The Face Machine

[NolanSW]

Oooo, first post. Awesome Raf. Damn you and your smartness. hehe. The video is messed up though. Could you re-post it? Looks cool though. Can't wait to check it out.

-Sean

[rafanzovin]

Changed the link to be a smaller QT version...this should work better.

--Raf

[thematt]

looks really awesome, i've develloped something in the tone of that, although yours is much better hehe..I wonder how you managed to have the double transform on thecontroler so pulling one, influence the other..also it's pretty cool they don't go all over the place like it's the case in my setup, it hard to constraint it only to the face.

looking forward to it though.I f you need beta tester..i'm your man..hehe.(on ca nalways try right).:)

Anyway great job

[Firas]

looks very promessing... please show the setup steps .. i understand its wip process .. but to know how complecated it is.

thank u.

[yolao]

To say it in one word... "AMAZING", this looks truly fantastic, i hope it will be released soon.
You mention that you can not show the interface because you still working on that, but it would be nice after you work more on the interface to have a video that shows the process and steps to rig the face...

Again, it looks wonderful:thumbsup:

[kamsvag]

Brilliant!

[digones]

Hi Raf

I've been using TSM 2.0 here at the studio for almost a year, and it's a must!

can't wait to put my hands on this one, if you're in need of a beta tester, drop me an email ;)

cheers and good luck with the new product

[captainboom]

Off the hook, I am huge user of TSM 2.0. I have used it on many projects and freelance work. and it has not let me down yet...I am looking forward to adding The Face Machine, to my tool set.

Any idea when you will be releasing it?


Later
Jason

[rafanzovin]

Thanks! We're hoping to go to beta in a couple of weeks, so the most likely release date will be begining of April.

--Raf

[yolao]

Thanks! We're hoping to go to beta in a couple of weeks, so the most likely release date will be begining of April.

--Raf

a release in a couple of months...that`s great:).

I`m sure it will have a huge success just as TSM2 has.:thumbsup:

-Does face machine will have the option to use blendshapes?.... to have a combination of joint based deformations and blendshapes.

[rafanzovin]

-Does face machine will have the option to use blendshapes?.... to have a combination of joint based deformations and blendshapes.

No, the Face Machine doesn't use Blendshapes at all. It uses some joints but it's mostly deformed directly by a set of curves.

--Raf

[yolao]

No, the Face Machine doesn't use Blendshapes at all. It uses some joints but it's mostly deformed directly by a set of curves.

--Raf

thanks for the answer.

I was asking about blendshapes because it will be nice to also have the option to sculpt the shapes for some expresions...

So you can sculpt some shapes on top of the default deformation that FACE MACHINE does automatically ( wich looking at the video are GREAT)

Would face machine have the option to sculpt the shapes, so when you move a locator you can have some specific shape at that time.. like FACEROBOT does.?

thanks... face machine will be GREAT:thumbsup:

[thematt]

hello raf,
Like I said in a previous post the setup looks great.Although if you're using curve that are skin as infleunce object, make sure you test it maya 8.5 as there is an issue with it once they are setup with controller..where before there wasn't any double transform and you could move them freely with the rig now a double transform occurs.I don't know if you're using that but just in case i warn you.

Also if you've already found a solution to that problem please share as we are a few that are stuggling with that issue just now.

Anyway once again great job, hope I can make my hand on the product one day.

best of luck

matt

[rafanzovin]

You could use driven keys fairly easily to add blendshapes on top. We've done a couple of tests with 8.5 but we haven't encountered any issues with using curves as influences yet. For future reference, though, do you have any more info on the problem you're describing?

--Raf

[thematt]

no problem seriously..oouch I've try a bunch of solution and none work.As soon as I add some cluster and controler to the curve that is influence I can't move the parent joint anymore without having a double transform.
before up until 8.0 you could parent everthing under your top parent joint or node and move that around with the curve following, now on 8.5 there is an issue with the curve giving a double transform and I can't find a way to correct that.

maybe you could send a simple scene that work for you also look at this thread to see the problem I'm talking about.If you find a solution that would be very helpfull actually thanks.

http://forums.cgsociety.org/showthread.php?f=7&t=455632

Cheers

[Buexe]

Double transforms can usually be fixed by turning off "inheritTransform" on one of the related nodes in the Attribute Editor.
Hope that helps

[Ramteen]

Repeated .Sorry .

