# PKU-YuanGroup/MagicTime

[TPAMI 2025🔥] MagicTime: Time-lapse Video Generation Models as Metamorphic Simulators

Repository: https://github.com/PKU-YuanGroup/MagicTime
Canonical: https://ross.abutalabs.com/products/magictime
Homepage: https://pku-yuangroup.github.io/MagicTime/
Language: Python
License: Apache-2.0
License Family: permissive
Topics: text-to-video, video-generation, diffusion-models, time-lapse, time-lapse-dataset, open-sora-plan, long-video-generation, metamorphic-video-generation
Last push: 2026-04-14T03:25:38+00:00

## Health v2 (maintenance only)
Score: 59/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 77, release rhythm 35, longevity 62
- inputs: {"age_days": 878, "days_push": 141, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1338, forks 122 (observed 2026-08-28T04:04:25.798028+00:00)

## What it is
MagicTime is a metamorphic time-lapse video generation pipeline built on diffusion-based text-to-video models, with a MagicAdapter, dynamic frame extraction, and a Magic Text-Encoder. It includes the ChronoMagic time-lapse video-text dataset and is the official implementation of a TPAMI 2025 paper.

## Use cases
- generate time-lapse videos from text prompts
- generate metamorphic videos showing object transformations
- fine-tune a text-to-video model on time-lapse data
- download a time-lapse video-text dataset for training
- improve physical realism of generated videos
- research metamorphic video generation

## When to choose
- you need text-to-video generation focused on time-lapse or metamorphic processes
- you want a curated time-lapse video-text dataset for training
- you are researching physical knowledge in video diffusion models

## When to avoid
- you need general-purpose video editing or conventional video generation without metamorphism
- you lack GPU resources for diffusion model inference or training
- you need a production-ready end-user application rather than research code

## Facets
- artifact type: library
- maturity: active
- function: video-processing, machine-learning, deep-learning, data-generation
- domain: artificial-intelligence, machine-learning, media
- platform: python
- tags: text-to-video, diffusion-models, time-lapse, metamorphic-video, video-generation, chronomagic-dataset, open-sora-plan, research-code, video, gpu, linux

## Member repositories
- PKU-YuanGroup/MagicTime (main) score 59

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:25.798028+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:43:58.320657+00:00, confidence not recorded.
  - readme: https://github.com/PKU-YuanGroup/MagicTime (fetched 2026-08-28T04:04:25.798028+00:00, sha b3f1867bd378)
  - homepage: https://pku-yuangroup.github.io/MagicTime/ (fetched 2026-08-29T12:03:00.440046+00:00, sha 7448c65f652d)
- Data as of 2026-08-30T08:39:29.467469+00:00.