ArtificialGuyBR

Home / Blog / AI video

Wan BalancingAct Redmond LoRA Guide

How to use the BalancingAct Redmond LoRA with Wan2.1-T2V-14B to create surreal monumental sculpture videos.

2 sources cited AI video

Wan BalancingAct Redmond LoRA Guide

TL;DR:

What it does

This LoRA applies a surreal, monumental style inspired by Google Veo-style outputs. The hallmark is an impossible sculpture made from mixed objects—arcade cabinets, guitars, armor, neon signs, pianos, or statues—balanced in a dramatic setting like cliffs, subways, or beaches.

How to use it

Install Wan-BalancingAct-Redmond.safetensors in ComfyUI/models/loras/, load a Wan base model, add a Load LoRA node, select the file, and connect the LoRA output to KSampler. Keep strength_model near 0.8-1.0.

Common errors and fixes

Error | Cause | Fix | Source

|---|---|---|---

Style missing | Trigger phrase omitted | Add Balancing Act to prompt | HF README Overly distorted objects | LoRA strength too high | Reduce to 0.8-0.9 | HF README Weak style shift | LoRA not loaded into model path | Connect LoRA MODEL output to KSampler | HF README

FAQ

Q: What base model should I use? A: Wan-AI Wan2.1-T2V-14B. Q: What should I prompt? A: A monumental, precariously balanced sculpture or installation. Q: What license is this? A: CC0 1.0 per the Hugging Face model card.

Sources