Page 2 of 3

Posted: Fri Dec 17, 2004 4:51 pm
by solmax
i love to hear your proposals. i proposed the slider thing some time ago, but without big response.

here is how i see the animation system:

1 - values (loc, rot, alpha..)
2 - animated values -> IPO's
3 - multiple IPO's ->strips (let users define which IPO's should go in!)
4 - multiple strips ->actions (not the current actions)

let's call everything above 2 high-level animation objects. each of them can be controlled by it's own time-IPO or a slider going from 0-100. this means, as soon as i have an animation defined by keyframes, i can keyframe it's state ("local" time) over the global scene timeline. this should be possible for every animation object (IPO, strip, action).

furthermore, we need some sort of general slider object (a new window-type for custom sliders would be great - either linked to a scene or to objects). any animatable parameter can be linked to a slider, and one slider can control multiple parameters/animation objects. this is a generalized driven key approach, or let's call it channel linking.

so what results from that? blender burns XSI, maya and max altogether. no, really, this would enable most complex animations setups for virtually any case. what is crucial is how fast state evaluation is (a slider controls 3 sctions, each consisting of multiple strips, each consisting of multiple IPO's) and how deep this hierarchy should go.

from the users point of view, for setting up such a system i'd like to see an aproach as in maya, or XSI, where the user works with keys (slider_1@0%: action_2 to 20%, action_3 to 0% as an initial state, and slider_1@100%: action_2 to 80% and action_3 to 90%, for example). to make it clear: we map a slider value to a time in the local timeline of an animation object. the slider itself is animated by it's own IPO in the global scene timeline.

additionally, i'd love to see better armatures, with lockable axes and axis-limits, both visualized right in the viewport. this would speed up IK-solving a lot, and make animation easier (as "forbidden" states are more or less avoided right from the beginning)

phew, i hope that was somhow clear.

any suggestions?

best regards


Posted: Mon Dec 20, 2004 11:03 pm
by ZanQdo
just to add a couple of ideas:

1- Effects NLA strips: prepare an effect, and then convert it to a strip so it doesn't play in linear time, just when and where you add it in NLA window

2- Meta strips in NLA: Well basicaly the same meta strips that the sequence editor has, allowing to group varius strips (armature, material, effects :P) all together so you can re-use complete complex animations just by duplicating an NLA Meta Strip, So simple and powerfull :D

3- RVK simple feature: When you are in any morph target you can copy other targets vertex positions to current target selected vertices. Example: select some vertices of a smile target and copy the base target to them, so you clear their positions, There are many possibilities :)

See you

Posted: Tue Dec 21, 2004 5:16 am
by harkyman
1 and 2 - since any IPOs (material, mesh, etc.) can be put into an Action strip, they will be able to appear in the NLA as a cohesive unit.

Posted: Wed Dec 22, 2004 2:44 pm
by malefico
What I would like to see implemented as workflow is the chance to do "layers" of character animation. So, for instance, a group of animators make first animation pass, setting up some basic keyframes, then a second group works over these adding acting to these basic keyframes and so on...
So you would end up with several NLA strips, each one representing these different layers of animation.

This is very difficult to achieve today, because once you start adding keyframes for your "second pass", the action overlaps any previously NLA strips, and you can only see what you've actuially did by converting action to strip which leads you to a very cumbersome workflow.

Hopefully with Harkyman's approach, this can be fixed... don't really know.

Posted: Thu Dec 23, 2004 1:02 am
by harkyman
This would work like a dream is my proposal makes it to implementation.

Posted: Thu Dec 23, 2004 1:58 am
by joeri
solmax wrote: additionally, i'd love to see better armatures, with lockable axes and axis-limits, both visualized right in the viewport. this would speed up IK-solving a lot, and make animation easier (as "forbidden" states are more or less avoided right from the beginning)
I like the Maya pole vector contraint alot better then locking axes or setting limits. I never set limits on my character animation in maya, there is no need 'cause maya does not bring the bones in the wrong spot. The pole vector can point to a different spot over time, so the bones never flip to the wrong side.
And for knee inward/outward just move the constraint, the pole will move, so will the bones of the knee. ...Hard to explain. Maybe add some pictures?

Posted: Thu Dec 23, 2004 3:25 am
by malefico
harkyman wrote:This would work like a dream is my proposal makes it to implementation.
That would reaaaallly please me :D

Since everybody is adding feature requests to your proposal... here goes mine:

I always wanted lattice and armature animation could work together (simultaneously) on character animation. For instance to add squash and stretch in a proper way to the head or stomach of a character while it's walking.

Maybe with this "NO POSE MODE" aproach, something could be worked out in this direction. For instance we have wave effects on top of armature available now... maybe we're not so far away...



Posted: Sun Jan 16, 2005 8:29 pm
by ton

Just to notify I am studying your proposals and find it higly useful :)
Apart from the IK/FK and constraint issues (which I feel very unfamiliar with still) all looks very feasible here. Stilll studying more feedback... will come back later to it.


Posted: Mon Jan 17, 2005 4:31 am
by harkyman
Awesome. I know you're working hard on this, and there is a lot of stuff out there. I tried to keep my proposals within what I thought was feasible, having banged my head against the character animation code for several months.

Posted: Thu Feb 17, 2005 3:52 pm
by wolfmanyoda
I love the idea of onion skinning. That would be so helpful.

Posted: Sat Feb 19, 2005 3:19 pm
by bullx
does somebody lnows how other apps behave when you have 2 interacting objects?
in example a tentacle that have to spin around an arm of a moving caracter, or more simple interactions like feet and floor, or bringing une object in one hand.
i feel lost everytime i need to animate something like this because my objects always intersecate them during the animation.

Posted: Sat Feb 19, 2005 3:47 pm
by rcreel
I think that anyone working on the new animation system should take a look at Animation:Master. I have found it easier to model with polygons than Hash patches, but most people agree that the animation system in A:M is excellent. I don't know how difficult the same sort of system would be to implement in Blender, but combining it with the speed and stability of Blender would be great.


rigging proposal

Posted: Sat Feb 26, 2005 8:52 pm
by anubis4d
This is a proposal for Rigging systems in blender.
I disagree with the full code rewriting ideas around this. I wrote about this in a PDF file, tell me what you think.


anubis4d, from Argentina 8)

Posted: Tue Aug 16, 2005 12:42 am
by tommy
Sorry for bumping this, but I just wanted to bring this thread back to people's attention. Especially in coherence with the recode for the Orange Project, harkyman's suggestions might of value.

Personally, I think those functions, especially the NLA stuff and Animation paths, will help n00b animators like myself to get into animating and will make workflow a lot more effective. I especially like the animation paths, which I think I requested a long long time ago, after seeing Project:Messiah for the first time a few years ago. I've tried making walk cycles a number of times and have always failed to produce anything that didn't look robotic and chunky, simply because I find it much more intuitive to see the motions on screen and to be able to rotate around them and perfect them from any angle.

Just my 2 cents.