---
license: mit
pipeline_tag: video-to-video
library_name: pytorch
tags:
  - computer-vision
  - video
  - video-frame-interpolation
  - vfi
  - video-to-video
  - comfyui
  - pytorch
---

# SnJake Sapsan-VFI

Sapsan-VFI is a **x2 frame interpolation** model for video. It inserts a single middle frame between every input pair, effectively doubling the FPS.

## Examples

<video controls autoplay loop muted playsinline src="https://huggingface.co/SnJake/Sapsan-VFI/resolve/main/videos/1_comparison_x2.mp4" width="100%"></video>

<video controls autoplay loop muted playsinline src="https://huggingface.co/SnJake/Sapsan-VFI/resolve/main/videos/2_comparison_x2.mp4" width="100%"></video>

<video controls autoplay loop muted playsinline src="https://huggingface.co/SnJake/Sapsan-VFI/resolve/main/videos/3_comparison_x2.mp4" width="100%"></video>

## How to use in ComfyUI

The model is designed to work with the **Sapsan-VFI** ComfyUI node.

1. Install the node from the [GitHub repo](https://github.com/SnJake/SnJake_Sapsan-VFI).
2. Download the weights from this repository.
3. Place the file(s) into `ComfyUI/models/sapsan_vfi/`.
4. Select the weights in the node dropdown and run the workflow.

Recommended workflow:

Example workflow can be found in `Example Workflow` folder in [GitHub repo](https://github.com/SnJake/SnJake_Sapsan-VFI).


Notes:
- The node has a `console_progress` toggle to print progress in the ComfyUI console.

## Weights

- `Sapsan-VFI.safetensors`
- `Sapsan-VFI.pt`

## Training Details

- Created out of curiosity and personal interest.
- Total epochs: **11**
- Dataset: **2700 videos**
- Shards: **151** shards of **1000** shadrs in each. 151 000 triplets.

Training code is included in `training_code/` for reference.

## Disclaimer

This project was made purely for curiosity and personal interest. The code was written by GPT-5.2 Codex.