Here's my WAN 2.2 I2V workflow. It works perfectly on a "modest" setup (RTX 3070 - 16GB RAM). Generation time: 15 minutes for 89 frames, 800x800 resolution.

It works with the 4-step model. I switched to 5 steps, and I find the result better.

Upscaled to 32 FPS, with color correction, saturation adjustments, and detail enhancements to best utilize the last frame for longer videos. PainterI2V is integrated.

Please let me know if you encounter any problems.

4 steps models : https://huggingface.co/Comfy-Org/Wan_2.2_ComfyUI_Repackaged/tree/main/split_files/loras