01
Model & rig
Editable Blender files connect the body, joints and skin weights. Earlier states remain available for comparison.
From my workshop
NULL // RELOAD
When does a character become a believable movement?
A browser game needs more than an idea for its world. Right now I am working on the characters behind it: bodies, joints, skin and movement that still hold up under close inspection.
In development · as of 6 October 2026Running is not the same as looking right.
A knee bends unnaturally. A hand loses its shape. These failures make the work interesting: where does the difference between the model in Blender and the image in the browser arise?
02
Motion sample
Selected movements and individual poses are exported. One successful pose is not an approved animation.
03
The same viewpoint
Before and after are compared with the same camera, including slow motion and a visible skeleton.
04
Correct & test again
Changes are checked against the same sample. Visible defects and unsuccessful attempts remain part of the result.
In development · as of 6 October 2026
The workshop is ahead of integration.
The project documents a separate character and animation workbench with A/B comparisons. This does not automatically put the new character into the main game. Anatomical defects, partner combat and game integration remain open.
A personal development project, not scientific validation. The planned campaign and autonomous game AI are not completed features.
The next test
Can a corrected character perform the same movement from several viewpoints without introducing a new problem elsewhere? That is the comparison I want to show, not just a beautiful single frame.
Research, separate from the game