on each side of our head we locate sound by whether it reaches one ear before the other. As you pan right, your brain feels as if that sound is coming more from the right side.

The positioning of instruments in front of the listener is a critical production decision. Sounds that actually move in the stereo field create the illusion of someone moving in front of the listener.

Decision
It is a production decision as to which adjustment (Volume or Velocity) is going to work best. Non-percussion instrument’s can get louder during a held note... so musical crescendo for a String or Horn player is very different from the hammer-struck-plucked instrument (in most cases)... crescendo and decrescendo can be accomplished without altering Velocity, in fact, Velocity modification may be not at all effective to accomplish your goal. Non-percussion instrument’s are considered the self-oscillating family - where a Pressure is applied to create the musical vibration (typically by bowing or blowing) and as long as that pressure is maintained the sound continues. This allows the instrument to be adjusted in volume *during* the Note event. Velocity modification will only work if there is a phrase of notes for it to be applied to. Since velocity is a one time event that is read at Note-on.

Once a percussion sound is started typically the player’s role in the encounter is over (these are not 100% things but, in general).

By the way, a special Fade Out is available using “SYSEX” under “Create Continuous Data” JOB in MOXF... It uses the Sysex message for Universal Real-time Volume to “fade” the entire MOXF Master Volume (great for designing those sci-fi Fade-Outs or Fade-In’s that affect all Parts together!

Hope that helps. The JOBs may look complex at first glance but they are actually fairly easy to use. Experiment with them... if you get stuck or have specific questions, post back to this thread.

Lots of good information here --- I'm sure you've offered several solutions!

Thanks
Todd