Free Reads
Sign in to view your remaining parses.
Tag Filter
Diffusion Transformer
SeedVR: Seeding Infinity in Diffusion Transformer Towards Generic Video
Restoration
Published:1/3/2025
Diffusion TransformerVideo RestorationLong Sequence Video ModelingSpatio-temporal Window AttentionCausal Video Autoencoder
SeedVR employs a diffusion transformer with shifted window attention, enabling efficient restoration of arbitrarylength and resolution videos. It supports variablesized spatialtemporal windows and integrates causal autoencoding and mixed training, outperforming prior methods o
06
Direct3D-S2: Gigascale 3D Generation Made Easy with Spatial Sparse
Attention
Published:5/23/2025
Spatial Sparse AttentionDiffusion TransformerSparse Volumetric Representation3D Generation FrameworkVariational Autoencoder (VAE)
Direct3DS2 employs Spatial Sparse Attention to efficiently generate gigascale 3D shapes using sparse volumetric data, combining a unified sparse VAE design that boosts training efficiency and stability while drastically reducing computational costs.
07
DreamClear: High-Capacity Real-World Image Restoration with Privacy-Safe
Dataset Curation
Published:10/24/2024
Real-World Image RestorationDiffusion TransformerPrivacy-Safe Dataset CurationText-to-Image Diffusion ModelsMultimodal Large Language Model Assisted Restoration
This work introduces GenIR for privacysafe, largescale dataset generation and DreamClear, a DiT model leveraging generative priors and multimodal LLMs for adaptive, highquality realworld image restoration.
07
Effective Diffusion Transformer Architecture for Image Super-Resolution
Published:9/29/2024
Diffusion ModelsImage Super-resolutionDiffusion TransformerMulti-Scale Hierarchical Feature ExtractionFrequency-Adaptive Time-Step Conditioning Module
DiTSR introduces a Ushaped diffusion transformer with frequencyadaptive conditioning, enhancing multiscale feature extraction and resource allocation, achieving superior superresolution without pretraining compared to priorbased methods.
08