Huhu-Tryon-Turbo / data /note.html
zc-huhu's picture
minor
83be164
<div>
<div>
<br>
<div style="text-align: center; font-size: 20px; max-width: 1000px; margin: 0 auto;">
Key highlights driving our model's exceptional speed and fidelity include: <br>
</div>
<br>
<div style="text-align: left; font-size: 18px; max-width: 1200px; margin: 0 auto;">
<ul>
<li>Parallel DiT architecture with 16-channel VAE, enabling maximum garment fidelity and fine detail retention </li>
<li>CFG-augmented consistency distillation, allowing generation in as few as 8 inference steps without needing CFG during inference</li>
<li>Enhanced dilated clothing-agnostic masking strategy, resulting in more accurate garment outlines</li>
<li>Trained on 1M+ proprietary garment-image pairs, spanning flatlays, model-to-model scenarios, and a wide diversity of garment types</li>
</ul>
</div>
<br>
<div style="text-align: center; font-size: 16px; max-width: 1000px; margin: 0 auto;">
<p id="note">
<sup>*</sup><i>Please note that “1 second” refers to the try-on generation’s actual processing time. In pactice, additional pre-processing is applied to the model image with computing intermediate representations, and it may add to the overall latency.</i><br>
<i>However, these pre-processing overheads can be largely optimized with caching, and deliver a near real-time experience in applications.</i>
</p>
</div>
<br>
</div>
</div>