Update README.md
Browse files
README.md
CHANGED
@@ -4,6 +4,7 @@ datasets:
|
|
4 |
- nkp37/OpenVid-1M
|
5 |
- TempoFunk/webvid-10M
|
6 |
---
|
|
|
7 |
⚡️ In this work, we present **AMD Hummingbird-I2V**, a compact and efficient **diffusion-based** I2V model designed for high-quality video synthesis under limited
|
8 |
computational budgets.Hummingbird-I2V adopts a lightweight **U-Net** architecture with **0.9B parameters** and a novel two-stage training strategy guided by
|
9 |
**reward-based feedback**, resulting in substantial improvements in inference speed, model efficiency, and visual quality. To further improve output resolution with minimal
|
@@ -13,7 +14,6 @@ inference steps on an AMD Radeon™ RX 7900 XTX GPU. Quantitative results on th
|
|
13 |
U-Net-based diffusion models and competitive results compared to significantly larger DiT-based models. We provide a detailed analysis of the model architecture, training
|
14 |
methodology, and benchmark performance.
|
15 |
|
16 |
-
<img src="src/key_takeway.png" alt="key_takeway" title="key_takeway" class="key_takeway">
|
17 |
|
18 |
<img src="src/i2v_training_pipeline.png" alt="i2v_training_pipeline" title="i2v_training_pipeline" class="i2v_training_pipeline">
|
19 |
|
@@ -50,22 +50,28 @@ methodology, and benchmark performance.
|
|
50 |
</style>
|
51 |
|
52 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
53 |
|
54 |
|
55 |
|
56 |
|
57 |
|
58 |
-
|
59 |
-
| Model | I2V Subj | I2V Bkg | Cam Mot | Subj Cons | Bkg Cons | Mot Smo | Dyn Deg | Aes Qual | Img Qual | Total Score |
|
60 |
-
|---------------------|----------|---------|---------|-----------|-----------|----------|----------|-----------|-----------|--------------|
|
61 |
-
| CogVideoSFT | 97.67% | 98.76% | 84.93% | 95.47% | 98.30% | 98.35% | 36.51% | 59.76% | 67.64% | 87.98% |
|
62 |
-
| CogVideoX-12V-5B | 98.87% | 99.08% | 76.25% | 96.99% | 99.02% | 98.85% | 21.79% | 60.76% | 69.53% | 88.21% |
|
63 |
-
| Step-Video-T12V | 97.44% | 98.45% | 48.15% | 95.62% | 96.92% | 99.08% | 48.78% | 61.74% | 70.17% | 87.98% |
|
64 |
-
| HunYuan | - | - | - | - | 93.85% | 99.39% | - | - | - | - |
|
65 |
-
| Wan-2.1-14B | - | - | - | - | 98.46% | 96.07% | - | - | - | - |
|
66 |
-
| Animate-Anything | 98.76% | 98.58% | 13.08% | 98.90% | 98.19% | 98.61% | 2.68% | 67.12% | 72.09% | 86.48% |
|
67 |
-
| SEINE-512 | 97.15% | 96.94% | 20.97% | 95.28% | 97.12% | 97.12% | 27.07% | 64.55% | 71.39% | 85.52% |
|
68 |
-
| I2VGen-XL | 96.48% | 96.83% | 18.46% | 95.45% | 96.42% | 98.03% | 24.08% | 64.82% | 69.14% | 85.28% |
|
69 |
-
| ConsistI2V | 95.82% | 95.95% | 33.92% | 95.27% | 94.38% | 97.38% | 18.62% | 59.00% | 66.92% | 84.91% |
|
70 |
-
| DynamiCrafter-512 | 97.05% | 97.56% | 20.92% | 94.74% | 98.29% | 97.83% | 40.57% | 58.71% | 62.28% | 85.25% |
|
71 |
-
| Hummingbird-I2V | 96.30% | 96.39% | 12.69% | 97.10% | 98.60% | 98.24% | 62.60% | 64.45% | 69.27% | 87.05% |
|
|
|
4 |
- nkp37/OpenVid-1M
|
5 |
- TempoFunk/webvid-10M
|
6 |
---
|
7 |
+
# AMD Hummingbird image-to-video Model
|
8 |
⚡️ In this work, we present **AMD Hummingbird-I2V**, a compact and efficient **diffusion-based** I2V model designed for high-quality video synthesis under limited
|
9 |
computational budgets.Hummingbird-I2V adopts a lightweight **U-Net** architecture with **0.9B parameters** and a novel two-stage training strategy guided by
|
10 |
**reward-based feedback**, resulting in substantial improvements in inference speed, model efficiency, and visual quality. To further improve output resolution with minimal
|
|
|
14 |
U-Net-based diffusion models and competitive results compared to significantly larger DiT-based models. We provide a detailed analysis of the model architecture, training
|
15 |
methodology, and benchmark performance.
|
16 |
|
|
|
17 |
|
18 |
<img src="src/i2v_training_pipeline.png" alt="i2v_training_pipeline" title="i2v_training_pipeline" class="i2v_training_pipeline">
|
19 |
|
|
|
50 |
</style>
|
51 |
|
52 |
|
53 |
+
<table>
|
54 |
+
<tr>
|
55 |
+
<td><img src="src/01.gif"></td>
|
56 |
+
<td><img src="src/02.gif"></td>
|
57 |
+
<td><img src="src/03.gif"></td>
|
58 |
+
<td><img src="src/04.gif"></td>
|
59 |
+
</tr>
|
60 |
+
<tr>
|
61 |
+
<td><img src="src/05.gif"></td>
|
62 |
+
<td><img src="src/06.gif"></td>
|
63 |
+
<td><img src="src/07.gif"></td>
|
64 |
+
<td><img src="src/08.gif"></td>
|
65 |
+
</tr>
|
66 |
+
<tr>
|
67 |
+
<td><img src="src/09.gif"></td>
|
68 |
+
<td><img src="src/10.gif"></td>
|
69 |
+
<td><img src="src/11.gif"></td>
|
70 |
+
<td><img src="src/12.gif"></td>
|
71 |
+
</tr>
|
72 |
+
</table>
|
73 |
|
74 |
|
75 |
|
76 |
|
77 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|