I’m good at Python, so I don’t mind using a bit of programming if-necessary to accomplish this…
I want a character to be able to “pick up, point and carry” an object. If the object is heavy, it pulls down on his arm and he needs to “shift his weight” to adjust.
In other words, I’m looking for somehow-physics-based smart animation, whereby the computer calculates realistic motions without requiring me to laboriously create them using stop-action.
I want to say, “Igor, go pick up that lead weight, carry it over to the table and put it down.”
Wow.
Coding that sounds more laborious than hand animating it.
I recently read about a large company doing something similar. I can’t quite remember the name. The point is that this would be cutting edge stuff.
Good luck!
With all of the “physics engine” stuff that we have now, it might not be so difficult as it may seem. We already have a very well-developed actions-based NLA system, collision detection and so-on. All of which could be used.
You see, for me, keyframing is very difficult, and anytime something must be changed (however slightly), a lot of work has to be done all over again. For the kind of work I am doing (museum displays…), being able to direct an “actor” (“hit this mark in five seconds”) would be a huge win. Enabling the computer to generate plausible movements would be a win. I see things like “ragdoll physics” and wonder. My Python-coding fingers start to itch…
Physics sims are very good for stuff that’s 100% physics: water, cloth, rubber can be simulated as it just reacts to gravity and the world around it. If the cloth isn’t just reacting to gravity but also moving deliberately on its own accord you run into problems. That’s why rag-dolls work. They person is basically unconscious and their legs are just flopping. People and animals are more complex: they will lean into a strong wind to maintain balance. If they are picking up a heavy object they may try to lift with their legs if they are worried about hearting thier backs. Can you code THAT in python? some people are working on that kind of code somewhere.
my personal aprotch would be to animate a walk sequance, and a walking-carrrying-heavy-object sequence. Search for reference diagrams for both. As for picking stuff up and putting it down? Oh - I’d just not show that particular bit.
Don’t get me wrong…
If you can pull it off, good luck to you!
As I mentioned, there’s a recent technology that does all of this and more. I can’t find the link (or remember the company name), but it was posted on BlenderNation about 6 (?) or so months ago.
The problem with calculated human motion simulation is that it is just that – calculated. It can act fully only within the coded parameters, so it tends to look robotic unless some hugely in-depth physiological as well as physical simulation algorithms are developed. How do you code spontaneity, the occasional mis-balance that people correct for reflexively all the time, the small motions that occupy all moments of human activity? Without the nuances, human motion tends to look right of Uncanny Valley central casting. If you can write something like that, why are you doing museum displays? :eek:
for me, keyframing is very difficult, and anytime something must be changed (however slightly), a lot of work has to be done all over again.
I found its easier to focus on the basics (such as keyframing) and then build upon what I’ve learned (for more complex things)
You might want to try putting some time into the “basics” (keyframing) , and learn some easy tricks that make keyframing more enjoyable (for instance, having a keyframe on every 10 frames for all your bones will make it easier to go back and fine-tune your motions)
But yeah, I do agree that having a plug-n’-play keyframe engine would be friggin’ sweet! However, it would probably take many hours of dedication and research into different aspects of physics/ movement to get it to look right…
good luck!!!
