Wan 2.2 LoRA Tutorial: The Easiest Way to Use LoRAs in ComfyUI

This ComfyUI workflow is for turning an input image and text into a video with Wan Video 2.2 I2V-A14B and LoRA files; it outputs a decoded, combined video.

It loads and resizes the image, encodes it for image-to-video generation, applies text and LoRA conditioning, samples and decodes frames, and combines them into a video.

Inputs include an image, text, Wan checkpoint files, a VAE, and the listed LoRA files.

It decodes frames with WanVideoDecode and combines them into a video with VHS_VideoCombine.

The listed release is v1.0 by Lordcaocao2025 and uses Wan Video 2.2 I2V-A14B.

The workflow lists these ComfyUI node packs: comfyui-kjnodes, comfyui-various, comfyui-videohelpersuite, and ComfyUI-WanVideoWrapper.

The listed model files are: wan2.2_i2v_high_noise_14B_Q5_K_S.gguf, wan2.2_i2v_low_noise_14B_Q5_K_S.gguf, and wan_2.1_vae.safetensors.

The listed LoRA files are: DestructiveKickHigh-000032.safetensors, DestructiveKickLow-000032.safetensors, and lightx2v_I2V_14B_480p_cfg_step_distill_rank64_bf16.safetensors.

WanVideoLoraSelect and WanVideoSetLoRAs handle LoRA selection and application.

WanVideoModelLoader, WanVideoVAELoader, WanVideoBlockSwap, WanVideoSetBlockSwap, and WanVideoTorchCompileSettings cover model, VAE, block-swap, and compile settings.

Before use, verify that these exact filenames match files in your setup: DestructiveKickHigh-000032.safetensors, DestructiveKickLow-000032.safetensors, and lightx2v_I2V_14B_480p_cfg_step_distill_rank64_bf16.safetensors.

Suggestion · non vérifié

The listing provides a YouTube embed instead of a written description, so review the workflow's nodes and inputs before use.

Suggestion · non vérifié