[Ramteen]

Wow .Great . I can`t wait to see it . amazing stuff . but I have some question . I wanna know if you guys have used any sort of expressions . Or it`s just only constrains and ifluence objects ? :-?
thanks very much .

[rafanzovin]

Wow .Great . I can`t wait to see it . amazing stuff . but I have some question . I wanna know if you guys have used any sort of expressions . Or it`s just only constrains and ifluence objects ? :-?
thanks very much .

Nope, no expressions.

No clusters either, which is probably why we're not getting the double-transform problem. We're moving the curves around with direct connections into each control point.

--Raf

[thematt]

Nope, no expressions.

No clusters either, which is probably why we're not getting the double-transform problem. We're moving the curves around with direct connections into each control point.

So that's the trick right.:) I wonder how that work.
thanks for the info

cheers

[Juss]

This is really good~ Oh, Raf, if only you could provide the scene file~
Very useful technique for facial rigging. Thanks for sharing! :thumbsup:

[karabo]

this is awesome, when is it coming out? How much will it cost?

[rafanzovin]

We're hoping to go beta this week. After that, we'll see how many bugs show up. :)

The price is still being determined.

--Raf

[thematt]

I'd be happy to beta test if possible .. thanks for the good work

cheers

[karabo]

I'd be happy to beta test if possible .. thanks for the good work

cheers

that goes for me too. If you are looking for beta testers, I'd be more than happy to volunteer

[jason108]

Sweet!

Can you include a tutorial on how to model a good face for the face machine to work well too?

thanks

[VM]

fantastic stuff!! did you guys started beta testing the beast? can I join in?

[Wilson-3d]

Sent you a PM rafanzovin

Looks exciting!

Cheers

[yolao]

hey Raf, can you please, when you have some time, make a video of how you setup a face with The Face Machine.

Thanks

[VM]

yeap, a video or a pdf on how to setup a face w face machine would be great!!

[rafanzovin]

Soon, I promise. We want to get all the interface elements working first so you can really see how it works.

--Raf

hey Raf, can you please, when you have some time, make a video of how you setup a face with The Face Machine.

Thanks

[yolao]

Soon, I promise. We want to get all the interface elements working first so you can really see how it works.

--Raf

thanks Raf, i will be waiting for the video and the release of The Face Machine.

[VM]

I hope it will not be as slow as it appears in the demo video. Doing lipsynch with a crawling rig is not practical; I have no idea how animators deal with Face Robot for example. TSM2 is also a slow rig, and I guess it's because of the large number of nodes and connections, but.... I don't really know for sure (and I mean the rig itself, with no geometry attached, is slow, far from realtime).

I also hope you will allow people some flexibility with that setup. Your setup machine for example has those widgets, and I for one have never managed to figure out how to position them precisely, so I get a predictable result when you turn them into joints. I think a serious rig should allow us to create the joints ourselves, as many as we need to, not a default number and structure, and also, of course, give us the ability to position those joints exactly where needed. Just a little bit off and the deformations could very well get messed up. My workaround in this case was that I had to connect the real skeleton to the TSM2 skeleton, which is an absurd solution, but I couldn't think of anything else.

[rafanzovin]

One of the main things we've been working on in the months since the first video was released was heavy optimization of the rig--it performs many times faster now (this will be clear once the new videos are released). We're also working on optimizing TSM rigs for TSM3. The key, in both cases, has been to write our own nodes that perform the functions of a long chain of Maya nodes, but hundreds of times faster.

If you want to see the joints directly in TSM2 while positioning them, it's not that hard to do--look under the main widget control and you'll see a hidden node called "JOINTS." Unhide it and you can see the joints being positioned by the widgets in real time.

