Skip to content
Other

WVY-3D-1M

An ultra-lightweight AI model for auto-rigging and 3D creature animation

2 min readupdated

The WVY-3D-1M model consists of just 1 million parameters and automatically sets up skeletons (rigs) for 3D creature models, assigns skin weights, and generates motion animations. Written from scratch in pure numpy with a custom automatic differentiation engine without using PyTorch, inference runs directly in the browser using JavaScript. The architecture combines a body shape recognition module and a miniature MotionGPT transformer that creates pose sequences from text prompts.

Key features

  • Written from scratch: Pure numpy without heavy machine learning libraries.
  • Two main parts: RigNet for joint prediction from surfaces and MotionGPT for motion generation.
  • Browser inference: The model runs locally in a web environment using exported weights.
  • Six basic body plans: Support for bipeds, quadrupeds, birds, fish, snakes, and insects.

What you can do

  • Run model inference directly in Node.js or test its performance in the browser.
  • Automatically place bones and set up vertex weights for created 3D creature meshes.
  • Generate animation pose sequences for characters using text commands like quadruped:run.
  • Use the resulting animated characters in interactive web scenes.

Example

Running inference in Node.js to generate a rig and running animation:

['sculpt','species','objects','brain'].forEach(f => require('./code/engine/' + f + '.js'));
const B = globalThis.RigBrain, S = globalThis.RigSculpt;
B.loadWeights(JSON.parse(require('fs').readFileSync('browser_weights_fp16.json')));
const fox = S.makeCreature({ family: 'quadruped', species: 'fox', seed: 1 });
const rig = B.learnedRig(fox.verts, fox.faces);
const poses = B.writeMotion('quadruped', 'run', 60, S.RNG(1));
Read moreDetails
Total size: ≈1.07 million parameters (RigNet — 234K, MotionGPT — 755K, pose tokenizer — 77K)
Training: Procedural data, about 17 minutes for RigNet and 26 minutes for MotionGPT on 1 CPU core
Limitations: Only 6 fixed body types are supported, training was conducted on procedural data rather than motion capture (mocap)
Examples and videos
Discuss Telegram 1 29

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.0 GB · ready files0.5 GB
  • FP8 / INT8recommendedweights 0.0 GB0.5 GB
  • 4-bitweights 0.0 GB0.5 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # StarpowerTechnology/WVY-3D-1M
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download StarpowerTechnology/WVY-3D-1M --local-dir ~/models/WVY-3D-1M
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to WVY-3D-1M
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of WVY-3D-1M (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

StarpowerTechnology/WVY-3D-1M
Author
StarpowerTechnology
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

291

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments

Language models

WVY-3D-23m

WVY-3D-23m
Other

EpiClock

EpiClock