promising !
his that anyway connected with CAT ?
or totally self-Working ?

The last one looks cool It would be great if eventually it is an open system that can be used in a lot of contexts, aside from the viz one. Having a good crowd tool could help VFX work a lot, for the tasks that aren't worth going to Massive for.

that's looking so promising...Hats off for all of you working on these technologies..too bad there is no rss email subscription. Area should integrate something to let us updated.

zombies.

I do believe we'll tackle the VFX problem at some point, but, we're trying to first do something no else has done: believable pedestrians that are painted into your space and walk/interact like you'd expect pedestrians to behave. We've seen all the other systems and they all seem to break down when presented with path conflicts. It's fine if you're sweeping the camera through the scene so fast you can't spot the artifacts, but what if you want a lingering shot? I don't know if we can solve this, but we seem closer than any other system we're aware of. It's all driven by an "ocean-of-motion" that can be key-framed or mocap. The core of the effort is generalized, but the behavioral layer right now is custom, but modifiable. Our vision includes selecting a single character and "directing" them as part of this.

Yes, they look like "zombies" right now, but I'm not showing you the final set of capabilities (even for the Lab Release).

for some reason all I can think of is how often this will crash when you release it.

that's really positive, thanks for taking the time to share it

This is going to be AWESOME for zombie movies. Unfortunately by the time it will be released and hotfixed enough to work properly, there will already be so many zombie movies out there people will be tired of them already :/

hi!
what is the technical positionning comparing to Miarmy?

looks cool ken, would be cool to see maybe some vids of the crowds running and walking/running on some uneven surfaces.

nice... notice me when SP3 of Geppetto is out and it will not crash every 10 minutes.
the way we know autodesk it will be a pain to use in it´s first iterations because AD tend to releases unfinished stuff.

and if AD likes to get some reputation back.... i would focus on making things WORK when you sell software to your customers... not 3 month after the release.

there is no need to complain about comments like jonas, when AD constanty messes up software releases.
the 3d max 2012 release was a shame, after all the hype AD has created around nitrous.

I'd love to see some more challenging avoidance, e.g. escalators or avoiding traffic.

I'm guessing that it's not within the scope of it to have people getting ragdolled by cars, but having the crowd solvers drive vehicles so that they stop or avoid pedestrians would be cool.

The one slight thing that bothers me on one of the video is that the avoidance can be too perfect, almost like they're a ground-based aerobatic display team who know exactly what speed to hit to just avoid the others. Something more chaotic would look more natural, but I guess that will come.

Thanks,

Steve

Ken. Pull your coders off this and have them fix Max.

I don't really see 'avoidance' here. Just planned timing. It's cute but nothing groundbreaking.

You guys have a mess called CAT and you're working on this?

Ken. Have you ever really had to use 3dsmax daily to pay the rent?

Nice, the lane changing now works in the first video

How much control will we have over a character? I assume the purpose of Gepetto is to simulate crowds but can I directly control a character as well, as in the game 'The Sims'?

Lets say I want to give the viewer a guided tour by a virtual tour guide.. It would be brilliant if I could make that character walk where I want, force certain actions or poses and let Geppetto figure out how to animate, layer the motion clips and walk from a to b.

And is there going to be a beta test program for this?

Re: zombies

@Ken
I was just joking, but yeah, zombies. Dont expect the bipeds to actually interact with each other, but would be nice if so ie. Like in Grand Theft Auto ( or what most real humans do in SoCal much less Detroit), drinking, fighting, fornicating, etc you get the picture.

Good stuff though.... (actually at time of v2011, thought v2012 would have this; but obviously seems a SAP only feature for 2012) BUT have to agree with Jona vark on the current state of 2012.

Nitrous is a awesome one of a kind milestone... But QC really needs a relentless, unforgiving a$$ beating.

It would be interesting to see some samples that has larger social distances (some of these people make sport of almost colliding with others). See here for some general guidelines: http://changingminds.org/techniques/body/social_distance.htm

Also, I hope it will be easy to add more complex head motions. For example, people should look at one another occasionally when they are close, and also look in the direction of a turn a little while before making the turn. It's a bit robotic in that aspect at the moment

Love the stuff that has come out of Labs so far, nice to see the fairly expensive subscriptions leading to software we can test. This looks great!

Hi Ken I don't know if you can say or not at the moment but, are we seeing more than just avoidance and paint here? I know you all are hard at work on the fruits of the Biped/CAT system, is this update related to that, part of that? If so rigs that understand other rigs would really a coup. A lot of the "biped like" systems seem to set up off of the idea of 1 character in a seen and stop at that. Crazy as it may be I think it would be great if it's not already in if one character rig could use/understand the info of another. Example more than just avoidance/interception but simple things like trying to maintain eye contact in a way that can be messaged and not just an object to object link.
Ah well pie in the sky. Looks good to me.

Head-turning/gaze focus is on the agenda, but not in the Labs release. There will be some orientation stuff for idle behaviors.

The first phase is "orchestration" allowing you to direct the crowd. The second phase is "direction" allowing you to pick a performer and direct them to do things. Layered animation is part of the research. Select someone and have them wave, etc.

We are planning the Labs release by the end of September. We're bug fixing/stabilizing it right now. Lot's of activity. We've gotten some great feedback from beta testers.

Look great. Is there any news on when this exciting project will be released as we are now into October...

Sorry! We found some problems we wanted to fix before we release. The new target date is October 10th. We plan on making some noise about it, so you shouldn't miss it.