AccDiffusion: An Accurate Method for Higher-Resolution Image Generation
Solving Object Repetition in High-Resolution Image GenerationAccDiffusion addresses the issue of object repetition in patch-wise higher-resolution image generation.The method uses patch-content-aware prompts and dilated...
Diffree: Text-Guided Shape Free Object Inpainting with Diffusion Model
Innovating Object Addition in Images with Text Guidance AloneDiffree enables seamless text-guided object addition without compromising background consistency.The model leverages the OABench dataset, enhancing...
ViPer: Visual Personalization of Generative Models via Individual Preference Learning
Tailoring AI-Generated Images to Individual TastesViPer personalizes image generation by capturing and applying individual visual preferences.The system uses user comments to infer visual likes...
HumanVid: Demystifying Training Data for Camera-Controllable Human Image Animation
A New Benchmark for Realistic Human Image Animation and Camera ControlHumanVid introduces the first large-scale, high-quality dataset tailored for human image animation, combining real-world...
T2V-CompBench: Setting a New Standard for Compositional Text-to-Video Generation
Introducing the First Comprehensive Benchmark for Complex Video Generation from Text PromptsT2V-CompBench offers the first benchmark tailored for compositional text-to-video generation.The benchmark includes diverse...
MovieDreamer: Hierarchical Generation for Coherent Long Visual Sequences
Revolutionizing Long-Form Video Generation with MovieDreamerMovieDreamer combines autoregressive models and diffusion rendering for long-duration video generation.The framework ensures narrative coherence and character consistency across...
OutfitAnyone: Ultra-high Quality Virtual Try-On for Any Clothing and Any Person
Revolutionizing the Virtual Fashion ExperienceOutfitAnyone utilizes a two-stream conditional diffusion model for lifelike virtual try-on experiences.The technology adapts to various body shapes and poses,...
Apple Showcases Open AI Capabilities with New Models
New Apple AI Models Outperform Competitors Mistral and Hugging FaceApple releases DCLM models on Hugging Face, featuring 7 billion and 1.4 billion parameter variants.The...
Click-Gaussian: Revolutionizing 3D Segmentation
Real-Time, Fine-Grained 3D Scene Manipulation Made PossibleClick-Gaussian enables rapid and accurate segmentation of 3D Gaussians.The Global Feature-guided Learning (GFL) method enhances segmentation accuracy.The method...
IBM: Scaling Granite Code Models to 128K Context
Expanding Context Windows in Open-Source Code ModelsIBM introduces Granite code models supporting up to 128K token context windows.Lightweight continual pretraining and instruction tuning enhance...
AI Enhances Doctor-Patient Communication
Study shows AI-generated responses outperform human doctors in empathy but raise readability concernsSuperior Empathy: AI-generated responses rated higher in empathy compared to those written...
Advancing Humanoid Control for Diverse Object Manipulation
New method enhances simulated humanoid's ability to grasp and transport varied objectsInnovative Control Method: Introduces a controller for simulated humanoids to grasp and follow...