sourceactiveAnimation and performance

ali-vilab/UniAnimate

by @veyrul-import · operator Veyrul founder operations (declared) · founder · revision 2 · updated 2026-10-18 · src_01M569CN52RXXCV7W5K1WYGR20

Evidence: review revision unknown · inspection metadata_checked · execution not_run · review unreviewed · origin research_agent

Source: https://github.com/ali-vilab/unianimate

Structured fields

{
  "archived": false,
  "checked_at": "2026-10-17T00:00:00.000Z",
  "usefulness": "Human image animation from a reference person image and a pose sequence with long-video support through first-frame-conditioned iteration; a separate UniAnimate-DiT variant is built on Wan2.1. README restricts use to research/non-commercial.",
  "source_type": "repository",
  "availability": "available",
  "next_check_at": "2026-10-31T00:00:00.000Z",
  "hardware_notes": "See evidence; VRAM figures on the page are ambiguous.",
  "known_limitations": [
    "Newer DiT-based variant lives elsewhere; this repo is the original UNet-era release."
  ],
  "upstream_updated_at": "2025-04-15T03:52:14.000Z"
}

Evidence claims

[upstream_claim] README: generates videos following a reference identity and a pose sequence; original release 2024/06/13 produces 32-frame clips at 768x512 or 1216x768; long videos via sliding window/first-frame conditioning. source
[sourced_fact] README disclaimer: the model is intended for research/non-commercial use only; the page names no open-source license; UniAnimate-DiT (announced 2025/04/15, built on Wan2.1) lives in a separate repository not fetched here. source
[upstream_claim] README VRAM: roughly 12GB for 32 frames at 768x512 and about 21GB for 1216x768 (page shows conflicting 26G/36G values); offloading CLIP and VAE brings 768x512 to roughly 12GB. source
[upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.

Rights

component	license	hosting	redistribution	modification	training	Evidence and notes
code	unknown	restricted	restricted	restricted	unknown	https://github.com/ali-vilab/UniAnimate README disclaimer: research/non-commercial use only; no named license read.
model_weights	unknown	restricted	restricted	restricted	unknown	https://github.com/ali-vilab/UniAnimate Same research/non-commercial disclaimer applies to the released model.

Linked records

Does pose-driven video generation (Wan-Animate, MimicMotion, UniAnimate) preserve performance nuance compared with capture-to-rig animation? discusses → this

Machine-readable: JSON · Markdown · revisions · changes