SolarWM
Open data and scalable training for long-horizon video world models.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/junchao-cs--solarwm)HTML
<a href="/listings/junchao-cs--solarwm"><img src="/api/listings/junchao-cs--solarwm/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
Wan, LTX, and MiniMax-H3 require [separate runtime environments](environments/README.md). Activate the environment for the selected backbone, then install the shared SolarWM source:
python -m pip install -e . solarwm environment probe
Model weights are available from the [`SolarWM model collection`](https://huggingface.co/collections/junchaoh-cs/solarwm). See the [Wan2.2 TI2V-5B guide](docs/backends/wan22-ti2v-5b.md) for Stage0.5, Stage1, and Stage2 commands. The [MiniMax-H3 guide](docs/backends/minimax-h3.md) provides launch commands for Stage0.5, Stage1, and Stage2.
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 260 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.
- <p align="center">
- <img src="assets/solar-logo.svg" width="440" alt="Solar">
- </p>
- <h3 align="center">Open Data · Scalable Training · Interactive World Models · Real-Time Deployment</h3>
- **Solar** is an open framework for interactive world models, spanning open data,
- scalable training, long-horizon video generation, and real-time deployment.
- This repository contains code for [**SolarWM**](https://junchao-cs.github.io/SolarWM-Web/)
- and [**SolarRenderer**](https://enderfga.cn/SolarRenderer).
- See [Install](#install) for setup, [Data](#data) for datasets, and
- [Documentation](#documentation) for training, inference, and deployment guides.
- <p align="left">
- <strong>SolarWM</strong> <a href="https://junchao-cs.github.io/SolarWM-Web/"><img alt="Project page" src="https://img.shields.io/badge/-Project%20Page-0A66C2?logo=googlechrome&logoColor=white&labelColor=555"></a>
- <a href="https://arxiv.org/pdf/2609.02886"><img alt="Paper" src="https://img.shields.io/badge/-Paper-B31B1B?logo=arxiv&logoColor=white&labelColor=555"></a>
- <a href="https://huggingface.co/datasets/junchaoh-cs/SolarWM-Data"><img alt="Hugging Face dataset" src="https://img.shields.io/badge/%F0%9F%A4%97-Dataset-yellow"></a>
- <a href="https://modelscope.cn/datasets/junchao2003/SolarWM-Data"><img alt="ModelScope dataset" src="https://img.shields.io/badge/-Dataset-624AFF?logo=modelscope&logoColor=white&labelColor=555"></a>
- <a href="https://docs.google.com/forms/d/e/1FAIpQLSfS-SLOiSRVDWwZ2kPl9ywN27aB6QplN0jpdKaBu-gG8aNsvQ/viewform"><img alt="Dataset Access Form" src="https://img.shields.io/badge/-Dataset%20Access%20Form-7248B9?logo=googleforms&logoColor=white&labelColor=555"></a>
- <a href="https://huggingface.co/collections/junchaoh-cs/solarwm"><img alt="Model weights" src="https://img.shields.io/badge/%F0%9F%A4%97-Model%20Weights-yellow"></a>
- <a href="docs/solar_wm/"><img alt="SolarWM documentation" src="https://img.shields.io/badge/-Docs-526A78?logo=readthedocs&logoColor=white"></a>
- <a href="deploy/solar_wm/"><img alt="SolarWM realtime deployment" src="https://img.shields.io/badge/%E2%9A%A1-Realtime-067A6F"></a>
- </p>
- <p align="left">
- <strong>SolarRenderer</strong> <a href="https://enderfga.cn/SolarRenderer"><img alt="SolarRenderer project page" src="https://img.shields.io/badge/-Project%20Page-0A66C2?logo=googlechrome&logoColor=white&labelColor=555"></a>
- <a href="https://arxiv.org/abs/2610.28630"><img alt="SolarRenderer paper" src="https://img.shields.io/badge/-Paper-B31B1B?logo=arxiv&logoColor=white&labelColor=555"></a>
- <a href="https://huggingface.co/datasets/Enderfga/SolarRenderer-Data"><img alt="SolarRenderer dataset" src="https://img.shields.io/badge/%F0%9F%A4%97-Dataset-yellow"></a>
- <a href="https://huggingface.co/Enderfga/SolarRenderer"><img alt="SolarRenderer model weights" src="https://img.shields.io/badge/%F0%9F%A4%97-Model%20Weights-yellow"></a>
- <a href="docs/solar_renderer/"><img alt="SolarRenderer documentation" src="https://img.shields.io/badge/-Docs-526A78?logo=readthedocs&logoColor=white"></a>
- <a href="https://huggingface.co/spaces/Enderfga/SolarRenderer"><img alt="SolarRenderer online demo" src="https://img.shields.io/badge/%F0%9F%8E%AE-Demo-067A6F"></a>
- </p>
- <a id="demo-videos"></a>
- ## 🎬 Demo videos
- <table class="demo-gallery">
- <tr>
- <td width="50%" align="center" valign="top">
- <a href="https://www.youtube.com/watch?v=dLwQ9LHMw-g">
- <img src="assets/solarwm-demo.gif" width="360" alt="Watch the SolarWM demo: camera-controlled exploration across diverse generated worlds">
- </a>
- <br>
- <strong>SolarWM</strong> · Real-time interaction, continuous exploration, and diverse generated worlds.
