Instruments for DFM MS are combinations of three sinusoidals with a volume 
drop-off function. This produces waveforms of surprising variety. The 
equation is of the form:

sin(a*sin(C1*b*x) + c*sin(C1*x)) * exp(-pow(C2*x*d + e, f) - C2*x*d + g)

where x is the current sample number (relative to the note's start sample),
C1 & C2 are constants that you don't need to know about, and a - g are 
parameters.

MIF files have 9 instrument parameters, yet the equation includes only 7.
Here's a parameter table to explain:

Param #	Letter	Value 	Description

0	d	~2.3	Dampening - describes how much of the volume drop-off
			is used, and thus how fast the fall-off is.

1	a	var.	These specify
2	b	var.	the "voiceprint"
3	c	var.	of the instrument

4	e	-1.15	These specify 
5	f	var. 	the volume
6	g	0.45	drop-off function further

8	-	var.	Main volume - how loud the pure instrument is
9	-	var. 	Effect volume - how loud the special effect is

The special effect is a cubic frequency interpolator to smooth the freq 
transition between two notes. In reality it produces a percussion-like 
effect when used on multiple notes simultaneously. The result is freaky at
the least. See geocities.com/ahel/EMS.html for some samples (try 'Dance of 
Static Electricity' or 'Path of War'). 

You won't need to toy with parameters 0-6 very much after you've specified 
your instruments, but the capability to use a trillion different
smoothly blending instruments in a piece is still there. 

A graphing calculator or graphing software is nice when toying with the these
options. You may want to try and plug in one of the sides of the equation
(sin(sin+sin) or exp) and play around with the parameters. 

Here are a few tentative instrument specs:

Approx. name	0	1	2	3	4	5	6

Horn		2.25	1.13	0.5	1.5	-1.15	12	0.45
Piccolo		2.25	0.23	2	1.3	-1.15	12	0.45
Pipe Organ	2.2	0.5	0.25	1.4	-1.15	80	0.45
Soft Saxophone	2.25	2	2	1.2	-1.15	6	0.45

Of course, none sound like the real thing, both because the volume function
is not too flexible and has not been properly fitted, and because DFM synthesis
has serious limitations. Still, they sound rather nice, and should be plenty
enough for the first time. I owe the definition of DFN synthesis, and the 
instrument parameters 1-3 to Seum-lim Gan. His work describing their derivation
in detail can be found at: 
www.media.mit.edu/~gan/Gan/Education/NUS/Physics/MScThesis/