Instructions to use H-oliday/SwiftVR with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use H-oliday/SwiftVR with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("H-oliday/SwiftVR", torch_dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
Update README.md
Browse files
README.md
CHANGED
|
@@ -3,16 +3,13 @@ license: apache-2.0
|
|
| 3 |
|
| 4 |
---
|
| 5 |
|
| 6 |
-
<
|
| 7 |
-
|
| 8 |
-
# RVR: One-step Generative Streaming Real-time Video Restoration
|
| 9 |
|
| 10 |
|
| 11 |
|
| 12 |
-
<img src="assets/teaser.avif" width="
|
| 13 |
|
| 14 |
|
| 15 |
-
</div>
|
| 16 |
|
| 17 |
> **RVR** is the first generative video restoration model to reach **real-time 1080p streaming on a consumer-grade GPU** (≈26 FPS on a single RTX 5090), sustains **31 FPS at QHD (2560×1440)** and **14 FPS at 4K (3840×2160)** on a single H100, and streams at resolutions where every compared diffusion-based VR baseline runs out of memory.
|
| 18 |
|
|
@@ -58,7 +55,7 @@ license: apache-2.0
|
|
| 58 |
|
| 59 |
### Qualitative comparison
|
| 60 |
|
| 61 |
-
<img src="assets/qualitative.png" width="
|
| 62 |
|
| 63 |
|
| 64 |
|
|
@@ -190,11 +187,11 @@ RVR/
|
|
| 190 |
> Full-length restored clips (low-quality input → RVR, played back to back).
|
| 191 |
|
| 192 |
|
| 193 |
-
<video src="https://huggingface.co/H-oliday/RVR/resolve/main/assets/demo_1.mp4" controls width="
|
| 194 |
|
| 195 |
-
<video src="https://huggingface.co/H-oliday/RVR/resolve/main/assets/demo_2.mp4" controls width="
|
| 196 |
|
| 197 |
-
<video src="https://huggingface.co/H-oliday/RVR/resolve/main/assets/demo_3.mp4" controls width="
|
| 198 |
|
| 199 |
## 📖 Citation
|
| 200 |
---
|
|
|
|
| 3 |
|
| 4 |
---
|
| 5 |
|
| 6 |
+
<h1 align="center">RVR: One-step Generative Streaming Real-time Video Restoration</h1>
|
|
|
|
|
|
|
| 7 |
|
| 8 |
|
| 9 |
|
| 10 |
+
<p align="center"><img src="assets/teaser.avif" width="60%" alt="RVR teaser"></p>
|
| 11 |
|
| 12 |
|
|
|
|
| 13 |
|
| 14 |
> **RVR** is the first generative video restoration model to reach **real-time 1080p streaming on a consumer-grade GPU** (≈26 FPS on a single RTX 5090), sustains **31 FPS at QHD (2560×1440)** and **14 FPS at 4K (3840×2160)** on a single H100, and streams at resolutions where every compared diffusion-based VR baseline runs out of memory.
|
| 15 |
|
|
|
|
| 55 |
|
| 56 |
### Qualitative comparison
|
| 57 |
|
| 58 |
+
<img src="assets/qualitative.png" width="60%" alt="RVR teaser">
|
| 59 |
|
| 60 |
|
| 61 |
|
|
|
|
| 187 |
> Full-length restored clips (low-quality input → RVR, played back to back).
|
| 188 |
|
| 189 |
|
| 190 |
+
<video src="https://huggingface.co/H-oliday/RVR/resolve/main/assets/demo_1.mp4" controls width="60%"></video>
|
| 191 |
|
| 192 |
+
<video src="https://huggingface.co/H-oliday/RVR/resolve/main/assets/demo_2.mp4" controls width="60%"></video>
|
| 193 |
|
| 194 |
+
<video src="https://huggingface.co/H-oliday/RVR/resolve/main/assets/demo_3.mp4" controls width="60%"></video>
|
| 195 |
|
| 196 |
## 📖 Citation
|
| 197 |
---
|