Icon Unrolling Rotations


Icon Animation Blend Spaces without Triangulation


Icon Quaternion Weighted Average


Icon BVHView


Icon Dead Blending Node in Unreal Engine


Icon Propagating Velocities through Animation Systems


Icon Cubic Interpolation of Quaternions


Icon Dead Blending


Icon Perfect Tracking with Springs


Icon Creating Looping Animations from Motion Capture


Icon My Favourite Things


Icon Inertialization Transition Cost


Icon Scalar Velocity


Icon Tags, Ranges and Masks


Icon Fitting Code Driven Displacement


Icon atoi and Trillions of Whales


Icon SuperTrack: Motion Tracking for Physically Simulated Characters using Supervised Learning


Icon Joint Limits


Icon Code vs Data Driven Displacement


Icon Exponential Map, Angle Axis, and Angular Velocity


Icon Encoding Events for Neural Networks


Icon Visualizing Rotation Spaces


Icon Spring-It-On: The Game Developer's Spring-Roll-Call


Icon Interviewing Advice from the Other Side of the Table


Icon Saguaro


Icon Learned Motion Matching


Icon Why Can't I Reproduce Their Results?


Icon Latinendian vs Arabendian


Icon Machine Learning, Kolmogorov Complexity, and Squishy Bunnies


Icon Subspace Neural Physics: Fast Data-Driven Interactive Simulation


Icon Software for Rent


Icon Naraleian Caterpillars


Icon The Scientific Method is a Virus


Icon Local Minima, Saddle Points, and Plateaus


Icon Robust Solving of Optical Motion Capture Data by Denoising


Icon Simple Concurrency in Python


Icon The Software Thief


Icon ASCII : A Love Letter


Icon My Neural Network isn't working! What should I do?


Icon Phase-Functioned Neural Networks for Character Control


Icon 17 Line Markov Chain


Icon 14 Character Random Number Generator


Icon Simple Two Joint IK


Icon Generating Icons with Pixel Sorting


Icon Neural Network Ambient Occlusion


Icon Three Short Stories about the East Coast Main Line


Icon The New Alphabet


Icon "The Color Munifni Exists"


Icon A Deep Learning Framework For Character Motion Synthesis and Editing


Icon The Halting Problem and The Moral Arbitrator


Icon The Witness


Icon Four Seasons Crisp Omelette


Icon At the Bottom of the Elevator


Icon Tracing Functions in Python


Icon Still Things and Moving Things


Icon water.cpp


Icon Making Poetry in Piet


Icon Learning Motion Manifolds with Convolutional Autoencoders


Icon Learning an Inverse Rig Mapping for Character Animation


Icon Infinity Doesn't Exist


Icon Polyconf


Icon Raleigh


Icon The Skagerrak


Icon Printing a Stack Trace with MinGW


Icon The Border Pines


Icon You could have invented Parser Combinators


Icon Ready for the Fight


Icon Earthbound


Icon Turing Drawings


Icon Lost Child Announcement


Icon Shelter


Icon Data Science, how hard can it be?


Icon Denki Furo


Icon In Defence of the Unitype


Icon Maya Velocity Node


Icon Sandy Denny


Icon What type of Machine is the C Preprocessor?


Icon Which AI is more human?


Icon Gone Home


Icon Thoughts on Japan


Icon Can Computers Think?


Icon Counting Sheep & Infinity


Icon How Nature Builds Computers


Icon Painkillers


Icon Correct Box Sphere Intersection


Icon Avoiding Shader Conditionals


Icon Writing Portable OpenGL


Icon The Only Cable Car in Ireland


Icon Is the C Preprocessor Turing Complete?


Icon The aesthetics of code


Icon Issues with SDL on iOS and Android


Icon How I learned to stop worrying and love statistics


Icon PyMark


Icon AutoC Tools


Icon Scripting xNormal with Python


Icon Six Myths About Ray Tracing


Icon The Web Giants Will Fall


Icon PyAutoC


Icon The Pirate Song


Icon Dear Esther


Icon Unsharp Anti Aliasing


Icon The First Boy


Icon Parallel programming isn't hard, optimisation is.


Icon Skyrim


Icon Recognizing a language is solving a problem


Icon Could an animal learn to program?




