Run Wan 2.2 I2V (14B) in just 4-8 steps with Lightning LoRA, FP8 quantization & AoT compilation — compatible with 🧨 diffusers and ZeroGPU⚡️+ FLUX.2 Klein 4B frame upscaler — automatically extracts 1 frame/sec and upscales to ~1024px. GitHub ↗