# 3D İlgili Makaleler

HTX Haber Merkezi, kripto endüstrisindeki piyasa trendleri, proje güncellemeleri, teknoloji gelişmeleri ve düzenleyici politikaları kapsayan "3D" hakkında en son makaleleri ve derinlemesine analizleri sunmaktadır.

Show me 'The Lord of the Rings', Karpathy Recommends New Benchmark for Large Model Evaluation

In a new benchmark for evaluating large language models, Andrej Karpathy proposes replacing the once-popular "pelican riding a bicycle" SVG test with a more complex challenge: generating a 3D scene from the opening text of *The Lord of the Rings*. Using Anthropic's Opus 5 model and the Three.js library, the task consumed approximately 1 million tokens, 2 hours, and 5,500 lines of code to produce a rudimentary, low-polygon animation of the Shire. While the output is visually crude with notable glitches like floating characters, it demonstrates the model's ability to parse narrative text and translate it into a functional, programmatic 3D world with defined objects, cameras, lighting, and basic animation. This "Lord of the Rings benchmark" is argued to test a model's capacity for long-horizon project planning, spatial reasoning, and maintaining consistency across thousands of code lines—capabilities not fully captured by simpler single-output tests. The initiative has sparked community experimentation, with users generating other 3D worlds like a low-poly San Francisco, a data-driven New York City model, and even a virtual Kanye West concert. Karpathy suggests a future pipeline where code-generated scenes provide the structural "bones" for video-to-video models to enhance visual fidelity. While some debate the computational cost and specificity to Three.js, proponents see it as a test of a model's general ability to structure its understanding of the world into an executable form. The shift signals a move towards evaluating how well models can not only generate code or images but also comprehend and construct interactive, multi-element digital environments.

marsbit2 gün önce 08:54

Show me 'The Lord of the Rings', Karpathy Recommends New Benchmark for Large Model Evaluation

marsbit2 gün önce 08:54

Claude Opus 5 Leaks, First Wave of User Tests Arrive

Claude Opus 5 appears to have been leaked, with early test results circulating online. Users report generating highly detailed 3D scenes, such as a catapult attacking a castle with intricate parameter cards, dynamic weather interfaces, and realistic kitchen renders. Comparisons show Opus 5 outperforming the current Fable 5 in detail density for similar prompts. Other demos include impressive Minecraft recreations and detailed SVG graphics. Evidence of the leak includes sightings in Cursor's model selector (under the code "Honeycomb EAP") and Google Vertex AI, followed by reports of the model appearing for some users, though officially labeled as version 4.8. Speculation is growing that Opus 5 could match Fable 5's capabilities at half the price per token, though concerns exist about potentially higher token consumption. The model's formal release seems imminent.

marsbit07/24 07:53

Claude Opus 5 Leaks, First Wave of User Tests Arrive

marsbit07/24 07:53

1600 Lines of Code Create an Underwater Manhattan, Fable 5 Leaves Karpathy Stunned

Title: Fable 5 Stuns Karpathy with 3D Worlds Built from 1600 Lines of Code An AI model, Fable 5, has demonstrated a remarkable leap in generating complex, interactive 3D worlds with minimal code. Showcased by Peter Gostev of Arena.ai, the model created 63 diverse 3D environments across themes like immersive cityscapes, explorable famous paintings, natural wonders, and cosmic phenomena. Many were generated in a single attempt. A standout creation is a detailed, submerged Manhattan built with only 1600 lines of Three.js code. Other highlights include traversable versions of Van Gogh's "Starry Night," a bear catching a salmon with realistic physics, and a split Red Sea. The model's ability to cohesively manage vast numbers of elements within a scene represents a significant technical advancement. Andrej Karpathy, who recently joined Anthropic's pre-training team, expressed amazement, particularly at how the model intuitively understood and rendered complex real-world interactions like a fish struggling when caught. He coined the term "fablemaxxing" to describe this qualitative leap. While Fable 5 excels at world-building, Gostev notes current limitations in creating engaging, long-form gameplay. The model also sometimes requires prompting to fully utilize its capabilities. Having topped the Agent Arena benchmark for real-world task completion, Fable 5 signals that the boundaries of AI-generated content are rapidly expanding, with its full potential yet to be discovered.

marsbit07/06 09:48

1600 Lines of Code Create an Underwater Manhattan, Fable 5 Leaves Karpathy Stunned

marsbit07/06 09:48

SemiAnalysis Dissects Huawei's Kirin 9030: Process Technology Halted, So They Folded the Chip

