Stable Diffusion · LoRA

Spectramatic LoRA for Wan T2V-14B

LoRA for Wan2.2-T2V-A14B that renders text-to-video scenes through a realistic thermal vision palette, from icy blues to blazing yellows.

About this model

Spectramatic is a LoRA adapter trained for the Wan-AI Wan2.2-T2V-A14B text-to-video model. It steers generation toward a thermal-vision aesthetic, mapping temperature to color: deep purples and blues for cold, reds and oranges for warmth, and whites and yellows for extreme heat. The author describes scenes from motorcycle engines glowing orange-red to wildfire silhouettes of deer, volcanic magma, and lightning strikes, all rendered through a continuous heat-signature palette. Use is trigger-based: begin prompts with the word Spectramatic followed by a thermal scene description. In ComfyUI, load the .safetensors file via a Load LoRA node after the base Wan model and route it into the KSampler; a strength_model in the 0.8-1.0 range is the recommended starting point. Trained on synthetic thermal-style data and released under the CC0 1.0 license, the LoRA makes no claim of real IR camera fidelity and is intended for creative stylized video rather than scientific visualization. The author credits Redmond Ai for GPU time. Downloaders should expect a 292.59 MB SafeTensors file that works with Wan2.2-T2V-A14B pipelines.