I think I can answer this question - it is actually fairly straightforward. Latry is using the Tutti (without the chamades) at the outset - this includes the five clavier choruses, with their chorus reeds. A significant reduction at around 1'.20" uses a variety of sounds - including quiet fonds 8ft. and 4ft. At about 3'.34", he moves his right hand up to the third clavier (Récit-Expressif) where, in addition to the full Récit, the Boisseau Chamades (8ft. and 4ft.) are brought into play for the fanfare-like motif; (this appears, modified into single notes, again at 4'.34" - 5'.40" - possibly with the Grand-Chœur coupled and the Tuba Magna 16ft. in use - there is a distinct, strong sub-unison reed tone, here). Although the 'new' chamades (added in the 1990-92 restoration) are also available on this clavier, I believe that it is those by Robert Boisseau (which were installed around 1970*) used here. They have a distinctive sound. Those added more recently were deliberately modelled closely on those constructed by Cavaillé-Coll for the organ of S. Sernin, Toulouse. These stops (8ft. and 4ft.) have more body to the sound and are less 'sharp' in timbre.
Clearly there are a number of crescendi and diminuendi throughout the improvisation. The most powerful of these (at around 6'.00") uses the chamades (on the Récit) against most of the rest of the instrument. However, when he drops down to the first clavier (G.O.) at 6'.22", the ensemble, whilst being fairly full for most divisions of the instrument, has yet to include the chamades coupled through.
A diminuendo, commencing at about 6'.35", reduces the instrument to quiet fonds 8ft. and 4ft. At 7'.16", as far as I can tell, just 8ft. Bourdons are in play. At 7'.19", his left hand moves up to the Positif (unenclosed), where he uses the Unda Maris against the G.O. Bourdon 8ft., with high mutations on the Pédale Orgue.
At 7'.44", both hands play on the G.O. (lowest clavier), utilising mezzoforte fonds 8ft, with one or two quiet sub-unison ranks (perhaps the 16ft. Bourdons of the G.O. and Positif).
The final cresecendo begins at 7'.56", and progresses through the choruses and mixtures (from about 8'.04" onwards), to reeds (added at approximately 8'.17"). At 8'.38", most sections are 'full' - including the Pédale Bombarde 32ft., but excluding the chamades. A brief reduction at 9'.24", leads to the addition of (at least) the Boisseau chamades at 9'.31", for a grand statement, in octaves. By about 9'.34", the Tutti Général is in use (now including the chamades) - and remains so for the conclusion of the piece. I can assure you, from personal experience on a number of occasions, that this is nothing less than devastating from the console.
* I am not sure of the exact date of their installation. They were planned by the Organ Commission in 1959. However, they do not 'appear' on recordings prior to 1970, as far as I can recall. There is no doubt whatsoever that Pierre Cochereau would have used them, had they been available prior to this date.