You can't, however, change the number of joints. That's part of the nature of TSM2 as a "black box" solution--you give up a certain amount of control in return for ease of use. At the moment, we're begining work on a companion tool to TSM, tentatively entitled "Anzovin Rig Tools," that is based on the same technology as TSM3 but is not a black-box tool--it will, if things go as planned, offer full control of the rigging process around your own joints, up to the point of being fully mel-scriptable itself.

I hope it will not be as slow as it appears in the demo video. Doing lipsynch with a crawling rig is not practical; I have no idea how animators deal with Face Robot for example. TSM2 is also a slow rig, and I guess it's because of the large number of nodes and connections, but.... I don't really know for sure (and I mean the rig itself, with no geometry attached, is slow, far from realtime).

I also hope you will allow people some flexibility with that setup. Your setup machine for example has those widgets, and I for one have never managed to figure out how to position them precisely, so I get a predictable result when you turn them into joints. I think a serious rig should allow us to create the joints ourselves, as many as we need to, not a default number and structure, and also, of course, give us the ability to position those joints exactly where needed. Just a little bit off and the deformations could very well get messed up. My workaround in this case was that I had to connect the real skeleton to the TSM2 skeleton, which is an absurd solution, but I couldn't think of anything else.

[VM]

that's great news. in fact, I was exagerating when I said TSM2 is slow, sorry for that. it's not a slow rig, it's just not realtime, but that's... I think... more of a Maya problem. most things are not trully realtime in Maya (or is it just my machine), and probably geting a body rig to perform real-time is maybe impossible... (?)

but the facial rig does look slow in the videos, I'm happy to hear you've optimised it.

anyway, thanks for the insight on how to deal with joints!

[rafanzovin]

that's great news. in fact, I was exagerating when I said TSM2 is slow, sorry for that. it's not a slow rig, it's just not realtime, but that's... I think... more of a Maya problem. most things are not trully realtime in Maya (or is it just my machine), and probably geting a body rig to perform real-time is maybe impossible... (?)


It's not impossible, but your rig has to be pretty stripped-down--you can't have a rig with TSM's comprehensive stretchy/deformable tools without taking a significent performance hit--at least, not if you're using standard Maya nodes to do it.

--Raf

[bazuka]

when can we try face machine? :)

[splintah]

what can i say

i am a fan !

so where do i sign for that ?
any news on release dates of face machine and tsm3


on a sidenote
is tsm3 able to export as fbx or into game engines ?

[JYoung]

I'm also curious to know when face machine will be released. :

[rafanzovin]

As you all may have guessed, initial projections of release were widley off. :) At the moment, the Face Machine is being used internally for production, and we've been ready for a while to send it to a few beta testers and post the new movies--we just haven't had a break from our production work in which to do so! We'll be meeting a large deadline next week, and after that I hope we'll have a chance to start packaging it for release.

--Raf

[doodlyD]

great news !!
keep up the good work

[JoseLuis3D]

aaaaaand? any news on face machine? we're almost july....maybe for siggraph?

[VM]

Nope, no expressions.

No clusters either, which is probably why we're not getting the double-transform problem. We're moving the curves around with direct connections into each control point.

--Raf

Wowowowow, thanks rafanzovin!!
This is great tip, such a simple and elegant solution (or is there a dark side I don't know about...?)
Tutorial (hahaha, this is what I found for now, very simple stuff):
1.Load a curve's shape node in the Connection Editor / Inputs and look down for Control Points.
2.Load, say... a locator (the controller) in the Outputs.
3.Connect the locator's Translate to one of the curve's Control Points.
Now the locator drives that curve's control point. Works with both CV and EP curves ;) Sooo simple..... For a head, group the locators, and curve/s (wires) together, so that they move along with the head.

[rafanzovin]

Wowowowow, thanks rafanzovin!!
This is great tip, such a simple and elegant solution (or is there a dark side I don't know about...?)

