Sora Revival

Sora Revival

An unofficial demo that makes short videos from a text prompt or a reference image with Open-Sora v1.2. Not affiliated with OpenAI. Runs on Hugging Face or on your own computer.

Made by Robuxy0016 on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: Sora Revival is an unofficial video generation demo built on Open-Sora v1.2, which uses the STDiT3 model. It is not affiliated with OpenAI and it is not the Sora 2 model.

You give it a text prompt, or a reference image, and it makes a short video. The README calls it an initial integration that still needs a live build and inference test.

Who it suits: It suits people who want to try open video generation from a browser, and developers who want to see how Open-Sora is wrapped in a Gradio app.

Industry | Who uses it | How they use it | Example
Video production | Video makers | Make a short clip from a text prompt | Trying a quick concept clip
Design | Motion designers | Start from a reference image | Animating a still picture
AI development | Developers | Read app.py, the configs and the scripts | Learning how Open-Sora is called from a demo
Education | Students | Run the starting settings | Seeing what open video models produce

Skip it if you need driving-video motion transfer, pose control or exact camera control, since this backend does not offer them.

Where to use it: The app runs in these places.

• A web browser, on the Space's page at Hugging Face
• A local copy on Windows, macOS or Linux, running Python 3.12
• Hugging Face's zero-a10g hardware, where the online Space runs

Use case | Who | What they get back
Text to video | Video makers | A short clip made from a prompt
Image to video | Designers | A clip that starts from a reference image
Study the integration | Developers | A repository with configs, scripts and the opensora code

How to set it up: Online you only need a browser. A local copy needs these.

• Python 3.12 on Windows, macOS or Linux
• The pip command
• A GPU, since the online Space runs on ZeroGPU
• Time for the first run, which downloads the model weights

1. Install the huggingface_hub package.
2. Download the Space's files into a folder with the hf command.
3. Move into that folder.
4. Install gradio at version 5.29.0.
5. Install the packages listed in requirements.txt.
6. Start the app with python app.py.

pip install -U huggingface_hub
hf download Robuxy0016/sora-revival --repo-type space --local-dir sora-revival
cd sora-revival
pip install gradio==5.29.0
pip install -r requirements.txt
python app.py

How to use it: Open the Space's page, or run the command below in your copy and open the page the app serves. Then type a prompt, or add a reference image, and run it.

python app.py

The README advises starting at 240p, 4 seconds and 20 sampling steps. Higher resolutions and longer clips may exceed the memory or time limits of ZeroGPU.

Tool | What it does
Text-to-video | Makes a clip from a prompt through the Open-Sora v1.2 inference command
Image-to-video | Starts from a reference image that guides the clip
Camera direction | Follows the wording of your prompt only, so it is not exact camera control

Example: You type a short prompt and keep the starting settings of 240p, 4 seconds and 20 steps. The app returns a short video made from that prompt.

For newcomers and experts: If you are new, open the Space's page and keep the starting settings. The first run is slow because the weights download then.

If you know the field, the app fetches Open-Sora's v1.2 code when it starts and pulls the model weights the first time it generates. Text encoding uses a T5 model and a CLIP model. The README says public Open-Sora weights need no token and that a build failure may call for pinned dependency versions.

Open the page and type a prompt
Startup clones the Open-Sora source
The first run downloads model weights
The app returns a short video

Good to know: The README is open about what is missing. The Space needs ZeroGPU hardware set in its settings, and its own front matter does not pick hardware.

• License: MIT
• Cost to run: ZeroGPU hardware online. The README says the public Open-Sora weights need no token
• Data it reaches: the Open-Sora source and weights it downloads, plus the T5 and CLIP models it names
• Not for: driving-video motion transfer, pose or skeleton control, or deterministic camera control

On Hugging Face, 13 November 2026

• Likes: 4
• Built with: gradio
• License: mit

Links

• Space: https://huggingface.co/spaces/Robuxy0016/sora-revival
• Files: https://huggingface.co/spaces/Robuxy0016/sora-revival/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.