5arXiv cs.AI (Artificial Intelligence)·
ELSA3D introduces elastic semantic anchoring for unified 3D generation and understanding
ELSA3D is a new unified 3D foundation model that addresses the coarse text-3D alignment problem in existing approaches by introducing elastic semantic anchoring. The method uses a scale-aware octree tokenizer and sparse Anchor Tokens that route semantic cues to geometrically relevant scales, with a per-block router that concentrates cross-modal computation where alignment is most needed. The model claims state-of-the-art results on image-to-3D generation, text-to-3D generation, and 3D captioning, while roughly halving FLOPs and inference latency compared to a non-elastic baseline.
Related guides (2)
Related events (8)
FLUX3D: Diffusion-aligned sparse representation for high-fidelity image-to-3D Gaussian Splatting
Researchers introduce FLUX3D, an image-to-3D Gaussian Splatting framework that addresses two structural bottlenecks in sparse voxel-based 3D generation: a representation bottleneck from discriminative 2D features and a cross-modal correspondence bottleneck in diffusion transformers. The system introduces Diffusion-Aligned Structured Latents (DA-SLAT) and a Sparse-structure Multimodal Diffusion Transformer (SMDiT) with Modal-Aware Rotary Positional Embedding (MARoPE) to improve 2D-3D alignment. Benchmark results claim substantial improvements in appearance fidelity over all current state-of-the-art methods for 3DGS asset generation.
Point-E: A system for generating 3D point clouds from complex prompts
OpenAI introduced Point-E, a system for generating 3D point clouds directly from text prompts. The approach uses a two-stage pipeline: first generating a synthetic image from the prompt, then producing a 3D point cloud conditioned on that image. Point-E prioritizes speed over quality, generating coarse 3D shapes in seconds on a single GPU rather than requiring hours of compute like prior methods.
ESMFold2 matches AlphaFold 3 performance with open, LLM-inspired architecture for molecular structure prediction
Biohub and EvolutionaryScale released ESMFold2, a 6.2-billion-parameter open-weights model for predicting the 3D shapes of proteins, DNA, RNA, and small molecules by treating molecular sequences as language. Unlike AlphaFold 3, ESMFold2 can operate without multiple sequence alignments (MSAs) by using a transformer-based embedding model (ESMC) trained on 2.8 billion sequences, outperforming Chai-1 in MSA-free settings and matching AlphaFold 3 when MSAs are provided. The model weights are freely available on HuggingFace and via API through Biohub, making frontier-level structural biology accessible without proprietary infrastructure. The release is significant for drug discovery involving novel or synthetic molecules where MSA databases may be sparse.
EvoStruct: Bridging Evolutionary and Structural Priors for Antibody CDR Design via Protein Language Model Adaptation
EvoStruct addresses vocabulary collapse in GNN-based antibody CDR design by combining a frozen protein language model with an E(3)-equivariant GNN through a cross-attention adapter. The method introduces progressive PLM unfreezing and R-Drop consistency regularization to recover functionally important amino acid diversity. On CHIMERA-Bench, EvoStruct improves sequence recovery by 16%, reduces perplexity by 43%, and achieves 2.3x greater amino acid diversity compared to the best GNN baselines.
OrbitForge: Text-to-3D scene generation via reconstruction-anchored video synthesis using Gaussian Splatting
OrbitForge is a new method for converting text-generated videos into 3D Gaussian Splatting scenes without task-specific fine-tuning or score-distillation optimization. The approach uses a frozen video diffusion model as a prior, performs an initial 3D reconstruction via Deformable Gaussian Splatting, detects missing viewpoints from a prescribed orbit, and completes only those views before final reconstruction. On a 300-prompt T3Bench-derived audit, OrbitForge achieves a 359-degree median orbit span and substantially improves coverage quality over a MedianGS-only baseline. The work also argues for coverage-aware evaluation metrics in text-to-3D tasks.
OneCanvas achieves state-of-the-art 3D scene understanding via panoramic reprojection in VLMs
OneCanvas is a new method for 3D scene understanding in Vision-Language Models that aggregates multi-view patch features onto a single equirectangular panoramic canvas using depth and camera pose, avoiding complex geometry encoders or large training budgets. A 3D position embedding restores metric depth information lost during angular projection, and a spatial pretraining curriculum generates on-the-fly supervision for spatial reasoning tasks. The approach achieves state-of-the-art results on SQA3D and VSI-Bench benchmarks while using an order of magnitude less training compute than competing methods, and supports situated reasoning relevant to robotics and embodied AI.
Hierarchical Text-Conditional Image Generation with CLIP Latents (DALL-E 2 / unCLIP)
OpenAI published research on hierarchical text-conditional image generation using CLIP latents, the technique underlying DALL-E 2. The approach uses a prior network to map text embeddings to image embeddings, then a diffusion decoder to generate images from those embeddings. This represented a significant advance in text-to-image generation quality and semantic fidelity at the time of release.
ALIGNBEAM: Training-free safety alignment transfer across model families at inference time
ALIGNBEAM is a training-free inference-time method that transfers safety alignment from a safe anchor model to a domain-fine-tuned target model, even when the two models have different vocabularies. It works by translating anchor logits into the target model's vocabulary token-by-token at each decoding step, then using a small LLM judge to select the safest among K candidate continuations. The method addresses a known vulnerability where domain fine-tuning degrades safety, and demonstrates substantial refusal improvements on adversarial benchmarks without retraining either model or incurring prohibitive inference overhead.