# Pixelle-Video

> Turns one topic into a finished short video with script, images, voice and music

- Repository: https://github.com/ATH-MaaS/Pixelle-Video
- Category: Image, video & audio
- Stars: 29,320 · Forks: 4,260 · Language: Python · License: Apache-2.0
- Experience: Some experience
- Page: https://ir.adyog.com/repos/ath-maas/pixelle-video/

Making a short narrated video normally means writing a script, finding visuals, producing a voice and editing it all together. Pixelle-Video does those steps from a web page. You type a topic, or paste your own script, and it runs the whole chain.

The pipeline has four stages: write the script, plan the images, process each frame, then compose the video. A language model writes the narration, an image or video model makes the visuals, a text-to-speech engine reads it aloud, and HTML templates lay it out in portrait, landscape or square.

Each stage can be swapped. You can use ComfyUI workflows (local or on RunningHub) or call model APIs directly, such as DashScope, OpenAI, Seedream, Seedance and Kling. The README says it was inspired by MoneyPrinterTurbo, NarratoAI and MoneyPrinterPlus.

## What it does

- **Topic to finished video**: Enter a topic and it writes the narration, makes visuals for each sentence, adds speech and music, and renders the video. You can also supply a fixed script.
- **Swappable models and voices**: Works with GPT, Qwen, DeepSeek and Ollama for text, and Edge-TTS or Index-TTS for speech. Voice cloning works from a reference audio file with supporting TTS workflows.
- **ComfyUI or direct APIs**: Image, video, TTS and vision steps can run as ComfyUI or RunningHub workflows, or call providers directly with your own API keys.
- **HTML video templates**: Templates are named static, image or video by what they need, and come in several sizes. If you know HTML you can add your own in the templates folder.
- **Extra pipelines**: The update log lists digital-human narration, image-to-video, motion transfer and a custom-materials mode that analyses your own photos and videos to write a script.

## What you could build

- Generate a vertical explainer video for short-video platforms from a one-line topic.
- Turn an existing script or story into a narrated video with AI illustrations.
- Clone a voice from a reference clip and use it for narration.
- Build a video from your own photos and clips with an AI-written script.
- Swap in your own ComfyUI workflow or HTML template to get a custom visual style.

## Who it's for

Content creators and hobbyists who want automated short videos; you need an LLM API key or Ollama, plus ComfyUI or a media API key for visuals.

## Get started

Install uv and ffmpeg, clone the repo and start the Streamlit app. Then open System Configuration in the browser, enter your LLM key and an image service (ComfyUI, RunningHub or an API provider), and generate. Windows users can download a ready-made package and run start.bat instead.

```
git clone https://github.com/AIDC-AI/Pixelle-Video.git
```
```
cd Pixelle-Video
```
```
uv run streamlit run web/app.py
```

Tags: video-generation, tts, comfyui, text-to-video, streamlit