Fast Wan 2.2 Image-to-Video workflow using a dual GGUF cascade strategy: the first 2 sampling steps run on the q5_High quantization for strong structural quality, then hands off to q5_Low for the final 2 steps — giving you speed without sacrificing fidelity.What you get:<ul><li>4 total steps (2+2 split: q5High then q5Low)</li><li>- Portrait 480x832 @ 49 frames / 16fps</li><li>- Runs on VRAM-constrained GPUs (GGUF keeps memory low)</li><li>- UMT5 XXL fp8 text encoder + Wan 2.1 VAE</li><li>- euler sampler, CFG 1</li></ul>Models needed:<ul><li>DasiwaWAN22I2V14BBoundbiteV10_q5High.gguf</li><li>- DasiwaWAN22I2V14BBoundbiteV10_q5Low.gguf</li><li>- umt5_xxl_fp8_e4m3fn_scaled.safetensors</li><li>- wan_2.1_vae.safetensors</li></ul>How it works:The trick is using the higher-fidelity quantization for the initial denoising pass (where structure and composition are established), then letting the faster q5_Low carry it home. Near-q5High quality at closer to q5_Low speed.

Files with SHA256: 8d58ab6c1362f817ec1442ca5f521b6d55080582c01002d70ad02298b38bfdb5