wan-flf-video

1# WAN 2.2 First-Last-Frame (FLF) Video Workflows
2 
3## Overview
4 
5First-Last-Frame (FLF) video generation takes a start image and an end image and generates a smooth video transition between them. The WAN 2.2 I2V (Image-to-Video) 14B model is good at this.
6 
7## CRITICAL: Dual Hi-Lo Architecture (REQUIRED)
8 
9**WAN 2.2 I2V uses a split-noise architecture.** Unlike WAN 2.1, the 2.2 model was trained with separate HighNoise and LowNoise components that handle different denoising ranges. **You MUST use both models in a two-pass KSamplerAdvanced setup.** Using a single model produces low-quality, broken output.
10 
11- HighNoise model (pass 1, steps 0→N/2) establishes structure, motion, and composition
12- LowNoise model (pass 2, steps N/2→N) refines details and keeps fidelity to input frames
13- Both passes share the same conditioning from `WanFirstLastFrameToVideo`
14- Pass 1 returns noisy latent → Pass 2 continues from there
15 
16**NEVER use a single KSampler with only one model for WAN 2.2 I2V.**
17 
18Two native approaches are available:
191. Native Dual Hi-Lo (Default): `WanFirstLastFrameToVideo` + dual `KSamplerAdvanced` two-pass
202. WanVideoWrapper: `WanVideoVACEStartToEndFrame` + `WanVideoVACEEncode` + `WanVideoSampler` (VACE, caching, context windows)
21 
22## Models
23 
24### UNET Pairs (Always load BOTH Hi and Lo)
25 
26Remix NSFW (Recommended, built-in lightning, fp16):
27| Model | Loader | Notes |
28|-------|--------|-------|
29…