It has been a fear of mine that most if not all studios would give up on making stylized human characters in favor of scans. Some of the most memorable characters (for me) in games have been hand modeled because the freedom it allows to make truly unique looking faces that wouldn't exist in real life. IMO one of the biggest issues with scans is that if the rendering and rigging isn't good enough (so pretty much all real time graphics at this stage) the result is almost always off-putting, especially if the actor is well known outside of the game (Kevin Spacey in advanced warfare for instance).
So as budgets keep reaching towards a never ending horizon, its inevitable that actors outside of games will become more and more common. However, I hope that talented artists will still keep creating beautiful unique characters that wouldn't exist outside their minds because not everything needs to be a scan.
There are two sides to this, and both have a lot of pro and con stuff to weigh when a studio makes this decision.
I guess it's quite obvious that you'll have a lot more freedom when you only hire the talent for the performances. You get access to a wider pool of actors as you don't have to match the looks with the ingame character. Also, and I really don't mean to offend with this, there are a lot of talented actors who are not as good looking as movie and TV productions would need them to be, but would probably be more than capable for the part. And, since they have less opportunities in those industries, they aren't as demanded and probably cheaper to hire. It's easier to book people and match schedules for capture sessions, and there's also a good chance that the talent will already be familiar with the process. Again, just stating the facts, no judgments.
Another obvious advantage is the complete artistic control over the look of the character. Handsomeness, age, build and so on are of no importance, you can go as far as having two actors play twins, whatever. You can also go the other way and make someone look uglier, deformed, handicapped, or an alien or anything else; and you don't have to rule out scanning either. The best known example would probably be Halo 4 and 5, where Jen Taylor can do the voice for Cortana who's captured from another actress and scanned from at least two other people; but she can also do capture and voice for Halsey who's a hand crafted character.
Also, you can allow character customization for the player, which is a big deal for RPG type games. This is actually becoming harder to do these days and I wouldn't be surprised to see the new Mass Effect game ditching this option - but at least it's still a possibility with this approach.
However, creating a character from scratch at today's standards is pretty demanding, takes a LOT of time and effort and lots of really talented artists. You have to iterate many many times to get the looks right, you have to hand craft all the facial expressions as well, and there's little room for automation in the workflow. It's just not feasible to build a large ensemble cast this way; you'll notice that Uncharted 4 for example has shown less than 10 main characters so far and it's probably not going to get much higher with the final game. This has also been the case with the previous games as well, it's always just a few characters and lots of noname guys.
On the other hand, going with celebrity talent will offer a few advantages from the start and somewhat fewer drawbacks, at least from a production point of view.
Obviously you'll get good acting skills, likable looks, and a good voice. There are a lot of proven actors around Hollywood, in the TV industry as well, and AA productions can afford to make good use of it. You can nail the character from the start with good casting, and you can get really good performances; even rely on the talent to add to the project using skills and experience. Good directors are not yet that common in the game industry, so this can mean a lot. Also, getting a big name on board will generate a lot of publicity and I'm quite sure that it can transfer into sales as well.
So I still think that hiring Kevin Spacey was a good move, regardless of how well the ingame (or CG
) character worked in the end. I'd also like to point out that his involvement has sort of made video games "cool" for movie and TV talent, and thus we've seen a rise of interest in the past two years. The results may be mixed, but still, a lot highly talented people are now easier to get into video games.
Relying on scan data also allows for a lot of proceduralism: you can write code to process it and generate geometry and textures, increasing your efficiency and accuracy. Once you have a good pipeline in place, asset creation can get really fast; as I've mentioned elsewhere, 343i was able to get a fully rigged and textured face into the game within something like 2 days. I've seen some games with a huge cast of characters at this point, and that wouldn't be possible without relying on scanning. Also, scanning means that you don't have to make any guesses about the facial expressions, you will always have perfect reference for every aspect of the character. Extreme facial expressions can be really hard to get right without a guideline, but you get that with scans.
I'd also say that real life people can actually be far more varied in their looks than hand crafted characters. As I've mentioned, handsomeness is a really tricky thing - there are a lot of people with facial features that would be unpleasant on paper, but somehow we still find them to be attractive. Nailing this stuff is just incredibly hard, it'd take a lot of time and it's really hard to test the results; but if you go with real actors, you'll get the results instantly.
The obvious downside is that getting the likeness perfect can be really hard - well-known talent is always shot to make them look their best, they have professional make-up and hair artists, and so on. So you're not even aiming at the data acquired, but something a lot harder to grasp, basically a highly idealized mental image of the person. You're gonna need a really good art director to supervise this. It's also the same case with the performance, you have to study reference material really hard to catch those tiny ticks and gestures. However, I'm quite convinced that the technology is there now, and it's mostly up to the developers to make the most out of it.
So, all in all, I personally believe that there are a lot of merits with both approaches. Budgets and schedules are a very strong driving force, so scanning may become more common; but the various disadvantages will always be there, so we're probably going to see hand crafted characters as well, but less of them because of production constraints.