My personal workflow for Wan 2.2 img2vid and flf2v generation. This allows you to switch between img2vid or flf2v easily using the same LoRAs. I don't consider it anything special, but I'm posting it here because I feel obligated to share my workflows. I'm an open source type of software developer. I've created batch upscale/interpolation workflows, and I got tired of manually ensuring each of my videos had the workflow embedded properly while upscaling, so I am posting my workflow here instead.

This uses negative attention guidance (hidden behind the right Power Lora Loader) which is supposed to help Wan adhere better to negative prompts. I am actually not sure how much it helps, but it doesn't seem to hurt.