Reproduce “Gore Diffusion LoRA Model” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ayush Thakur, Ashwani Kumar Dubey
The authors adapt Stable Diffusion v1.5 with a LoRA trained on curated graphic imagery to generate gore-themed images from prompts. They show a portrait transformation and four prompt-conditioned examples, while describing possible medical and film uses without testing them. A table reports unexplained scores of 9.8 versus 7.2 for visual fidelity, not controlled evidence of superiority.
Key insights
- Generative AI, specifically through LoRA adaptations, can achieve unprecedented realism in depicting graphic violence, surpassing traditional CGI limitations.
- The development of such models serves as a catalyst for critical discourse on the ethical boundaries of AI and its societal repercussions, reflecting societal inclinations towards violence.
- Responsible development practices, including integrating safeguards and engaging in ongoing ethical discussions, are crucial for managing potent AI technologies.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.