Icon Pure Depth SSAO


Icon Synchronized in Python


Icon 3d Printing


Icon Real Time Graphics is Virtual Reality


Icon Painting Style Renderer


Icon A very hard problem


Icon Indie Development vs Modding


Icon Corange


Icon 3ds Max PLY Exporter


Icon A Case for the Technical Artist


Icon Enums


Icon Scorpions have won evolution


Icon Dirt and Ashes


Icon Lazy Python


Icon Subdivision Modelling


Icon The Owl


Icon Mouse Traps


Icon Updated Art Reel


Icon Tech Reel


Icon Graphics Aren't the Enemy


Icon On Being A Games Artist


Icon The Bluebird


Icon Everything2


Icon Duck Engine


Icon Boarding Preview


Icon Sailing Preview


Icon Exodus Village Flyover


Icon Art Reel




Icon One Cat Just Leads To Another

Dead Blending Node in Unreal Engine

Created on Aug. 13, 2023, 10:37 a.m.

Unreal Engine 5.3 comes with a new experimental "Dead Blending" node that acts as a drop-in replacement to the Inertialization node.

In general I've found it creates better looking blends than the standard Inertialization node, in particular when you have transitions between poses that are very different:

Or when you are triggering transitions very frequently:

However you may find that sometimes it tends to damp out the motion a bit more than the normal Inertialization node.

The way it works is a similar to the basic exponential decay method I described in my previous article on Dead Blending, with a few small differences.

First, instead of updating the extrapolated pose each frame we just store the pose at the point of transition and use an implicit extrapolation function to get the extrapolated transform of the joint at an arbitrary time in the future past the point of transition (given some half-life used to control the rate of decay of the velocity).

Here is some similar extrapolation code adapted for quaternions and allowing an individual half-life per rotation axis:

void extrapolate_decay_rotation(
    quat& x,
    vec3& v,
    vec3 halflife,
    float dt,
    float eps = 1e-8f)
    vec3 y = LN2f / (halflife + eps);
    x = quat_mul(quat_from_scaled_angle_axis(
        (v / (y + eps)) * (1.0f - fast_negexp(y * dt))), x);
    v = v * fast_negexp(y * dt);

Although it doesn't make much difference to the final result, this feels like a more elegant solution to me since it reduces the amount of state being mutated each frame and more closely resembles the setup for normal Inertialization.

The big difference in this implementation comes from having individual half-lives per degree of freedom for each joint of the character. This allows us to safely extrapolate joints on axes that we are sure can move far, while limiting the extrapolation for joints on axes that we are not certain can move.

These half-lives aren't set by hand. Instead we use a simple heuristic to compute them. At the point of transition we look at the source and destination animation and determine which joint axes are moving toward the destination animation and at what rate. Axes which are moving toward the destination animation with a low velocity can be extrapolated for a long time, while axes that are moving with a fast velocity, or away from the destination animation we want to give a short half-life.

In code it looks something like this:

float max_mag(float x, float eps)
        x >= 0.0f && x <  eps ?  eps :
        x <  0.0f && x > -eps ? -eps : x;

vec3 max_mag(vec3 x, float eps)
    return vec3(
        max_mag(x.x, eps),
        max_mag(x.y, eps),
        max_mag(x.z, eps));

vec3 estimate_halflife(
    vec3 src_dst_diff,
    vec3 src_vel,
    float halflife_scale = 0.3f,
    float halflife_min = 0.1f,
    float halflife_max = 1.0f,
    float eps = 1e-8f)
    return clamp(
        halflife_scale * (src_dst_diff / max_mag(src_vel, eps)), 

Here src_dst_diff is the difference between source and destination animations (rotations are encoded as scaled-angle-axis), src_vel is the velocity (or angular velocity) of the source animation, halflife_scale controls the overall expected scale of the half-lives, and halflife_min and halflife_max control the overall min and max half-life to allow.

If we visualize the extrapolations we can see how this heuristic affects them - some joints are allowed to extrapolate for a long time while others are brought to a halt almost instantly. In this way we almost always avoid violating joint limits while maintaining long extrapolations where we can.

This is still an experimental node, so the implementation and interface are subject to change, but I hope it will provide useful to developers who are having a bit of trouble getting good results out of the normal Inertialization node.

And if you'd like to try it out yourself I've added the same heuristic as an option in my previous article's demo code.

github twitter rss