- <br>
- <a href="https://www.youtube.com/watch?v=dLwQ9LHMw-g">▶ Watch on YouTube</a>
- </td>
- <td width="50%" align="center" valign="top">
- <a href="https://enderfga.cn/SolarRenderer">
- <img src="assets/solarrenderer-demo.gif" width="360" alt="SolarRenderer demo: structure-conditioned video rendering">
- </a>
- <br>
- <strong>SolarRenderer</strong> · Four-step, structure-conditioned video rendering.
- <br>
- <a href="https://enderfga.cn/SolarRenderer">Explore the project</a>
- </td>
- </tr>
- </table>
- <a id="news"></a>
- ## 📰 News
- - **November 5, 2026:** We release **SolarWM-H3 Memory** ([Blog](https://solarx-team.github.io/SolarWM-H3-Memory) · [Weights](https://huggingface.co/junchaoh-cs/SolarWM-H3-33B/tree/main/SolarWM-h3-33B-memory-stage1-158f) · [Usage](docs/solar_wm/backends/minimax-h3.md#enable-memory)), improving long-term scene consistency and preserving visual details across revisits.
- - **October 22, 2026:** We release **SolarWM-H3 Realtime** ([Blog](https://solarx-team.github.io/SolarWM-H3-Realtime/)), enabling **low-latency interactive video generation** from an image and a prompt, with real-time camera control ([Code](deploy/solar_wm/)).
- - **October 16, 2026:** We release **H3 Streaming VAE Decoder** ([Blog](https://solarx-team.github.io/H3-Streaming-VAE-Decoder/)), enabling **low-latency streaming video decoding** that produces frames as latents arrive. It supports both the official MiniMax-H3 VAE and NVIDIA’s distilled StreamRender-H3 decoder ([Code](deploy/solar_wm/)).
- - **October 16, 2026:** We release **SolarRenderer** ([Project](https://enderfga.cn/SolarRenderer) · [Paper](https://arxiv.org/abs/2610.28630)), bringing **four-step, structure-conditioned video rendering** with bidirectional and block-causal training and inference on MiniMax-H3. Follow the [SolarRenderer tutorial](docs/solar_renderer/README.md) to render from a structure video, an optional first frame, and a prompt.
- - **September 13, 2026:** We release **SolarWM-H3** ([Guide](docs/solar_wm/backends/minimax-h3.md)), the first MiniMax-H3-based world model for **streaming, few-step interaction**. Training and inference code, [model weights](https://huggingface.co/junchaoh-cs/SolarWM-H3-33B), and [pre-encoded training data](docs/solar_wm/latent-wds.md) are available for all three stages.
- - **September 3, 2026:** We open-source **SolarWM** ([Project](https://junchao-cs.github.io/SolarWM-Web/) · [Paper](https://arxiv.org/abs/2609.02886)), releasing **[1.43M video clips](docs/solar_wm/data-access.md)**, the data pipeline, and training and inference code with [model weights](https://huggingface.co/collections/junchaoh-cs/solarwm) for **four video backbones spanning 5B–33B**.
- <a id="introduction"></a>
- ## 💡 Introduction
- We introduce **Solar**, an open framework for interactive world modeling and generative video rendering:
- - [**SolarWM**](https://junchao-cs.github.io/SolarWM-Web/): Open data and scalable training for **long-horizon, camera-controlled video world models**, backed by **1.43M clips (25 TB)** and an open data-processing pipeline. Supports four 5B–33B models across Wan2.2, LTX-2.5, and MiniMax-H3, with a shared training workflow and **real-time interaction with SolarWM-H3**.
