The Emily Project
As much as this sounds like another open source project, it’s not. Rather, The Emily Project is a very impressive life-like human representation by Image Metrics.
Thanks to Michael Cuffe for the article suggestion!
Shutter Shades RELOADED
I attended a panel of several leading studio animators who gave a discussion on 2d to 3d animation . Their reactions were interesting when they were asked about their feelings as animators working on a movie with MoCap. Most hated working with it, however T Dan Hofstedt http://www.t-dan.com/ said he had worked on the Polar Express. Rather than have a distaste for MoCap he realized that it needed tweaking and made the best of it. He said that he animated 80% of Tom Hanks facial moves in the movie.
Later when I was talking with another panelist, he confided to me that when he heard T Dan talking about his experience and how he took a bad situation (bad for an animator) and and turned it into something good, that it changed his perspective. As the saying goes, “if you are given Lemons, then make it into Lemonade”
ok, it’s a bit old news, but one thing has always bothered me with this one: it’s only a facial rig. I mean — to claim to be across the uncanny valley, one has to create a full body animation.
impressive, yes.
The thing that bugs me about this is that they captured her movement and just put a cg version of her face back over the original. For this to be truly impressive, I’d want to see an entirely cg character in a different scene.
It’s cool, but just seems overrated to me (it’s been around for a while now).
Wow, thats pretty CooL!
just saw this a few days ago. simply amazing
looks photoshoped
Hmm i’m torn about this post, as it’s nothing we as Blender users can actually sink our teeth in. While it’s nice, the information on the website is sparse and I would have liked more information, examples and stuff. If only it was an Open source project or publicly available rig, then it would be nice. The example clearly shows its only the frontal face and not the rest (ears, mair etc.).
What the connection with Blender would or could be escapes me unless you want to inspire some developers and artists to make a lifelike human or something. It’s clearly a closed source product with commercial aspirations. While that’s not bad perse, it almost looks like an advertisement, yet no Blender connection, unless ofcource they used blender to render it :)
Cool but off-topic.
It’s indeed impressive, but as BlendRoid noticed, it would be even more impressive if the captured data could transpose to other faces and character.
At least you can capture the face of an actor to animate the virtual face of the actor model in 3D. But what if we wanted to animate a caricature of the actor, or a anthropomorphic animal caricature of that actor?
I don’t know if there is any info about that. It doesn’t make the feat less impressive. It’s just that there seems to be still some work before it become useable and mainstream…
…
Any chance we could have that for Blender in 2011? :)
Should have been included in the article: http://image-metrics.com/project/see-how-emily-was-made
It’s easy with Emily, my face would crash the computer.
This definitely falls into the “why?” category. If you’re going to physically recreate an exact duplicate of a person, then why not just use the film? I understand this is probably falling into the category of “we needed to see if we could do it” (which is not much of a question as it’s just a function of more data points and/or a better algorithm), and is more a demo of the technological potential, but, as others have said, it’s close to useless IMO until they can make me believe that she is real, as some CG character, or a completely “different” looking realistic human. This just goes to show that the technologists working for film, and all too often film itself, miss the mark that this digital technology is essentially useless if you don’t create something “different” with a visual style people will find beautiful, and if it’s a film, a good story trumps all. What’s the next application, a remake of the movie “Singles” with a digital representation of each character exactly as they were in the original, with the same exact voice/soundtrack?
Silly.
Let alone, this overall push towards realism really misses the mark. The whole point of animation used to be that it could do “more” than real life, and it could be pushed and sometimes the subtle human mistakes of the animator…stole the show. Now we strive to digitally recreate life and do a crap job of it.
Uncharted. Saw that already on CG Society a half of year or even more time ago.
Oh, I thought that it was obvious that this ‘visually tracked’ motion data could be applied to any arbitrary face rig. If it is not, then this is all of course pretty useless. I just didn’t think of the possibility that a company creates such a useless product :)
Well it could also be useful to make people look younger (like in movies where the actor must act mutiple period of his life).
You track the actor face, and modify the texture to alter the age.
Just thought of that case after thinking of the movie about Benjamin Button…
RedBeer you may want to check out one of the related videos on youtube, which shows that the analysed data is flexible enough to be used on other meshes.
Indeed, I mean they have videos about how IMAGE METRICS will fillm the person, and then produce the results for any general rig for the program of your choice. They show Maya, SoftImage, 3DS Max, Houdini, and they show one actor doing the facial patterns for a warrior of some sort. They also show the remapping of Fred and Ethel’s faces for that drug prescription commercial.
So there are some pretty amazing applications for this, I do wish however that a free method existed to allow those of us without money to make a lame attempt at the same thing.
