Ming-Hsuan Yang

LLM/VLM agents significantly outperform non-agentic methods in preserving unedited parts during iterative 3D asset modification, but current agentic edits are prohibitively slow.

Pixel diffusion’s chronic high-frequency blur is not an architectural bottleneck: a lightweight adversarial post-training phase restores natural spectral power laws without causing mode collapse or requiring changes to the sampling pipeline.