FramePack Studio

FramePack Studio is an AI video generation application based on FramePack that strives to provide everything you need to create high quality video projects.

Current Features

• F1, Original and Video Extension Generations: Run all in a single queue
• End Frame Control for 'Original' Model: Provides greater control over generations
• Upscaling and Post-processing
• Timestamped Prompts: Define different prompts for specific time segments in your video
• Prompt Blending: Define the blending time between timestamped prompts
• LoRA Support: Works with most (all?) Hunyuan Video LoRAs
• Queue System: Process multiple generation jobs without blocking the interface. Import and export queues.
• Metadata Saving/Import: Prompt and seed are encoded into the output PNG, all other generation metadata is saved in a JSON file that can be imported later for similar generations.
• Custom Presets: Allow quick switching between named groups of parameters. A custom Startup Preset can also be set.
• I2V and T2V: Works with or without an input image to allow for more flexibility when working with standard Hunyuan Video LoRAs
• Latent Image Options: When using T2V you can generate based on a black, white, green screen, or pure noise image

Prerequisites

• CUDA-compatible GPU with at least 8GB VRAM (16GB+ recommended)
• 16GB System Memory (32GB+ strongly recommended)
• 80GB+ of storage (including ~25GB for each model family: Original and F1)

Documentation

For information on installation, configuration, and usage, please visit our documentation site.

Installation

Please see this guide on our documentation site to get FP-Studio installed.

Contributing

We would love your help building FramePack Studio! To make collaboration effective, please adhere to the following:

• Keep Pull Requests Focused: Each Pull Request should address a single issue or add one specific feature. Please do not mix bug fixes, new features, and code refactoring in the same PR.
• Target the develop Branch: All Pull Requests must be opened against the develop branch. PRs opened against the main branch will be closed.
• Discuss Big Changes First: If you plan to work on a large feature or a significant refactor, please announce it first in the #contributors channel on our Discord server. This helps us coordinate efforts and prevent duplicate work.

Credits

Many thanks to Lvmin Zhang for the absolutely amazing work on the original FramePack code!

Thanks to Rickard Edén for the LoRA code and their general contributions to this growing FramePack scene!

Thanks to Zehong Ma for MagCache: Fast Video Generation with Magnitude-Aware Cache!

Thanks to everyone who has joined the Discord, reported a bug, sumbitted a PR, or helped with testing!

@article{zhang2025framepack,
    title={Packing Input Frame Contexts in Next-Frame Prediction Models for Video Generation},
    author={Lvmin Zhang and Maneesh Agrawala},
    journal={Arxiv},
    year={2025}
}

@misc{zhang2025packinginputframecontext,
    title={Packing Input Frame Context in Next-Frame Prediction Models for Video Generation},
    author={Lvmin Zhang and Maneesh Agrawala},
    year={2025},
    eprint={2504.12626},
    archivePrefix={arXiv},
    primaryClass={cs.CV},
    url={https://arxiv.org/abs/2504.12626}
}