@LOGAN: Thanks for the link. ;) It was neat to see the light sphere/3D scanner (whatever it’s exactly called,) and to hear more details.
True, copying real life can seem pointless, but we can learn a lot from it, and besides, if we have the power to draw the gap so thin that you can merge both mediums, live and CG, then you can go many places from there, including stylized or more traditional toon animation.
@Redbeer, name any actor that died too young or cool actors that died years ago. I sure would like to see a couple of Bruce Lee movies or Humprey Bogart movies more.
@toontje: Wow, good idea! :) That would be neat… digitally resurrected actors…
Part way through the video I guessed that it was digital, it looked funny. What about you all?
Anyhow, glad to see the industry moving forward… I hope someone will move up with it, though. Up towards an excellent animated film, and more…
This is ridiculous!
I send this as article suggestion back in early september this year. It was rejected.
And now… this news aren’t news anymore. It’s more than old.
@elliott: I know; this is one of the things you get when working with a team of people. I wouldn’t have chosen this one, but other people do seem to think it’s relevant enough!
making a super realistic head in itself is not useful. but it IS useful if you’d want the head to explode or her skull to open so you could see her brains (or other things real actors wouldn’t like you doing to them :P).
still to bad i won’t be able to use something like this for myself unless i win the lottery.
and i don’t gamble. ;)
The technology is interesting, but very believeable CG human have already been created for several years, without any special capture technology.
Liam Kemp has done an amazing work in this direction since at least 2003 :
http://www.liamkemp.com/news2008.htm
His movie “This wonderful life” is a master work, and his demo animations are awesome.
@elliot: Sorry to hear that. :(
@Bart: I’m still learning. ;) It does apply to 3D news, but maybe it didn’t fit for BN?
I’ve never liked this project. they got a model derived from automatic computer generated topology, which is often multi million vert and not very good for animation, then instead of using this as a good reference they just retopo’d it, used more auto computer power to animate it, which was only copying something that had been done in RL. This is really a lazy scientific project imo, the technology has been around for a long while now, it’s just that nobody has done it because it’s pointless and requires no talent. 3D is a tool. It’s designed for the artist, not for a computer to do things automatically. Now, I DO respect the programmers and people who developed this software. it is an amazing research project, and the animations created are AMAZING resources for learning about natural movements in detail.
digitally resurrecting dead actors? Already done… some commercials for example. And from the same company of Emily… Richard Burton for the great War Of The Worlds life concert. Check it out, they use a ‘hologram’ of Burton to narrate the story.
@LOGAN: OK. Thanks for the head’s up on that, I may check it out.
Someone was working on a Bruce Lee film a while back with a 3D character. They were hyping about how the Lee family agreed to do it, yada yada, but I never personally saw anything come from it except for a still that looked promising.
BTW its important to understand a few things about the project and the ‘performance capturing’ this company does. First they make a cyber scan of an actor using the rig they have. This rig most likely also makes shading references by projecting lights from all angles and this is captured on a file so they can virtually apply any lighting condition using the recorded lights.
Also they don’t have an automatic way of capturing the performance but the artists manually re-create a performance using the reference video. There’s no magic performance capturing software involved that the movie with the article suggests. The artists manually keyframe the animation for the performance, something is a bit lost in the article.
So people wishing to see ‘this’ in Blender by 2011, well, the animation tools are already there. The only thing they have is the equipment to create detailed models, something that’s hardware based.
I hope to see something in the area of performance capturing into blender though without a complete motion capture rig. Nowadays there’s software that can automatically track facial features, or think into the area of what ILM did with Pirates 3, using some suits with shapes that allow software to track the motion, it’s almost like those augmented reality tracking images. Maybe something like that is possible to get into blender one day.
Anyway, more related to blender and open source (that could possibly be put into blender) would be appriciated. Anyone know Open SOurce for facial tracking?
@LOGAN: Thanks for clearing that up. Yes, from the making-of video you linked here I saw the sphere with the projecting lights. I also noticed in the 3D app someone was using that he had the video right next to his 3D model so he could accurately mimic it. I thought the movement was automated, captured by the video somehow. If it was all animated by hand that is quite a bit more impressive, at least for the artist. An automated motion capture system that would capture her facial movements would be awesome and amazing as well, but on the coding and engineering side, not the artistic skill side.
Cheers!
-b
I really take issue with this, you will need real actors to get the motion capture to look realistic, and this in itself limits applications, why not just film the actor in the first place. As banor said above, if an animation system could easily and readily produce these results without a human actor it would be more impressive.
While there are things that this could be useful in, quite a few in fact, having a real actor there kind of begs the question “what’s the point, why didn’t you just film her in the first place.” A bit of “never mind the man behind the curtain” trickery.
That’s way more clever than I was excpeting. Thanks!