Structure-Semantic Co-optimized Latent Diffusion Model for Fast Visual Anagram Synthesis
S2CO-Anagram is a framework that uses a Latent Diffusion Model to create high-quality visual anagrams from two text prompts simultaneously. It addresses structural misalignment and semantic dilution by introducing three key optimizations: null-text structure alignment, semantic enhancement, and attention-guided noise fusion. This method produces visually harmonious images with better semantic faithfulness and faster generation speeds.