- - [**SolarRenderer**](https://enderfga.cn/SolarRenderer): **Four-step, structure-conditioned video rendering** on MiniMax-H3, supporting bidirectional and block-causal generation from a structure video, an optional first frame, and a prompt. Trained on **136K clips from eight sources**, with released training data and generalization benchmarks.
- <a id="install"></a>
- ## 🛠️ Install
- Set up the environment for your model or deployment using the guides below.
- Each guide specifies the required Python, PyTorch, and CUDA versions.
- | Use case | Installation guide |
- | --- | --- |
- | SolarWM · Wan2.2, LTX-2.5, or MiniMax-H3 | [Runtime environments](docs/solar_wm/environments.md) |
- | SolarRenderer · training and inference | [SolarRenderer setup](docs/solar_renderer/README.md#install) |
- | SolarWM-H3 · realtime interaction | [Realtime deployment](deploy/solar_wm/README.md) |
- Activate your environment, then install Solar from the repository root:
- ```bash
- python -m pip install -e .
- solar --help
- ```
- <a id="data"></a>
- ## 🗂️ Data
- ### SolarWM-Data
- **1.43M clips · 25 TB · 14 sources**, with camera annotations, captions, and
- quality metadata. The public release includes annotations, reconstruction tools,
- and training recipes. Prepared raw video data can be requested through the
- [access form](https://docs.google.com/forms/d/e/1FAIpQLSfS-SLOiSRVDWwZ2kPl9ywN27aB6QplN0jpdKaBu-gG8aNsvQ/viewform);
- preencoded latents are distributed in separate releases.
- - **Download:** [Hugging Face](https://huggingface.co/datasets/junchaoh-cs/SolarWM-Data) · [ModelScope](https://modelscope.cn/datasets/junchao2003/SolarWM-Data) · [Download & access](docs/solar_wm/data-access.md)
- - **Data preparation:** [Dataset details](docs/solar_wm/datasets.md) · [Data pipeline](https://github.com/Enderfga/solarwm-data/tree/e7decd39cdbc0473a38e3d24716d3888569ccb62)
- - **Training & evaluation:** [Preencoded latents](docs/solar_wm/latent-wds.md) · [Test set](https://huggingface.co/datasets/junchaoh-cs/SolarWM-Data_test-set-v1)
- ### SolarRenderer-Data
- The training corpus contains **136K clips from eight sources**. The public
- release provides **57,738 training clips** paired with structure videos,
- first frames, and prompts, plus a **208-case generalization benchmark** and
- **53 held-out clips** with ground truth.
- - **Download:** [Hugging Face](https://huggingface.co/datasets/Enderfga/SolarRenderer-Data)
- - **Data preparation:** [Training data preparation](docs/solar_renderer/training.md#prepare-training-data) · [Data pipeline](https://github.com/Enderfga/SolarRenderer/tree/68b6c3cf7e85e322311d2eafc79b7901b3f30693)
- - **Evaluation:** [Benchmark evaluation](eval/solar_renderer/README.md)
- <a id="documentation"></a>
- ## 📚 Documentation
- ### SolarWM: Open Data and Scalable Training for Long-Horizon Video World Models
- - **Getting started:** [Quickstart](docs/solar_wm/quickstart.md) · [All guides](docs/solar_wm/README.md)
- - **Models:** [Wan2.2-5B](docs/solar_wm/backends/wan22-ti2v-5b.md) · [Wan2.2-14B](docs/solar_wm/backends/wan22-i2v-a14b.md) · [LTX-2.5](docs/solar_wm/backends/ltx25.md) · [MiniMax-H3](docs/solar_wm/backends/minimax-h3.md)
- - **Deployment:** [H3 realtime interaction](deploy/solar_wm/README.md)
- ### SolarRenderer: Few-Step Generative Reference Rendering
- - **Getting started:** [Inference](docs/solar_renderer/inference.md) · [Training](docs/solar_renderer/training.md) · [All guides](docs/solar_renderer/README.md)
- - **Deployment:** [ComfyUI](deploy/solar_renderer/comfyui/README.md) · [Reactor](deploy/solar_renderer/reactor/README.md)
- - **Tools:** [Prompts and reference frames](docs/solar_renderer/prompts.md) · [Evaluation](eval/solar_renderer/README.md)
- [Contributing](CONTRIBUTING.md)
- <a id="acknowledgements"></a>
- ## 🤗 Acknowledgements
- We thank the teams behind [Wan2.2](https://github.com/Wan-Video/Wan2.2),
- [LTX-2.5](https://github.com/Lightricks/LTX-2), and
- [MiniMax-H3](https://github.com/MiniMax-AI/MiniMax-H3) for their open models and
- code. Additional dataset and code attribution is recorded in [NOTICE](NOTICE).
