TrendWhat rose on GitHub, every day at 07:30 KST · 한국어

GitHub · as of November 15, 2026 · View on GitHub →

Friedrich-M/UniMate

3D character animation model

+104 stars in 7 days

Generate natural motion for any skeleton from a single text prompt

Game characters, robots, animals, and even flowers. Enter a text prompt and the system generates natural motion in real time that fits the specific skeleton. There is no need to train or optimize separately for each skeleton; a single model understands various structures and creates the movement. Recently, a feature was added that allows users to directly animate rigged 3D assets they bring in, broadening its range of application.

A dataset unifying diverse skeletons

The UniML3D dataset, which forms the basis of this model, contains 7,430 distinct skeleton structures. It covers a wide range of subjects, including bipedal, quadrupedal, avian, marine life, insects, snakes, and articulated rigid bodies, all organized in a standardized format. Comprising 13,769 text-motion pair clips, it enables the model to learn general movement patterns without bias toward specific skeletons. The data processing pipeline automatically handles everything from downloading original assets to feature extraction, captioning, and joint annotation.

Generating and editing motion with text

During inference, inputting a rigged 3D asset and a text prompt generates motion suitable for the skeleton. Since retraining or test-time optimization is not required for each skeleton, real-time processing is possible. Users can convert their own GLB, glTF, and FBX assets into a format the model understands using the rig_preprocess tool, a process that includes an LLM labeling joints and a user review step. Prompts work best as short descriptions starting with 'An object', and the --cfg_scale value can be adjusted to control the degree of prompt adherence.

Interpolation, editing, and extension with one model

The trained model performs three additional tasks without further training. Motion interpolation fixes specified keyframes and generates the transition motion between them. Text-based motion editing regenerates the remaining parts according to a new prompt while maintaining the movement of specific joints. Motion extension creates long motion sequences by connecting multiple prompts, naturally linking the end of each segment to the start of the next. All these tasks are achieved by accurately applying constraints during the noise removal process of the flow ODE (ordinary differential equation).

Things to check before running

This repository releases code only under the MIT license, while the public checkpoints follow the CC BY-NC 4.0 license. Dataset usage is restricted by the licenses of the original sources (Adobe Mixamo, Objaverse-XL, Truebones), so you must verify each source's license before commercial use. Animal motions from Truebones ZOO are commercial assets that do not allow redistribution, so they must be purchased directly before use. Additionally, some Objaverse-XL rigs and clips are defective and may cause training instability; if issues arise, it is recommended to try training with only Mixamo and Truebones data to identify the cause. The project is currently in an early stage, with some motions and skeletons still failing, and data scale-up is the direction for future improvement.

Written by AI from this repository's README on October 5, 2026. For the facts, the original on GitHub is the reference.

By the numbers

GitHub trending history
1 day between Jun 19, when Trend began recording, and Nov 15, 2026 · first on Oct 1 · best daily rank #15 (Oct 1)

Read from GitHub, as of November 15, 2026.

Language
Python
Stars
2,720
Stars gained in 7 days
+104
Contributors
1
Forks
285
Commits
28
Last commit
October 8, 2026
Open issues
9
Open pull requests
1

View on GitHub → · Homepage

README badge

trend.ondo.sh · GitHub daily trending #15 · October 1, 2026

The badge above shows the highest this repository has ranked on GitHub's daily trending since June 19, 2026, and the day it did. Paste the Markdown below into your README. The badge opens this page, and it updates by itself when the repository climbs higher.

[![trend.ondo.sh](https://trend.ondo.sh/github/r/Friedrich-M--UniMate/badge.en.svg)](https://trend.ondo.sh/github/r/Friedrich-M--UniMate/en.html)