SemiAnalysis has published a detailed teardown report on the HiSilicon Kirin 9030 Pro chipset found in Huawei's Mate 80 Pro. Fabricated using SMIC's most advanced N+3 node without EUV lithography, the analysis reveals significant technical achievements and strategic shifts. The report indicates SMIC's N+3 has achieved transistor density comparable to TSMC's N6 (113.4 vs 107.7 MTr/mm²), primarily through aggressive use of Self-Aligned Quadruple Patterning (SAQP) for its metal layers. This results in a notably small 32.5nm M0 metal pitch. However, SemiAnalysis notes this achievement comes with significantly higher process complexity, cost, and potential yield challenges compared to competitors using more advanced tools. The Kirin 9030 design maximizes this constrained density. While its GPU performance has improved ~70% and matches Qualcomm's 2022 flagship level, the CPU core's IPC lags behind current top-tier designs from Apple and Qualcomm, a gap attributed to the underlying manufacturing technology rather than design capability. Facing long-term restrictions on advanced tools, Huawei is charting a new path. The report highlights the company's "LogicFolding" roadmap, a 3D stacking technique aimed at shortening signal paths to boost performance and efficiency. The goal is to reach 5GHz frequency and a projected density of 295 MTr/mm² by 2031. SemiAnalysis concludes that export controls have not halted China's chip progress but have fundamentally altered its trajectory, making it more expensive and complex. This has spurred innovation in alternative areas like 3D stacking and domestic EDA tool development, with Huawei's supply chain also beginning to integrate Chinese memory from CXMT.

marsbit06/15 06:52

SemiAnalysis Dissects Huawei's Kirin 9030: Process Technology Halted, So They Folded the Chip

marsbit06/15 06:52

After being noticed by IBM, the token of three.ws surged 50 times

IBM's public acknowledgment and collaboration with Solana AI project three.ws sparked a massive 50x surge in its $THREE token. Three.ws is building a "3D Agent Layer" for the internet, aiming to move AI beyond chat interfaces into interactive 3D digital characters that can be embedded on websites. These agents possess bodies (3D models), brains (LLMs), memory, skills, and optional on-chain identities and wallets for payments. The platform's architecture consists of four layers: a Viewer layer for 3D rendering, an Agent layer for AI behavior and skills, an optional Identity layer for persistent profiles, and an Embedding layer for deployment. For monetization, developers can set per-use fees for agent interactions, paid in USDC via the x402 protocol. Key to its enterprise strategy are partnerships with IBM and AWS. Integration with AWS Marketplace provides enterprise billing and deployment channels, while the IBM collaboration aims to combine three.ws's 3D agents with IBM's enterprise AI, cloud services, and Granite models. The challenge remains whether businesses will adopt these embodied AI agents and what utility the $THREE token will ultimately hold within the ecosystem.

foresightnews_api06/05 04:27

After being noticed by IBM, the token of three.ws surged 50 times

foresightnews_api06/05 04:27

three.ws Joins Forces With Google Cloud: Powering Browser-Native 3D On-Chain AI Agents

The open-source platform three.ws, which provides the 3D layer for on-chain AI agents, has partnered with Google Cloud to advance browser-native, embodied AI. This collaboration aims to move AI beyond text chat boxes by enabling agents to have 3D avatars in the browser without plugins or installs. Developers can connect these avatars to LLMs like Claude, endowing them with long-term memory, real-time emotions, and composable tools. Agents can be minted on-chain (via Solana or EVM) for verifiable ownership, wallets, and portable reputation. The platform includes a production-ready Model Context Protocol (MCP) server, listed in Anthropic's official registry, allowing AI assistants to control 3D avatars and execute tools via natural language. Google Cloud will provide the computational scale for real-time 3D rendering and LLM inference, global reach for low-latency interactions, and a trusted enterprise foundation. three.ws is already integrated across the ecosystem, including the Anthropic MCP Registry and Alibaba Cloud Marketplace, and is fully open source.

TheNewsCrypto05/27 13:09

three.ws Joins Forces With Google Cloud: Powering Browser-Native 3D On-Chain AI Agents

TheNewsCrypto05/27 13:09

From a Lunch Table to an Infinite Universe: Fei-Fei Li Bets on AI's Next Dimension

From a Lunch Table Conversation to an Infinite Universe: Fei-Fei Li Bets on AI's Next Frontier - Spatial Intelligence In an era dominated by large language models, AI pioneer Fei-Fei Li argues that true understanding requires spatial intelligence — the ability to perceive, reason, and interact within the physical 3D/4D world. She points to evolutionary history: spatial perception drove the Cambrian explosion 540 million years ago, while language is a far more recent, inherently "lossy" way to encode reality. Current models struggle with basic spatial tasks a child can do, like counting chairs in a video. Her company, World Labs, is pioneering this shift with "Marble," a model that generates navigable, consistent 3D worlds from text, images, or simple 3D inputs—distinct from video generators like Sora. Though smaller than models like GPT-5, due to scarce 3D data and early-stage scaling laws, Marble is already used in gaming, robot training (by NVIDIA), architectural design, and personalized therapy for conditions like OCD and acrophobia. Li envisions this technology enabling "infinite universes" for creativity, social interaction, and more. However, she cautions against utopian or dystopian extremes, advocating for a measured vision where AI enhances human dignity and prosperity, akin to how electricity transformed civilization. The journey is long — as evidenced by the 20-year path to viable autonomous vehicles — but the direction is clear: for AI to move from merely talking about the world to truly understanding and acting within it.

marsbit05/27 00:14

From a Lunch Table to an Infinite Universe: Fei-Fei Li Bets on AI's Next Dimension

marsbit05/27 00:14

活动图片

Bitcoinright-arrow

Ethereumright-arrow