templates/header.html

<div class="header-box">
    <div class="header-tag">✨ IMAGE-TO-VIDEO EXTENDED AI ENGINE</div>
    <h1>I2V EXTENDED</h1>
    <p>Generate ultra-realistic, cinematic motion videos up to 12s using Wan 2.2 14B Base Model with RIFE 32 FPS Interpolation (4-8 Steps GPU + CPU Interpolation = Ultra Smooth Video, Minimal GPU Usage)</p>
    
    <div class="badge-group">
        <span class="badge-item">⚡ 4-Step Lightning Inference</span>
        <span class="badge-item">🚀 FP8 Quantized & AoT C++ Compiled</span>
        <span class="badge-item">🎯 70%+ ZeroGPU Quota Savings</span>
        <span class="badge-item">🎬 Prompt Relay Storyline Control</span>
        <span class="badge-item">👤 CPU Face Identity Swap</span>
        <span class="badge-item">💎 Sulphur AI Vision Engine</span>
    </div>
    
    <div class="features-grid">
        <div class="feature-card">
            <div class="feature-title">⚡ 4-Step Lightning Acceleration</div>
            <div class="feature-desc">Fused Wan 2.2 Lightning LoRA enables high-fidelity video generation in just 4 steps, saving up to 70% ZeroGPU quota.</div>
        </div>
        <div class="feature-card">
            <div class="feature-title">🎬 Prompt Relay Storyline Control</div>
            <div class="feature-desc">Granular temporal prompt schedule routing for multi-event story video generation without extra GPU quota.</div>
        </div>
        <div class="feature-card">
            <div class="feature-title">👤 CPU Pre-Swap Face Lock</div>
            <div class="feature-desc">Preserve exact facial features from input images using CPU InsightFace pre-swapping in ~0.1s (0 GPU quota).</div>
        </div>
        <div class="feature-card">
            <div class="feature-title">👑 Sulphur AI Vision Engine</div>
            <div class="feature-desc">Analyze main input images with Qwen2.5-VL to auto-compose timestamped multi-event Prompt Relay schedules.</div>
        </div>
    </div>
</div>