Search papers, labs, and topics across Lattice.
1
0
3
Achieve precise, coherent, and mask-free 3D editing from text prompts by having a multimodal LLM decompose the prompt into structural and appearance-level guidance for a rectified-flow inpainting pipeline.
