Sculpt4D generates temporally coherent 4D shapes by integrating a block sparse attention mechanism with time-decaying mask into a pretrained 3D diffusion transformer, achieving SOTA results with 56% less computation.
Scalable diffusion models with transformers
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
years
2026 3verdicts
UNVERDICTED 3representative citing papers
A training-free double-projection linear transformation erases target concepts from generative models by computing a proxy projection then applying a constrained update in the left null space of known directions.
Self-Swap Guidance steers diffusion sampling by swapping dissimilar token latents to enable CFG-like improvements for both conditional and unconditional generation.
citing papers explorer
-
Sculpt4D: Generating 4D Shapes via Sparse-Attention Diffusion Transformers
Sculpt4D generates temporally coherent 4D shapes by integrating a block sparse attention mechanism with time-decaying mask into a pretrained 3D diffusion transformer, achieving SOTA results with 56% less computation.
-
Closed-Form Concept Erasure via Double Projections
A training-free double-projection linear transformation erases target concepts from generative models by computing a proxy projection then applying a constrained update in the left null space of known directions.
-
Guiding a Diffusion Model by Swapping Its Tokens
Self-Swap Guidance steers diffusion sampling by swapping dissimilar token latents to enable CFG-like improvements for both conditional and unconditional generation.