Computer Vision news and analysis on Web Pulse: 1801 curated articles tracking the latest Computer Vision developments, tools, and research, updated continuously from vetted sources.
Researchers introduced Prompt2Effect, a hypernetwork that generates effect-specific LoRA weights for Image-to-Video diffusion models in a single forward pass, eliminating the need for per-effect training. The method achi…
Researchers at an undisclosed institution introduced CaricHarmony, a training-free method for sketch-based caricature synthesis that resolves destructive interference between identity and shape conditions in diffusion mo…
Researchers introduced the Overhead Wildlife Locator (OWL), a weakly supervised density-estimation framework for aerial wildlife surveys that uses point-level labels instead of costly bounding-box annotations. OWL-D set …
Researchers introduce Temporal Backtracking Search (TBS), a test-time scaling method for generative video reasoning that searches over the temporal axis rather than denoising steps. TBS transforms video generation into a…
Researchers introduced i2L (image-to-LoRA), a framework that compresses style LoRA training into a single forward pass, enabling immediate style instantiation from reference images without per-style optimization. The met…
Researchers introduced CineOrchestra, a unified video diffusion model that simultaneously controls subjects, events, cameras, and shot transitions for cinematic video generation. The model uses entity-centric conditionin…
Researchers introduce a morphology-aware sample assignment method to overcome IoU insensitivity in surface defect detection. The approach adds morphological similarity metrics to refine positive sample selection, achievi…
A study of six self-supervised geospatial foundation models found that model rankings change across tasks and adaptation settings, with task-relevant information more accessible in intermediate transformer blocks than fi…
Researchers propose RT-VLA, a lightweight vision-language-action model for autonomous driving that uses knowledge distillation from the larger SimLingo model to achieve real-time performance. RT-VLA reduces inference lat…
Apple announced visionOS 27 at WWDC 2026, the fourth major OS update for Apple Vision Pro, with features including held object tracking, IR LED tracked accessories, unfoveated screen recording, and a Siri AI overhaul pow…
UC Berkeley professor Hany Farid, a leading digital forensics expert, told the New York Times he struggles to distinguish AI-generated media from real content, calling it 'like going blind.' Farid is leaving Berkeley to …
A LEGO-style LoRA for Wan2.1 14B T2V, released on Hugging Face as Remade-AI/Lego, offers the most direct open-source route for generating LEGO-style video, though it requires a heavy base model and does not automatically…
Researchers Chen et al. published a paper on arXiv 2025 introducing Janus-Pro, a unified multimodal model that uses separate encoders for understanding and generation within a single transformer. The model scales data an…
Researchers Tian et al. introduced Visual Autoregressive Modeling (VAR), a new image generation method that predicts the next resolution instead of the next token, achieving state-of-the-art results on ImageNet and outpe…
Researchers Deng et al. published a paper on arXiv in 2025 introducing a unified multimodal pretraining model that understands, generates, and edits content, with multimodal reasoning emerging as data scales. Rudrite Res…
Huang et al. published a paper on arXiv 2025 introducing Self Forcing, a method that bridges the train-test gap in autoregressive video diffusion by rolling out the model on its own outputs during training, enabling real…
NVIDIA published Cosmos World Foundation Model Platform for Physical AI on arXiv in 2025, introducing pretrained world models that use diffusion and autoregressive techniques to simulate the physical world. A free intera…
A team of developers built Cosmos Claw, an AI-powered social media manager that uses NVIDIA Cosmos 3 and GPT-4o to autonomously film, edit, and post promotional videos for venues, deployed on Nebius H200 GPUs during a Sa…
ImageToolbox, a free and open-source image editing tool developed by T8RIN, offers over 310 filters, AI-powered background removal, and batch processing for photographers and developers. The app is available on GitHub an…
AnimateDiff Turbo scored 7.7 in a head-to-head comparison against Marey Realism V1.5, which scored 16.3, as Marey Realism V1.5 demonstrated superior understanding of temporal coherence and event progression in video gene…