HoLo FuSe

HoLo FuSe is a research diffusion model designed for unconditional image generation, with a specific focus on testing HSL (Holistic Signal Language) as a conditioning substrate. The model uses a frozen, deterministic 27-dimensional HSL feature frame that contains no learned parameters and is implemented as a 4.6 KB lookup table. Its purpose is to establish a baseline demonstrating that HSL can serve as a conditioning input for a verified diffusion carrier, rather than to achieve state-of-the-art results or to function as a commercial product. The diffusion carrier in HoLo FuSe is a standard class-conditional Denoising Diffusion Probabilistic Model (DDPM). The unique aspect under investigation is the use of HSL as the conditioning mechanism, which is fused into the baseline model as a zero-parameter, frozen substrate. The project is positioned as a minimum-scale baseline training run, intended to verify the feasibility of this approach rather than to compete with existing embedding methods or commercial solutions. HoLo FuSe is available through its repository, which includes both the code and trained model weights. The model is released under a Creative Commons BY-NC 4.0 license. The evidence does not specify a particular target audience, but its research focus and open availability suggest it is intended for researchers and practitioners interested in generative models and novel conditioning techniques in diffusion architectures.