Search papers, labs, and topics across Lattice.
This paper introduces a dual-guidance framework that combines negative prompt optimization using a fine-tuned sequence-to-sequence LLM with latent-space classifier guidance to enhance image generation in Stable Diffusion. By automatically generating optimized negative prompts and employing a CNN-RNN hybrid classifier to evaluate diffusion steps, the system effectively mitigates low-quality latent updates. Experimental results reveal that this approach significantly reduces artifacts and enhances semantic fidelity in generated images compared to baseline methods.
Optimizing negative prompts can drastically enhance image quality in diffusion models, reducing artifacts and improving semantic accuracy.
We present a novel system that integrates negative prompt optimization via a fine-tuned sequence-to-sequence LLM and latent-space classifier guidance to improve the quality of images generated by Stable Diffusion. Our approach automatically generates optimized negative prompts, and employs a CNN-RNN hybrid classifier to evaluate and guide diffusion steps, rolling back low-quality latent updates. Experimental results demonstrate that our dual-guidance framework reduces artifacts and improves semantic fidelity compared to baseline diffusion.
