If the cross sectional form of your cutter is that of a rack of the correct dimensions, and it is long enough to encompass any teeth in mesh then surely by definition it must cut the correct profile.
Obviously the teeth in mesh will vary dependent on the tooth count
Not so Andrew!
In a gear planer the cutting rack moves tangentially between strokes and the gear blank rotates a small amount generating the involute form.
With just one pass of a rack form cutter at centre height you get:

Rotating the blank a tooth (in this case 30 degrees for 12 teeth) you get:

That's as good as it gets - the involute is approximated with 2 flats.
Add a pass where you rotate the blank 1/2 tooth and raise the cutter the equivalent of 1/2 tooth you get:

The next pass would be another 1/2 tooth rotation dropping the cutter down to it's original height producing:

The involute is now approximated with three flats but some horrible looking artefacts at the root of the teeth.
So to generate a perfect involute you'd have to do an infinite number of passes rotating the blank and adjusting the cutter height accordingly. A compromise somewhere in between would give a very acceptable gear.
I chose a 12 tooth gear because this would accentuate the problem.
Hope this makes sense.
Cheers.
Phil.