Yes, we are using lots of scanning now.
Advanced Warfare's main cast of all 12 speaking characters was based on head scans of the real life actors who played them, the data acquisition was performed by Activision's team. We've also got 40-70 facial expression scans and such, but all the assets obviously needed processing. We've also had to modify the characters in some cases, which I expect to become fairly standard procedure.
However we've also built our own photogrammetry rig and keep upgrading it constantly. On AW, yours truly was the stand-in for the dead body of Will Irons and for the arms of all the main soldier characters; although more or less modified (like biceps size
) it's still fun to see my vein patterns on the hands and forearms... I've also been the test subject for the full body scan pipeline, and man it feels pretty weird to have several people work on your digital self or see yourself in shorts on the movie screen in dailies
It wasn't vanity though - I wouldn't ask anyone to do anything I wouldn't do myself, including shaving... And I also wanted to know how exhausting the scanning process is; the answer is that it's surprisingly bad, for example having to stand as still as possible for 10-20 minutes while the camera array is calibrated.
On a current, unannounced project, we're now using in-house scanning very extensively; the hero character is a heavily modified version of the mocap actor / stunt guy, and the other main character is based on a body double and a different actress's facial likeness and performance. But we're also adding artistic changes to that character as well. Another unannounced project is going to use a lot of facial scans for both main and background characters, but once again we'll modify the data and also mix in hand sculpted character faces, bodies and so on.
Understand that scanning is not a turnkey solution, at least not in our workflow (it is possible to go with 4D scanning where you shoot at 30-60fps and capture full facial movement, like LA Noire did). The assets are noisy, hair, body hair, eyebrows and eyelashes can't be properly captured; facial expression scans can only cover a small percentage of the full range and usually need to be modified as well to fit into the rigging pipeline. You need to do a LOT of manual work and it takes an accomplished artist to apply it at a proper quality level.
Also, there's still a lot of subtle high frequency detail that you just cannot capture, so you'll need to manually paint the fine texture details. Then you also want the character to move, which opens up an entirely new world...
However, the real life subtleties that you can capture are simply invaluable. Even highly trained and experienced traditional artists are usually somewhat limited in their artistic interpretation, and they also prefer to idealize both facial features and body shapes; and another probably underapreciated element of living people is the inherent asymmetry, which is usually way beyond what you'd expect. For example it's very common for faces to have a "fat" and "slim" side, the height of the eyes and ears are usually offset by several millimeters, muscles on one side may be far more developed, and so on. It's also interesting how far real life anatomy usually is from the classical idealized looks, for example someone you'd consider to be a pretty strongly built person might look surprisingly underdeveloped in your content creation app. This is another important lesson we've learned: no matter how hard you try, or what complex shaders you use, the interactive display in Maya or Zbrush will never really show a good enough representation of your character. However, funnily enough, when you load scan data, it still manges to look more convincing even with a simple Phong shader.
There's also an additional result from working with scans, your artists will learn just how far they can push their craft and benefit immensely from the real life data.
And then I haven't even mentioned the use of scanning for props and environments. Since they're static, you can shoot more images to get a more accurate solve from the photogrammetry software, and even do stuff like put a camera on a quadcopter drone... So the possibilities are endless.
But in the end it's still another tool in the box, a very useful one - but I don't expect it to completely replace talented artists, there's always going to be a need for a human eye and touch.
