Heavy Rain used Passive Optical Motion Capture, both for body- and facial capture.
These two were done separately, recording the body capture in one take and the facial capture in a separate take.
Facial capture was most likely not used very heavily in intense action scenes, but rather for the more “intimate” scenes in between when the player gets to see the faces more up-close, inciting an emotional reaction.
I’m not sure how much the method cost Quantic Dream during development, but it appears that they purchased the equipment to be able to use it in house, which even caused some attention during the development. I’m assuming it was still fairly unusual to do so, or perhaps the equipment was rather sophisticated/expensive for a single studio seven years ago. As far as I could tell they were able to use at least three actors during takes, meaning a minimum of 12 cameras if I have understood correctly with their motion capture equipment, although I am guessing they used a bit more than that for a high end game.
Considering the nature of Heavy Rain, it was most likely very convenient for the development to purchase the equipment as much of the game consists of action time events where unnatural movements might hurt the player’s immersion.
Mainly they appear to have used the key-talents for the majority of the motion capture, however when it came to some of the more advanced movements, like when the characters jump from buildings etc., they turned to stunt-actors. Also the character Madison was, unlike the other characters, created visually from her stunt-actress while voiced by another actress. If this applied to the facial capture as well wasn’t very clear though, but it wouldn’t seem very clever to take the facial capture from one person and the voice recording from another.
During the body capture the major movements were captured, such as running, jumping, struggling with attackers or simply body language. The finer points though, like fingers etc. were not recorded, but was most likely animated afterwards.
The facial capture took place separately though, and recorded both the actors voice and facial expressions.
As far as I could tell the development team was very pleased with the outcome of the motion capture for this game, and they have received praise for how impactful and natural the characters are. Although when browsing the web more focus lies in the comparison between Heavy Rain and LA Noire, as they used facial scanning rather than passive optical motion capture soon after, which caused quite a fuzz over which was better.
I picked this game because it was the first one that I could really think of that I had played that used motion capture to a great extent, and in my opinion quite effectively.
Honestly I never really looked into how motion capture was implemented during development very much before, so I pretty much assumed that the actors and characters would look very different from one another, and not the character created from the actor. It was rather obvious that it made for a more natural performance though when looking at the facial capture of the detective in the game. It would most likely have been a bit trickier to get it quite so natural with different facial structures.
The best part of how they have used motion capture is probably how they applied it to the faces. They often appear surprisingly natural and engaging. And the worst most likely the times when the body language either just comes across as stiff or very much exaggerated. I’m not quite sure if this is a result of the motion capture though, or if it is a result of the direction for the actors/actresses.
The sources I used for looking into Heavy Rain’s motion capture were these: