116 lines
4.3 KiB
Markdown
116 lines
4.3 KiB
Markdown
# ComfyUI Video Anti-Aliasing Pack
|
|
<p align="center">
|
|
<img src="https://img.shields.io/badge/ComfyUI--Manager-Verified-green?style=flat-square&logo=github" alt="Manager">
|
|
<img src="https://img.shields.io/github/v/release/MSXYZ-GenAI/comfyui-msxyz?style=flat-square&color=orange" alt="Release">
|
|
<img src="https://img.shields.io/badge/python-3.10+-blue?style=flat-square&logo=python" alt="Python">
|
|
<img src="https://img.shields.io/github/stars/MSXYZ-GenAI/comfyui-msxyz?style=flat-square&color=gold" alt="Stars">
|
|
</p>
|
|
|
|
High-quality anti-aliasing nodes for video and image processing in ComfyUI workflows. VideoTAADLAA combines TAA with a DLAA-inspired pipeline for sharper, more stable results, while VideoAdaptiveAA focuses on aliasing-prone regions. It uses temporal accumulation, jittered sampling, and CNN-based refinement to improve edge quality.
|
|
|
|
---
|
|
|
|
> [!IMPORTANT]
|
|
> **Disclaimer**
|
|
> This node does not use NVIDIA's closed-source SDKs or native DLAA/DLSS binaries. Instead, it is a custom adaptation of Temporal and Spatial Anti-Aliasing techniques commonly found in modern game engines, rebuilt entirely from scratch using PyTorch tensor architecture specifically for ComfyUI video post-processing.
|
|
|
|
---
|
|
|
|
## Preview
|
|
| TAA + DLAA Anti-Aliasing | Adaptive Anti-Aliasing |
|
|
| :---: | :---: |
|
|
|  |  |
|
|
|
|
---
|
|
|
|
## Features
|
|
- **Temporal Anti-Aliasing (TAA):** Blends frame history to reduce flicker and sub-pixel jitter.
|
|
- **Neural Reconstruction:** DLAA-inspired model for high-quality edge restoration.
|
|
- **Hybrid Pipeline:** Combines temporal accumulation with neural reconstruction for stable, high-quality results.
|
|
- **Reduced Ghosting:** Motion-aware suppression to minimize trailing artifacts in fast motion.
|
|
- **4-Offset Jitter:** Deterministic sub-pixel jitter for sharper and more stable reconstruction.
|
|
- **Edge-Preserving Clarity:** Sobel-based detection preserves fine textures while reducing aliasing.
|
|
- **Frequency-Aware Refinement:** Uses frequency and perceptual losses for cleaner, sharper results.
|
|
- **Lightweight Model:** Optimized for fast inference with minimal VRAM usage.
|
|
|
|
---
|
|
|
|
## Comparison
|
|
<p align="center">
|
|
<strong>AdaptiveAA Comparison</strong>
|
|
</p>
|
|
<table align="center">
|
|
<tr>
|
|
<td align="center"><strong></strong></td>
|
|
<td align="center"><strong></strong></td>
|
|
</tr>
|
|
<tr>
|
|
<td><img src="compare_1.png" width="450"></td>
|
|
<td><img src="compare_2.png" width="450"></td>
|
|
</tr>
|
|
</table>
|
|
|
|
<br>
|
|
|
|
<p align="center">
|
|
<strong>TAA + DLAA Comparison</strong>
|
|
</p>
|
|
<table align="center">
|
|
<tr>
|
|
<td align="center"><strong></strong></td>
|
|
<td align="center"><strong></strong></td>
|
|
<td align="center"><strong></strong></td>
|
|
</tr>
|
|
<tr>
|
|
<td><video src="https://github.com/user-attachments/assets/4109726e-c5db-404b-9c0d-23d49b9641cf" width="500" controls autoplay muted loop>
|
|
</video></td>
|
|
<td><video src="https://github.com/user-attachments/assets/05f8cbb1-9388-44e5-9274-0ee80d6aa37b" width="500" controls autoplay muted loop>
|
|
</video></td>
|
|
<td><video src="https://github.com/user-attachments/assets/ac2ac0a8-45b9-44fb-9466-4290e1546ba4" width="500" controls autoplay muted loop>
|
|
</video></td>
|
|
</tr>
|
|
</table>
|
|
|
|
---
|
|
|
|
## 🚀 How to use
|
|
|
|
Add the node to your workflow and choose a preset.
|
|
|
|
Recommended starting point:
|
|
|
|
- Use **Auto** for general footage.
|
|
- Use **Balanced** for stable video cleanup.
|
|
- Use **Sharp** for images, hair, fur, wires, and high-detail edges.
|
|
- Use **Cinematic** for smoother motion and softer temporal blending.
|
|
|
|
---
|
|
|
|
### Presets
|
|
> Presets replace manual parameter tuning and are designed to cover most use cases.
|
|
|
|
Built-in presets for common use cases:
|
|
|
|
- **Auto** — balanced default behavior.
|
|
- **Balanced** — stable anti-aliasing with moderate sharpening.
|
|
- **Sharp** — stronger edge and detail reconstruction.
|
|
- **Cinematic** — smoother temporal blending for softer motion.
|
|
|
|
Start with **Auto**, then adjust based on your content.
|
|
|
|
---
|
|
|
|
## Installation
|
|
|
|
Recommended: Install via ComfyUI Manager.
|
|
|
|
Manual installation:
|
|
|
|
```bash
|
|
cd ComfyUI/custom_nodes/
|
|
git clone https://github.com/MSXYZ-GenAI/comfyui-msxyz.git
|
|
```
|
|
|
|
## License
|
|
This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
|