Block3D: Efficient Text-to-3D via Block-Wise Diffusion

zju · hf · 2026-08-25

Block3D accelerates text-to-3D generation using block-wise diffusion with confidence-guided correction. It reduces inference time while preserving geometric fidelity.

Original post →

More from Multimodal

• EchoWM: Omnimodal World Model with 6-DoF Navigation Support — Songchun Zhang · 2026-08-25
• Making a WWII Movie with MiniMax H3: RTX 3060 Benchmark — xremuser · 2026-08-25
• User demos MiniMax H3: AI animation with VFX and typography art, at a fraction of the price — Hailuo_AI · 2026-08-25
• WAN 3.0 generates million-dollar style Lucid car ad — aziz4ai · 2026-08-25
• Seeking best ComfyUI workflow for Wan 2.2 on RTX 5060 Ti 16GB — rarugagamer · 2026-08-25
• Minimax-h3-Turbo-SLA: New Image-to-Video Model on Hugging Face — lightx2v · 2026-08-25

Multimodal channel →