[Bf-committers] call for coder 1. integrate BioMotionLab walk cycle 2. go deeper in motion analysis

bjornmose bjornmose at gmx.net
Mon Feb 4 04:12:49 CET 2019


There are 2 projects I'd like to suggest


Make user interface for the python script 


Dive deeper into motion analysis, mathematics and explore data 
reduction, recovering relevant data. Mayor focus is on 'relevant'. In 
Video / Audio streaming these methods are common and the elder of us 
know a device called equalizer .. some knowledge from there can be used 
to modify animation data .. hum too much .. but ask if you think  you'd 
like to know more


I see! Motion capture disappeared a bit from my radar... it's really 
best if you try one of the dev channels to get people involved. That can 
be bf-committers or the new devtalk.blender.org forums!



On 03/02/2019 14:36, bjornmose wrote:
> Hi Ton,
> They (BioMotionLab) created a walk-cycle-generator based on captured 
> data. Their approach is controlling the cycle by intuitive parameters 
> like gender, weight, mood, excitement . Joeri and Bart pointed at the 
> page a few years ago. Now recently there was a java script on the BML 
> page and I grabbed, reverse engineered .. ported things to python. 
> Some more clumsy code generates the actions for the primitives in 
> blender. Having a proof of concept I did ask Mr. Troje for permission 
> using the data. A few days ago he agreed, if proper credits are given. 
> So for now I published a .blend and the .py script with working code 
> but no UI at blenderartists.
> https://blenderartists.org/t/biomotionlab-data-working/1144168
> python script is also there
> https://github.com/bjornmose/BioMotionLab2Blender
> I think this could be a nice addition to the animation system but 
> needs far better integration. Now, the basic work is done and my 
> schedule is too tight to keep working on it I could imagine some 
> talented coder might want to bring it to full usability.
> Further more I can offer support to a even more talented coder to 
> create similar data for other motion patterns. That is: reducing a set 
> of captured motion data to a few basic 'poses' and mixing them back to 
> a new animation using parameters like 'cool it' or 'weight', as the 
> papers cited below are fully in the scope of my mathematical 
> understanding. A possible tool using the same mathematics could be a 
> 'why does this animation look wrong?'-assistant which would be working 
> on the already existing path tool.
> -Jens-

More information about the Bf-committers mailing list