Trends · 研究趋势
数据来自 Hugging Face 论文的 AI 提取关键词,按月统计研究方向的增长与热度。
Ruisi Wang, Zhongang Cai, Fanyi Pu +11 authors
Diffusion-based video models demonstrate reasoning capabilities through denoising steps rather than frame sequences, exhibiting behaviors like working memory, self-correction, and perception-before-action within specialized transformer layers.
Shuai Yang, Wei Huang, Ruihang Chu +9 authors
LongLive is a frame-level autoregressive framework for real-time and interactive long video generation, addressing efficiency and quality challenges through causal attention, KV-recache, streaming long tuning, and short window attention.
Yoav HaCohen, Benny Brazowski, Nisan Chiprut +26 authors
LTX-2 is an open-source audiovisual diffusion model that generates synchronized video and audio content using a dual-stream transformer architecture with cross-modal attention and classifier-free guidance.
NextStep Team, Chunrui Han, Guopeng Li +47 authors
NextStep-1, a 14B autoregressive model with a 157M flow matching head, achieves state-of-the-art performance in text-to-image generation and image editing by processing discrete text tokens and continuous image tokens.
Jiyuan Wang, Chunyu Lin, Lei Sun +8 authors
RL3DEdit uses reinforcement learning with rewards from a 3D foundation model to achieve multi-view consistent 3D editing from 2D editing priors.
Xiangyan Qu, Zhenlong Yuan, Jing Tang +9 authors
Image-CoT methods are extended to image editing with ADE-CoT, which improves efficiency and performance through adaptive resource allocation, edit-specific verification, and opportunistic stopping mechanisms.
Aleksandr Razin, Danil Kazantsev, Ilya Makarov
LUA is a lightweight module that performs super-resolution directly in the latent space of diffusion models, improving efficiency without compromising image quality.
Shuai Yang, Yifan Zhou, Ziwei Liu +1 authors
A novel framework adapts image diffusion models for video by generating key frames with hierarchical constraints and propagating them using patch matching and blending, achieving high-quality and temporally-coherent videos.
Yijing Tu, Shaojin Wu, Mengqi Huang +4 authors
Stream-T1 enables efficient test-time video generation through streaming synthesis with temporal guidance, achieving improved consistency and quality over existing diffusion-based methods.
Kai Wang, Zhaopan Xu, Yukun Zhou +4 authors
Diffusion models can generate high-performing neural network parameters using an autoencoder, producing new subsets of network parameters with comparable or improved performance.
Justin Cui, Jie Wu, Ming Li +6 authors
A method is proposed to enhance long-horizon video generation by using sampled segments from self-generated long videos to guide student models, maintaining quality and consistency without additional supervision or retraining.
Haoyu Zhao, Xingyue Zhao, Siteng Huang +3 authors
A multi-modal 4D world model generates synchronized RGB, depth, and optical flow data from single RGB-D images and language instructions, enabling efficient robotic manipulation through unified diffusion processes and inverse dynamics policy learning.
Daniil Selikhanovych, David Li, Aleksei Leonov +6 authors
RSD is a distillation method for ResShift, a top diffusion-based super-resolution model, that achieves high-quality results with fewer parameters and GPU memory compared to other methods.
Yuchao Gu, Yipin Zhou, Yunfan Ye +5 authors
ROICtrl, an adapter for diffusion models using ROI-Align and ROI-Unpool, improves regional instance control in text-based visual generation with lower computational costs.
Kwanyoung Kim, Byeongsu Sim
PLADIS leverages sparse attention in cross-attention layers to enhance pre-trained text-to-image diffusion models, improving text alignment and human preference without additional training.
Runhui Huang, Qihui Zhang, Zhe Liu +3 authors
SpectraReward leverages pretrained multimodal language models as training-free reward models for image-generation reinforcement learning by measuring prompt recovery likelihood, with a self-improving variant that uses the policy's own understanding branch to close the loop.
Bingnan Li, Haozhe Wang, Haozhong Xiong +5 authors
On-policy diffusion distillation under classifier-free guidance suffers from branch-level error compensation, which is resolved by separately matching positive predictions and guidance directions.
Xiangpeng Yang, Linchao Zhu, Hehe Fan +1 authors
VideoGrain uses space-time attention mechanisms to enhance fine-grained video editing by improving text-to-region control and feature separation.
Lizhen Wang, Yongming Zhu, Zhipeng Ge +15 authors
FlowAct-R1 enables real-time interactive humanoid video generation with high-fidelity synthesis and low-latency responsiveness through MMDiT architecture and chunkwise diffusion forcing strategies.
Lital Binyamin, Yoad Tewel, Hilit Segev +3 authors
CountGen enhances text-to-image diffusion models by accurately controlling the number of depicted objects using features that carry object identity information, surpassing existing methods on benchmark datasets.
北京市昌平区探索星信息技术及软件开发工作室
京ICP备2026059466号