Thanks! No dark side that we've encountered yet...the main reason to do this for us is as part of our rig optimization--there's so much internal structure to the face machine that unless every part of it operates as efficiently as possible it gets to be way too slow, and cluster turn out to be very ineficient for this purpose.

We're working on getting something you can see for the face machine soon, once again internal production has gotten in the way. The good news is that once it's actually released it will have been used so much in-house that it should be rock-solid.

--Raf

[brendo2026]

Amazing stuff, this is great work, is it going to be a free plugin?:)

[JYoung]

I'm going to take a wild guess and say no. If the setup machine is any indication, I'm sure it'll be priced fairly for what it does.

[rafanzovin]

We haven't decided on a final price yet--it may be a little more then TSM but we're still aiming for something with a reasonable price range that would be affordable to students and freelancers.

--Raf

[thematt]

Hey raf, already said but great work..
question when is it going to be available? I'll soon be going into a production that would greatly benefits from such a pluggin or develeppement, so I sure hope it's soon, it would save me tons of time.
Also is it interfaced as something like Face robot, or more in the spirit TSM, with multiple script that you use together to get to something?

Anyway, waiting impatiently.

cheers

[rafanzovin]

Hey raf, already said but great work..
question when is it going to be available? I'll soon be going into a production that would greatly benefits from such a pluggin or develeppement, so I sure hope it's soon, it would save me tons of time.
Also is it interfaced as something like Face robot, or more in the spirit TSM, with multiple script that you use together to get to something?

Anyway, waiting impatiently.

cheers

Unforunately, we don't have a firm release date yet--the product works but we're still working out all the bugs: we really want this to be solid before release.

Although we investigated the idea of using multiple widgets like TSM while TFM was still on the drawing board, it would have been prohibitively time-consuming to make that work, so we just have one widget that fits an at least close to humanoid face (it can stretch a lot to do cartoony faces but won't work well for dogs or dinosaurs). We're still investigating the possibility of trying to make individual face element widgets if the face machine does well, it would just take a lot longer to develop.

--Raf

[thematt]

ok too bad there is no planning on releasing that soon, but I always apreciate when you buy something that is rock solid.
So I'll wait.
thanks for clearing the widget question up.
cheers

[splintah]

any news on releasing that thing ?

waiting for it from day one

[rafanzovin]

Unforunately, Maya 8.5 SP1 and possibly Maya 2008 as well break the Face Machine at a fundamental level (basically, it's now impossible to do component-level influences in a skin cluster with mutltiple types of influence without risking a crash). We've been talking to Autodesk about it but it looks like we'll have to code around it, most likely by coding our own form of skin cluster. We're in the process of doing that, but it will take some time before we can make a release estimate again.

The good news is, now that we've been forced to examine some of the basic assumptions about how the Face Machine is put together and make some alternate plans, we've found some ways to optimize it even more and make it significently faster, so when it finally is released it will be a better product.

--Raf

[munehausen]

still waiting for testing, keep develop it

[JYoung]

It's pretty damn sweet. Everything was really well thought out... nice and simple controls and setup, and great deformations.

[JoseLuis3D]

It's pretty damn sweet. Everything was really well thought out... nice and simple controls and setup, and great deformations.
What??? is there a demo or something????? Is it already released?????

[splintah]

any news ?

waiting waiting

[rafanzovin]

any news ?

waiting waiting

You might want to take a look at the "New Face Machien videos" thread.

--Raf

[brendo2026]

Hi guys the demo video's are amazing!

I was interested to see that Autodesk bought out the cMuscle system and that sounds like it would be using similar muscle influence to what you are working on, I'm interested to see if they will intergrate it into maya also, or if you have to buy it as a seperate plugin.

I also wanted to know, of course you will want to reap the benefits of all your hard work by selling your product, but in the process of working out all the issues, will you allow a demo version of your product with capped limitations like cMusle?
and if you have beta testers are there any projects out there were I can see your work in action apart from in house?

This is truley great work, keep it up and I wait with great enthusiasm for a public release

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.