A lot of the discussion in this thread has been about how many tubes of  foam to inject, which holes to inject it into, which layers to inject it  into, whether cavity foam is required and if so, where it should be  injected. Some of uncertainty arose when people started to look closely  at BMW's original procedure and question it. Also, conflicting  information has been posted in various forums.
Since there's so much conflicting information and injecting foam is  irreversible, I decided to take a look at this to satisfy myself how it  should be done before I do it. I wish I had an uninstalled RACP to  inspect, but since I don't, I looked for photos. There are a lot of  photos, but unfortunately, most are of the rear mounts. There aren't  very many useful photos of the front mounts. I used images from M3F (
here and 
here), as well as from posts on 
www.vrperformance.com, 
www.e46zone.com, 
m3cutters.co.uk and 
aussieexotics.com. Thanks to those who posted the photos, especially maruscmn and thomn8r.
Since there are so many good photos of the rear mounts, I started there.
Below is a photo of an "original" coupe RACP. I've highlighted some notable areas.
In the red, you can clearly see that at least in this view,  the driver and passenger side rear mounts appear to be mirror images of  each other. Without cutting an RACP open, I have no way to confirm if  they are mirror images inside, but I'd be surprised if they aren't. I  think it's safe to say that the claim that the passenger side mount is  different can be disregarded. Based on this photo, the procedure for  injecting foam into the rear mounts should be the same for both sides,  at least for the original design RACP.
In the yellow, you can see that if there is a passage from the cavity  between the rear mounts to the cavity between the front and rear mounts  (the blue area and the corresponding area on the passenger side), it's  quite small. Again, there's no way to confirm how big the passage is  without cutting an RACP open.
Based on this photo, in my  opinion, the best places to inject foam in the rear mounts are the  smaller outer openings on the driver and passenger sides (highlighted in  green). My reasons for this are because the areas to the far left and  far right are small volume, closed off (the foam has nowhere to go) and  the two openings are very close to the mounts. If all of the foam is  injected into these openings and allowed to flow inward from either  side, the mounts will be completely encapsulated and any air pockets  will be pushed in one direction, inward. It would be useful to know what  the volume of the cavity is to confirm the amount of foam required.  When / if I inject foam into the RACP on my car, I will probably close  off the larger openings (highlighted in purple) to force the foam to  stay within the internal cavity. I think this makes sense, because the  load on the rear mounts is tension (particularly on the driver side), so  putting foam above the structure doesn't serve any purpose, whereas  filling the entire lower cavity with foam will distribute the load all  the way across the RACP instead of only toward the sides.
Here is a close-up of the driver side rear mount. The highlighted area  is on the original, but note the area where the passage to the front  mounts is located. If there is an opening between the internal cavity of  the rear mounts, it's small.
Below is a photo of a new RACP. I highlighted the major difference,  which is what appears to be a barrier inside the rear cavity on the  driver side.
Here are a couple of close-up photos showing the difference. If it's a  barrier, as suspected, then it will prevent foam from flowing toward the  inner part of the rear mount structure. Perhaps the additional  spot-welds noted by Redish are an improvement, but I don't think this  barrier (if that's what it is) is an improvement. Since most people are  injecting 2 x 420 ML into this cavity, it would be useful to know what  the volume is. It would also be useful to know where excess foam would  try to exit the cavity if too much is injected. If someone wanted to  completely fill the rear cavity all the way across, it would might have  to be done from the passenger side, depending on how tightly the barrier  seals the cavity (assuming it is a barrier).
Below is a photo of the driver side rear mount inner cavity. You can  clearly see that if foam is injected to the left of the mount, it will  easily flow through the mount, fully encapsuutf8g it.
Below is the only photo I've found that shows the entire front of the  RACP from above. If anyone knows of a better photo, please let me know.  Unfortunately, the quality isn't very good, but you can see the layer  that forms a cavity around the front mounts. It's not completely clear  from the photo, but it appears that the passenger side and driver side  cavities are not connected. It also appears that there is no barrier  between the area under the rear seat and the cavities to the left and  right where the RTABs are located. Those cavities are quite large and it  would be wasteful if foam migrated into them, so it makes sense that a  substantial amount of cavity foam should be injected into the openings  on the outer sides of the seat.
Below are photos of the front passenger side mount. The red area is the  passage from the cavity above the front mount to the cavity between the  front and rear mounts. The yellow area is the seatbelt bolt hole. The  blue area is an internal hole that is not externally accessible, which  is unfortunate, because it otherwise would be an ideal place to inject  foam directly into the mount. Since the load on both sides is  compression, I think it's beneficial to completely fill the front mount  cavity.
Below is a photo of the front passenger side mount looking front to back.
Below is a photo of the front passenger side mount looking down from  above.  The red highlight around the mount is from the original photo.
Here is a photo of the seat area looking back to show the openings where  foam can be injected. Based on what I said above, I agree that the far  left and right latch bolt holes should be injected with cavity foam to  prevent the structural foam from migrating into the cavities where the  RTABs are located. I also agree that structural foam should be injected  into the next inner most holes until it appears at the seat belt bolt  hole. After that, I think foam should be injected into the seat belt  bolt holes until it starts to leak out of the latch bolt hole, at which  point the latch bolts should be replaced and foam injected until it  appears at the middle holes. I don't see any point to also inject into  the latch bolt hole because it's so close to the seat belt hole.
Quite a lot can be determined from photos, but the only way to determine  the volumes of the various cavities is to take measurements from an  RACP that is still fastened to the surrounding panels. Unfortunately,  due to the way the RACP fastens to the surrounding panels, a lot of the  metal would have to cut out. Basically, from the front edge rear seat  where it joins the floor, across to either side, all the way back around  the wheel wells and back across the trunk through the middle of the  "spare tire well'. When I first asked an autowrecker how much they would  charge for the labour to remove an RACP they said $300, but I suspect  it would be more based on how much of the surrounding panels would have  to be removed to preserve the volumes of the cavities. If anyone would  like to see this done and you are willing to share the cost, send me a  PM. I'm willing to do the work, but not to part with $400 or more.