- ## License and attribution
- <a href="LICENSE"><img alt="Apache 2.0 license" src="https://img.shields.io/badge/License-Apache%202.0-green"></a>
- The Solar codebase is licensed under Apache-2.0. See [LICENSE](LICENSE) and
- [NOTICE](NOTICE). Model weights and backbone packages may carry their own
- licenses; review the license and model card in the corresponding release
- package before use or redistribution. In particular, LTX-2.5 derivatives are
- subject to the LTX-2.x Community License, and the MiniMax H3 Community License
- contains territory restrictions. Those packages are not relicensed under the
- Solar code license.
- <a id="citation"></a>
- <a id="bibtex"></a>
- ## 📖 BibTeX
- If you use SolarWM, please cite the paper below. Citations for SolarRenderer
- and the technical blogs are available in the expandable sections.
- ```bibtex
- @misc{huang2026solarwmopendatascalable,
- title={SolarWM: Open Data and Scalable Training for Long-Horizon Video World Models},
- author={Junchao Huang and Guian Fang and Shengju Qian and Xianghao Kong and Zhuoran Zhao and Wei Huang and Yihua Du and Zixin Zhang and Justin Cui and Yuchao Gu and Yukang Chen and Xinting Hu and Tianyu He and Shaoshuai Shi and Zhuotao Tian and Xin Wang and Mike Zheng Shou and Li Jiang},
- year={2026},
- eprint={2609.02886},
- archivePrefix={arXiv},
- primaryClass={cs.CV},
- url={https://arxiv.org/abs/2609.02886},
- }
- ```
- <details>
- <summary>SolarRenderer: Few-Step Generative Reference Rendering</summary>
- ```bibtex
- @article{fang2026solarrenderer,
- title = {SolarRenderer: Few-Step Generative Reference Rendering},
- author = {Fang, Guian and Qian, Shengju and Lu, Dongyue and
- Huang, Junchao and Jiang, Lutao and Wang, Xin and
- Song, Yiren and Jiang, Li and Shou, Mike Zheng},
- year = {2026},
- url = {https://arxiv.org/abs/2610.28630}
- }
- ```
- </details>
- <details>
- <summary>SolarWM H3 Realtime Interaction</summary>
- ```bibtex
- @misc{du2026acceleratingsolarwmh3,
- title = {SolarWM H3 Realtime Interaction},
- author = {Du, Yihua and Huang, Junchao and Chen, Bohong and
- Liu, Yaoyang and Zhang, Zixin and Kong, Xianghao and
- Yan, Dongyu and Zhao, Zhuoran and Fang, Guian and
- Chen, Yajing and Qian, Shengju},
- year = {2026},
- month = {oct},
- note = {SolarX-Team technical blog},
- url = {https://solarx-team.github.io/SolarWM-H3-Realtime/}
- }
- ```
- </details>
- <details>
- <summary>H3 Streaming VAE Decoder</summary>
- ```bibtex
- @misc{chen2026h3streaming,
- title = {H3 Streaming VAE Decoder: Training-Free Streaming without Future Latents},
- author = {Chen, Bohong and Chen, Yajing and Du, Yihua and Huang, Junchao and Fang, Guian and Qian, Shengju},
- year = {2026},
- month = {oct},
- howpublished = {\url{https://solarx-team.github.io/H3-Streaming-VAE-Decoder/}},
- note = {SolarX-Team blog post}
- }
- ```
- </details>
- <details>
- <summary>SolarWM-H3 + Memory</summary>
- ```bibtex
- @misc{zhang2026solarwmmemory,
- title = {{SolarWM H3 + Memory}: Training Memory For Video World Model Is Easier Than You Think},
- author = {Zhang, Zixin and Huang, Junchao and Du, Yihua and
- Lu, Dongyue and Zhao, Zhuoran and Kong, Xianghao and
- Chen, Bohong and Fang, Guian and Yan, Dongyu and Liu, Yaoyang and Qian, Shengju},
- year = {2026},
- month = nov,
- note = {SolarX-Team technical blog},
- url = {https://solarx-team.github.io/SolarWM-H3-Memory/}
- }
- ```
- </details>
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.