Phong13

Balancer is working a little better. Can actually find a balance pose now.

Working on a new balancing system. Not working yet but getting closer.

Got my exercise bike arduino project working. Pedaling is throttle, handlebars steer.

weeeeeeeeeee

(suspension physics in FUMES) 🎮

Testing the physically simulated walking system with a much larger heavier powered ragdoll. #gamedev

Getting some groove!

Rewrote the IK functions so they generate angular velocity for each limb as well as rotation. This angular velocity makes the Inverse Dynamics a lot more accurate. Still not perfect but the gait is faster and more natural.

Lol, a tap dancing bug!

Still looks a bit drunk but he isn't falling down anymore.

Some progress. Works for a few steps.

Trying a new approach to estimating target body part velocities... Still lots of work to do.

Testing the walking controller with different leg geometry. #gamedev

The dino rig is walking a little better now. The biggest change was shrinking the foot colliders. #gamedev

Started setting up a dinosaur powered ragdoll rig to see if the walk controller can be used with different leg geometry. First steps. Surprised that it works as well as it does without any modifications for the very different rig. #gamedev

Improved a bit over the week. The hardest part is the wrist joints. Without the marionette fake forces the wrists are driving the movement of the whole body. A tiny error messes up the whole pose. #gamedev

Climbing powered ragdoll doesn't use marionette forces anymore. It is now using Inverse Dynamics to control the rig. #gamedev

Peizhuo Li, Kfir Aberman, Zihan Zhang, Rana Hanocka, Olga Sorkine-Hornung. GANimator: Neural Motion Synthesis from a Single Sequence (SIGGRAPH 2022) Project page: https://peizhuoli.github.io/ganimator Code: https://github.com/PeizhuoLi/ganimator Abstract: We present GANimator, a generative model that learns to synthesize novel motions from a single, short motion sequence. GANimator generates motions that resemble the core elements of the original motion, while simultaneously synthesizing novel and diverse movements. Existing data-driven techniques for motion synthesis require a large motion dataset which contains the de- sired and specific skeletal structure. By contrast, GANimator only requires training on a single motion sequence, enabling novel motion synthesis for a variety of skeletal structures e.g., bipeds, quadropeds, hexapeds, and more. Our framework contains a series of generative and adversarial neural net- works, each responsible for generating motions in a specific frame rate. The framework progressively learns to synthesize motion from random noise, enabling hierarchical control over the generated motion content across vary- ing levels of detail. We show a number of applications, including crowd simulation, key-frame editing, style transfer, and interactive control, which all learn from a single input sequence. [SIGGRAPH 2022] GANimator: Neural Motion Synthesis from a Single Sequence

Yummm.

youtu.be/OV9VoHMEeyI

[SIGGRAPH 2022] Physics-based Character Controllers Using Conditional VAEs High-quality motion capture datasets are now publicly available, and researchers have used them to create kinematics-based controllers that can generate plausible and diverse human motions without conditioning on specific goals (i.e., a task-agnostic generative model). In this paper, we present an algorithm to build such controllers for physically simulated characters having many degrees of freedom. Our physics-based controllers are learned by using conditional VAEs, which can perform a variety of behaviors that are similar to motions in the training dataset. The controllers are robust enough to generate more than a few minutes of motion without conditioning on specific goals and to allow many complex downstream tasks to be solved efficiently. To show the effectiveness of our method, we demonstrate controllers learned from several different motion capture databases and use them to solve a number of downstream tasks that are challenging to learn controllers that generate natural-looking motions from scratch. We also perform ablation studies to demonstrate the importance of the elements of the algorithm. https://sites.google.com/view/jungdam/publications [SIGGRAPH 2022] Physics-based Character Controllers Using Conditional VAEs

More bedtime reading:

youtu.be/6vZkzVvHUzg

Physically simulated walker now works on accelerating platforms. #gamedevelopment

Back from holiday and working on the sword swing. Integrated it with the full character controller... sort of, still a few kinks. #gamedev

Applying the sword swing IK to a skinned character. Still a few glitches. I think I need to include the shoulder and chest in the IK. He doesn't look committed to this. #gamedevelopment

Started working on a sword swing IK rig. #gamedevelopment

He is running a little better. Having trouble keeping the gait even though. Feels like each step has slightly different start conditions resulting in a lopsided gait and uneven steps. Not sure how to deal with this. #gamedevelopment

He can run! ... sort of. Pretty awkward but it is a start. Experimenting with shortening and extending the standing leg length through the step and boosting the standing leg strength. #gamedevelopment

He is getting some groove! Added a function to vary leg length through the walk cycle. Also switches from 2-bone IK to 3-bone IK partway through the step. This shifts ground hinge point from heel to toe. #gamedevelopment

The HipHeight gait parameter can be used to jump!

Added a 'hip height' gait parameter that controls the target hip height for the physically simulated walking system. #gamedevelopment

Added an attack animation to see if he can do actions and walk at the same time without falling down. Pretty clumsy but works better than I thought it would for an initial effort. #gamedev

Working better, can walk forward backward and turn well now. However there is still a problem where he gets stuck with his feet crossed or wide apart and can't get them back to normal.

Finally the found the problem. It was the shoulders after all. Much more balanced now.