Brief

Reallusion is integrating AI models into its iClone software to enhance 3D animation for filmmakers. This new approach allows artists to bypass traditional text-prompt limitations by utilizing AI models that offer more nuanced control. The integration specifically pairs iClone with ByteDance's Seedance 2.0, aiming to streamline the creation of professional-grade 3D content. AI

Researchers from ByteDance and HKUST have developed a new strategy for training AI models on long documents. Their approach, which utilizes multimodal question-answering, significantly outperforms traditional methods relying on raw OCR transcription. This advancement aims to improve AI's comprehension and processing of lengthy textual data. AI

A ByteDance study demonstrates that a 7B parameter model can effectively process and answer questions about lengthy, image-rich documents. This approach, which involves the model learning by answering questions and locating relevant passages, proved more reliable than traditional transcription methods, even for documents significantly longer than the model's training data. The research suggests this question-answering method enhances performance for large language models (LLMs) when dealing with extensive and multimodal content. AI

A new report from Quantum Bit Think Tank analyzes the evolving landscape of AI applications in China, shifting from simple chatbots to task-oriented agents. The report highlights a significant increase in AI application usage, with web traffic exceeding 900 million monthly visits and app downloads surpassing 240 million. Key trends include the rise of agents, the democratization of AI models, AI assistants becoming primary interfaces, the initial success of paid AI models, and the deepening penetration of AI in vertical business sectors. AI

ByteDance's Seedream 5.0 model generates high-quality images with impressive aesthetic detail and a deep understanding of photographic language. Users can reference specific film stocks, lenses, and lighting to achieve desired effects. The model also features an example-based editing capability, allowing users to demonstrate desired transformations with before-and-after image pairs. AI

ByteDance has released Lance, a new 3-billion parameter open-source multimodal model designed to run on consumer GPUs. This model can process both images and text, aiming to make advanced AI capabilities more accessible. Concurrently, the popular inference engine llama.cpp has received significant performance enhancements through Multi-Threaded Pipelining (MTP), which boosts local inference speeds. Additionally, a new open-source chat client called Horizon has been launched, offering cross-platform support for interacting with local models via Ollama, as well as cloud-based AI services. AI

Beijing University and SoundAI co-hosted a seminar on voice AI technology and industry development. The event showcased an AI Agent Mic, a joint innovation combining office and life applications, demonstrating real-time meeting transcription, summarization, and voice-activated assistance. This collaboration highlights the university's strategy to translate academic research into industrial applications and advance the formation of new quality productive forces in AI. AI

ByteDance has released Lance, an open-source model capable of both generating and understanding images and videos. The model is available on GitHub, allowing developers to integrate its multimodal capabilities into their applications. Lance aims to provide a unified solution for tasks involving visual data processing and creation. AI

ByteDance has released Lance, a new 3-billion parameter multimodal model, into the open-source community. This model is designed to process and understand both images and video content. The release aims to foster further development and innovation in the field of multimodal AI. AI

ByteDance's CapCut has partnered with Google Gemini to integrate its AI-powered video and image editing tools directly into the Gemini app. This collaboration aims to streamline content creation by allowing users to access CapCut's features without leaving the Gemini application. The integration is expected to enhance user experience by providing AI-assisted editing capabilities within a conversational AI interface. AI

ByteDance is testing paid subscription tiers for its AI application Doubao in China, offering premium features like PowerPoint generation and data analysis for power users. Separately, a new binary trust registry called GenGEO aims to enhance AI agent transactions by verifying merchant reliability in real-time. Additionally, Diploi utilizes AI to automatically detect and deploy complex monorepos and full-stack projects, differentiating itself from services focused on simpler front-end or serverless applications. AI