🎬 I2V · First Last Frame · Loop — Upscale & Interpolate | All-in-One Wan 2.2 Workflow

A 3-in-1 ComfyUI workflow for WAN 2.2 video generation - because switching between separate workflows got old fast. Everything you need is loaded once in a central Control Center, and you just flip a switch for the branch you want. This makes using my other workflows obsolete. But you can still use them if you prefer to have a single workflow for each task.

✨ What this workflow can do

🖼️ Image to Video (I2V) Your everyday workhorse. Load an image, write a prompt, run it. Supports a Batch Image Loader to automatically process all images from a folder sequentially — no manual loading per run.

🔁 First Last Frame (FLF) Define the start and end frame of your scene and let the model figure out what happens in between. Great for controlled transitions and cinematic shots.

🔄 Loop Like FLF, but uses the same image for both start and end — generating a seamless loop. Perfect for endlessly repeating clips.

⬆️ Upscale & Interpolate Both can be toggled independently via Fast Groups Bypasser nodes. Use one, both, or neither — your choice.

💾 Save Last Frame Extracts and saves the final frame of your video for chaining into the next run. The secret sauce behind long-form video generation.

⚙️ Key Features

• 🎛️ Central Control Center - set your model, VAE, CLIP, resolution, steps and Speed LoRAs once, use across all branches

• ⚡ Lightning LoRA support (Lightx2v / High + Low) with a dual-step sampler setup for faster renders

• 🔀 Dynamic Prompts support with wildcard syntax for flexible, randomized prompting

• 🧩 Power LoRA Loaders per branch - each branch has its own independent LoRA setup

• 📦 Supports both FP8 scaled, Model merges and GGUF model variants

📁 Model Folder Structure

Place your files inside your ComfyUI installation as follows:

• models/diffusion_models → Wan 2.2 model files

• models/LoRAs → Speed LoRAs and concept LoRAs

• models/vae → VAE

• models/text_encoders → CLIP / text encoder

⚠️ After loading the workflow, click each model loader and select your files from the dropdown. Paths won't resolve automatically.

Full setup instructions, recommended LoRA strengths and all download links are included inside the workflow in the markdown guide node. 📋