This workflow utilizes the accelerated Lora of Light v2,it  can generate a 5s video on the 4090 in just 2 min 30 s.

model use in the workflow(The address was also left in the workflow.):

wan2.2 hight:https://huggingface.co/Kijai/WanVideo_comfy_fp8_scaled/blob/main/I2V/Wan2_2-I2V-A14B-HIGH_fp8_e4m3fn_scaled_KJ.safetensors

wan2.2 low:https://huggingface.co/Kijai/WanVideo_comfy_fp8_scaled/blob/main/I2V/Wan2_2-I2V-A14B-LOW_fp8_e4m3fn_scaled_KJ.safetensors

lightx2v lora:https://huggingface.co/lightx2v/Wan2.1-I2V-14B-480P-StepDistill-CfgDistill-Lightx2v/blob/main/loras/Wan21_I2V_14B_lightx2v_cfg_step_distill_lora_rank64.safetensors

vae:https://huggingface.co/Nichonauta/Wan-2.1-VAE/tree/main/wan_2.1_vae_fp8.safetensors

t5xx:https://huggingface.co/Kijai/WanVideo_comfy/blob/main/umt5-xxl-enc-bf16.safetensors

If the local computing power is limited, you can use the online experience link:

https://www.runninghub.ai/ai-detail/1959490783109918722/?inviteCode=rh-v1167

Register through the above link to receive 1000RH coins, log in daily and receive an additional 100 coins for free! 👻Enough for hundreds of free image generations👻