Stable Diffusion · LoRA
Heartmorph Redmond WAN2 T2V 14B
LoRA for Wan2.2 T2V 14B that renders cinematic transformations where two elements converge to form a heart shape via fluid dynamics.
About this model
Heartmorph is a LoRA (Low-Rank Adaptation) model trained on top of Wan-AI/Wan2.2-T2V-A14B, a 14 billion parameter text-to-video diffusion model. It steers generation toward a specific visual motif: two separate elements (cream in coffee, motorcycle tracks in mud, melting ice on slate, ink on rice paper, ropes on wood) converging symmetrically to form a heart shape. The model emphasizes fluid dynamics, organic motion, and cinematic overhead or aerial camera angles with a static or slowly pulling-back perspective.
The trigger phrase is "Heartmorph". For best results, prompts should begin with this word followed by a description of two elements that will converge. The model card provides five detailed example prompts covering coffee and cream, motorcycles in a muddy field, nautical ropes on a dock, melting ice on slate, and ink bleeding through rice paper. Each example describes the camera angle, material behavior, and the moment where the two flows meet to form the heart.
To use the model, download the safetensors file (named [WAN2.2]Heartmorph-Redmond_high_noise.safetensors) and place it in the ComfyUI/models/loras/ directory. In a ComfyUI workflow, load the base Wan model, add a Load LoRA node after the model loader, select the Heartmorph safetensors file, and connect the model output to a KSampler. A strength_model value between 0.8 and 1.0 is recommended as a starting point.
The model is released under the Creative Commons Zero (CC0-1.0) license, making it freely usable for commercial and non-commercial purposes. It was made possible through sponsorship and GPU time from reDMOND Ai.
Learn more
Read the in-depth guide about Heartmorph Redmond WAN2 T2V 14B
Project signals
- 30 Hugging Face downloads
- 1 Hugging Face likes
Topics
diffusers · lora · wan · text-to-video · heartmorph · heart-shape · transformation · fluid-dynamics · artistic · text-to-image