The state of mo-cap, part two

Part two of Develop magazine’s feature on the state of facial motion-capture

“A lot of the trick to selling animation and mocap as reality is actually in the context in which it is shown,” insists RealtimeUK’s [Ian] Jones. “Many of our most convincing pieces use realistic environments, lighting and true to life camera work. We build a situation in which the animation is strongly supported by the visual direction of the movie.”

One of the article’s illustrations put me onto the new FaceFX Studio (standalone download, 37Mb) which is a specialist facial animation tool with very delicate controls. It’s save-disabled, with no FBX export of heads/animations, and no screenshot rendering — but is otherwise fully-functional. I couldn’t find any way to re-light or green-screen the two sample heads, although they can be fed new lines as audio files which can then be performed. And the resulting animations can then presumably be captured using FRAPS but — although the interface can be cleared for a clear view of the heads — the basic quality of the sample animation didn’t entice me to bother doing a capture. If the samples are the best it can produce, I really wasn’t convinced. OK for churning out thousands of basic game animations, maybe… but I can’t see it competing with Face Robot except on price.

Still, it might be useful for someone’s iClone sci-fi movie, if you need a screen-corder capture of an alien on a communicator screen or being scanned in a medical bay…

Maybe there will be plugins that hook into webcams, so as to allow actors to “drive” the faces in real-time? That might be one way of improving the animation. Like this…

Hands-free pinball

A few weeks ago Dulci asked of Z-Brush… “can I just think at the screen, and it does what I want?”. You might have chuckled, but here’s a recent video of a guy playing pinball by just thinking about the moving the paddles…

Machinima.com inks Fox TV deal

Machinima.com‘s Machinima Comedy Lab unit finally gets to peek through the door at the world of big TV, having hired a stable of 15 quality Hollywood writers six months ago in order to reach the… “highly desired 18 to 34 male audience”. I guess some of these guys must still be watching TV (that’ll be the ones without fast broadband?), since FOX has just commissioned a machinima series from Machinima.com. Heel will be… “a love story between a man and his sociopathic dog.” The dog loves the guy, but dislikes his master’s family. The pilot will feature 7 to 9 minute cartoons rendered in… “one of the major game engines” using real-time animation…

“The creative team can alter camera moves, lenses, angles, character performances and edits in real time, thereby rendering the episodic animated television show into a virtual sitcom style of production.”

Strip copyright Eric Hews, “Yo & Dude”.

Muvizu beta: “They have a cave-troll timeline…”

New “real-time view” animation production software, from Scotland — Muvizu 3D (free beta). Muvizu seems like a neat, if deliberately simplistic, Windows application for quickly making cartoony-looking narrative animations in 3D. The aim is to…

“allow you to make animated videos without fancy equipment and years of training”

It was released a couple of months ago by the same company which was once behind the “virtual newsreader” Ananova, and the “design your own woman” Fantasy Babe application — before running out of cash and going bust in 2007. Now it’s back, with Muvizu.

The public beta is save-enabled and doesn’t seem to be time-bombed or crippled in any way apart from a small corner watermark. With a little care in the set design, framing and camera FOV, the watermark could be cropped out, or hidden with a prop if using it as a video-background in iClone…

It appears to require an internet connection on install, but doesn’t need a serial and thereafter can run without connecting to the net (I firewall-blocked it, as a test). The ease-of-install and lack of anal-probing DRM seems like a very good start. And just a few days ago the developers added YouTube upload.

Prepare for a chunky 465Mb download, and you’ll need a decent Windows desktop PC + a good Nvidia gaming graphics-card + masses of memory to run it because…

“the software is based on the Unreal 3 gaming engine”

And very nice it looks too, once you start working the lighting. You can check out the quality on the website gallery, although the scripts appear to lean toward the drunken-student end of the spectrum…

Anything that makes the Unreal Engine accessible to ordinary mortals without having to read three 600-page books and watch weeks of video tutorials is certainly welcome. There’s a lot to like, including the lack of an “uncanny valley” because it uses 3D cartoony characters. It feels very easy and intuitive to an iClone user, but there are also powerful features such as a timeline, gravity and collision physics (currently a little too sensitive?), and lip-sync. There are primitives which you can colour-shade, but not add textures to.

Tutorials are here, but are delivered in a strong Glasgow accent which Americans may find takes some getting used to.

The license statement is anymation friendly…

“Muvizu 3D is free to use for non-commercial purposes, however if users want to use it commercially then we will organise a licensing agreement which is done on a case by case basis.”

Output can be up to 1280 x 760 pixels and is a standard .avi with a very basic choice of video codecs. There’s no anti-aliasing (I vaguely seem to remember that’s one of the major failings with the Unreal 3 engine, because it has to cater for the console-kiddies?). But on re-sizing the video by 50%, most of the jaggies vanish. Heavy YouTube compression will probably take care of any that remain.

My tests showed that characters can easily be green-screened and thus brought into iClone using popVideo Converter. Set designs can also be brought into iClone (albeit with some nasty jaggies) as background video textures or 2D stills.

So, what’s not to like? The main failing is the lack of import. You can import your own dialogue sound files [update: and background textures], but there’s no support for OBJs or a hook-up with Google 3D Warehouse. All the professionally Maya-created low-poly assets, including textures, are protected inside zip-style archives files which use the .upk (Unreal Dev Kit) format. Although I wonder if someone with the free Unreal 3 Devkit could write their own .upk archives containing their own assets pack for a movie, drop them in with the other .upk files, and have them available in Muvizu? I can’t immediately see anything in the .ini files that would prevent that.

Another failing is the lack of ability to caricature people, judging by the almost-unrecognisable range of characters which are supposed to be the main players in the forthcoming UK General Election. Although, given the UK’s libel laws, perhaps that’s for the best…

In terms of expressiveness, the eyes are the failing at the moment. They need catchlights.

Assuming Muvizu can build a strong and pleasant community of dedicated makers and developers, doesn’t run out of funding again, and can further expand the character-customisation module so there’s a huge Spore-like range of possible combinations and textures, it seems Muvizu might have a future.