Friday 20 January 2012

L.A.Noire and MotionScan


Recently I’ve been playing last years hit game L.A.Noire – and whilst I am enjoying it, I also cant wait to finish it and move on to something else. Here’s why.

Now at the risk of boring you with stuff you already know – I’ll sum up the background quickly. The game is set in post war L.A and you follow the story of a police officer called Cole Phelps who works his way up from a lowly beat cop through to traffic, homicide and various other departments. Whilst there is a huge recreation of late 40s Los Angeles to walk, drive and shoot in, the majority of the time you will either be searching around for clues at a crime scene or questioning/interrogating people.

Now the big draw of this game is the insanely detailed facial animation courtesy of the newly developed MotionScan technology – a way of recording an actor’s performance in incredible detail and reproducing it within the game. Sometimes it’s so good you occasionally forget you are playing a game and are just watched a TV program.

In fact, you could argue that the whole game is built around this technology and as a result, somewhat adversely, lives and dies by it. When you are questioning witnesses or interrogating suspects – a good portion of the game is spent doing this – you are supposed to carefully watch the facial animations to guess if they are lying or not as they spout their testimony. All nervous tics, shifty eyes and other subtle facial gestures are recreated well – it’s just a shame that the actual gameplay surrounding these amazing feats of animation are largely inconsequential.

Every so often after hearing them say something you have a choice of deciding whether or not they are telling the truth, if they are lying or (somewhat confusingly) if you choose to simply ‘doubt’ them – which means you think they are holding something back but you cant prove it. Getting the answer right will give you more details about the case, getting it wrong usually means you get less information. There have been a few times when I knew the person was lying but didn’t know which piece of evidence specifically that you need to refer to in order to confirm this (a wrong piece of evidence selected causes you to ‘fail’ the line of questioning, as would only choosing ‘doubt’ when you were supposed to choose ‘lie’).

Regardless, ‘failing’ the questioning has only very few times led to a ‘game over’ screen – the majority of times it is inconsequential to the flow of the story – the game just carries on to the next scene. On one hand the developers Team Bondi could have made this aspect of the game less cut and dry and really effect the overall outcome of the story – but on the other hand, seeing repeated game over screens and having to watch and listen to all the dialogue every time you fail is guaranteed to have you reaching for the power switch.

The facial animation tech is amazing to watch – it’s just a shame that it couldn't be implemented usefully in any other aspects of the gameplay.

There ARE cut scenes to watch – lots of them – and whilst the facial animation is still brilliant here, the rest of the body whilst adequately mo-capped, is not up to that high standard set by MotionScan, sometimes leading to a jarring effect and the occasional tumbling into the uncanny valley.

The game’s other activities, such as shooting and driving, whilst competent, have been done better in other games and the remainder of the time all you are doing is walking around crime scenes pressing ‘x’ at every object hoping to pick up all the clues so that the story can progress. Whilst at first this investigation can be exciting and intriguing – it becomes rote and mechanical after the thirtieth time you have to do it.

Whilst the city has the appearance of a highly detailed and bustling metropolis, there is sadly nothing to do out there apart from visiting the next location on the current case you are working on (and a handful of very short ‘street crime’ sidequests) – making the whole thing feel strangely empty. Most of the time I found myself skipping the driving sequences entirely by asking my partner to chauffeur me over to the next scene.

This all points back to the MotionScan. Clearly this is what the majority of the development effort went towards (apparently over 400 characters had their faces animated this way) and it’s just a shame that it didn't somehow play a wider role. Perhaps a sandbox detective game was an overly ambitious project to first utilise this technology in?

We can hope that facial animation of this calibre makes an appearance in future games in order to capture more realistic performances from actors – but sadly late last year, despite L.A. Noire selling well, Team Bondi went under and was shut down.

Despite this, I can bet that the technology is sure to return in some form in the future and be utilised in different ways in gaming – although it is sure to be an expensive and painstaking process – L.A. Noire reportedly took seven years to develop. It would be interesting to see who will take on the risk next.


Whilst L.A. Noire is for the most part an enjoyable experience thanks to excellent writing and intriguing story and setting – I don't think I could sit through its lengthy investigations and cut scenes again once I’ve finished it. Which is shame since MotionScan creates animation that is truly astonishing.


To see a demo of the tech and some spiel from the developer, click HERE.


18 Jan 2012


No comments:

Post a Comment