Welcome to the AI Daily section! Here is your daily guide to exploring the world of artificial intelligence. Every day, we bring you the hottest content in the AI field, focusing on developers to help you understand technical trends and innovative AI product applications.
Discover the latest AI products here: https://top.aibase.com/
1. The AI Video King Returns! Runway's New Gen3 Model Impresses Netizens Again
This article introduces Runway's latest video generation model, Gen-3 Alpha, which has significant improvements in fidelity, consistency, and action performance, taking an important step towards building a general-purpose world model. Gen-3 Alpha boasts several notable features and characteristics, becoming a rising star in the creative industry.
AiBase Summary:
⭐️ Gen-3 Alpha significantly improves in fidelity, consistency, and action performance, capable of generating expressive, realistic human characters.
⭐️ Supports various generation tools, such as text-to-video, image-to-video, and text-to-image conversion tools.
⭐️ Can perform fine temporal control, supporting various advanced control modes, including motion brushes, advanced camera controls, and director mode.
⭐️ Extremely stable lighting, maintaining high-quality output even in fast-moving scenes.
View more videos here: https://mp.weixin.qq.com/s/5LbM0NfkeiYFU0r4VDqpYA
Official website: https://top.aibase.com/tool/gen-3-alpha
2. Luma AI Releases Extend Feature, Extending Video Duration Beyond 10 Seconds
Luma AI has updated its Dream Machine video model, adding the Extend feature, which can extend the video duration beyond 10 seconds while maintaining the original video style and consistent characters. Although the Extend feature takes longer to generate extended videos, the style consistency is well-maintained.
AiBase Summary:
✨ Dream Machine updates with the Extend feature, extending video duration beyond 10 seconds, maintaining the original video style and consistent characters.
⏱️ Extending videos with the Extend feature takes longer, but the style consistency is good.
3. DeepSeek Releases Open-Source Model DeepSeek-Coder-V2
DeepSeek recently released the open-source model DeepSeek-Coder-V2, which surpasses GPT-4-Turbo in code and math capabilities, boasting global leading performance. The model adopts a MoE architecture, supporting multiple languages and longer context processing lengths. Users can use it for free in commercial applications without applying.
AiBase Summary:
🚀 Model performance is globally leading, especially adept at code generation and math arithmetic.
💡 Supports 338 programming languages and 128K context length, meeting more development needs.
🔗 Provides API services, priced the same as DeepSeek-V2, performing excellently in benchmark tests.
4. Adobe Acrobat Receives Major AI Upgrades, Supporting Multi-Document Analysis and Image Generation
Adobe is set to roll out a series of significant AI upgrades, enhancing Acrobat's AI assistant functionality and image generation capabilities, ensuring data privacy protection. This update will greatly improve office efficiency, bringing convenience to handling large volumes of documents and optimizing visual content.
AiBase Summary:
🚀 AI assistant functionality upgrades, supporting multi-document analysis and inquiries, enhancing user experience.
🖼️ New AI image generator added, users can generate new images or edit images in existing PDFs.
🔒 Commitment to data privacy protection, documents uploaded to the cloud for analysis but not used to train AI models, prohibiting third-party use.
5. Apple Releases 20 Core ML Models on Hugging Face Platform
Apple has released 20 new Core ML models and 4 datasets on the Hugging Face platform, showcasing significant progress in advancing AI development. This update includes exciting new models focused on text and images, as well as a wide range of applications, such as image classification, monocular depth estimation, and semantic segmentation. Apple emphasizes the importance of on-device AI, running optimized models on user devices to improve application performance while ensuring user data security and privacy.
AiBase Summary:
🚀 Apple releases 20 new Core ML models and 4 datasets on the Hugging Face platform, advancing AI development.
💡 New Core ML models cover a wide range of applications, including image classification, monocular depth estimation, and semantic segmentation.
🔒 Apple emphasizes the importance of on-device AI, running optimized models on user devices to improve application performance and ensure user data security and privacy.
Details: https://huggingface.co/apple
6. ElevenLabs Open-Sources Video Generation Sound Effects Tool, Uploading Videos Automatically Adds Voiceovers
ElevenLabs, a company specializing in audio generation technology, recently announced its entry into the video generation field, open-sourcing a project that can automatically add voiceovers to uploaded videos, generating suitable sound effects. They have introduced a new feature that allows users to generate various realistic music effects by inputting text, providing significant help to the film, gaming, and short video industries. In addition to sound effect generation, it also offers powerful features such as voice cloning and text-to-speech.
AiBase Summary:
🔊 Automatically adds voiceovers to uploaded videos, generating suitable sound effects
🎶 Input text to generate various realistic music effects, helping the film, gaming, and short video industries
🎤 Provides voice cloning and text-to-speech features, giving content a more vivid form of expression
Text-to-audio entry: https://top.aibase.com/tool/elevenlabs-wenbenzhuanyinxiaoapi
Video automatic dubbing entry: https://top.aibase.com/tool/elevenlabs-texts-to-sounds-effects-api
7. Tencent WeChat Video Account Plans to Restrict Digital Human Livestreaming Sales
Tencent's Video Account recently announced revisions to the "Video Account Showcase Talent Low-Quality Content Implementation Rules," aimed at strengthening content quality regulation and planning to prohibit digital human livestreaming sales. The revisions were open for public comment from June 7 to June 13 this year.
AiBase Summary:
⭐ Revisions aim to strengthen Video Account content quality regulation
⭐ Prohibits digital human livestreaming sales, clearly banning non-real livestream content
⭐ Platform will take corresponding punitive measures against violators
8. Stability AI's SD3 Faces Opposition Due to Licensing Issues, CivitAI Community Bans Related Content
Stability AI's latest major model, SD3, has sparked controversy over licensing issues, facing opposition from the AI community. The CivitAI community has banned content related to SD3, sparking licensing agreement disputes. The company has introduced a creator's license for consumers, restricting developer conditions and the number of images generated. SD3 has issues with not being able to generate specific human poses, and its future is uncertain. The CEO's departure and layoffs require the company to explain the impact of the new licensing agreement. The entire controversy has potential implications for the AI community and the development of open-source models.
AiBase Summary:
💥 SD3 licensing issues spark controversy, facing opposition from the AI community.
🔒 The company introduces a creator's license, restricting developer conditions and the number of images generated.
❓ SD3 has issues with not being able to generate specific human poses, and its future is uncertain.
9. LEGO Printer Pixelbot 3000
This article introduces a LEGO printer called Pixelbot3000 designed and manufactured by YouTube channel creator @Creative Mindstorms, using custom code and AI to generate LEGO mosaics. Users only need to input the name of an artwork, and after AI generates the image, Pixelbot3000 automatically assembles the mosaic.
AiBase Summary:
🤖 Pixelbot3000 can automatically generate LEGO mosaics using custom code and AI, simplifying the printing process.
🎨 Pixelbot3000 uses OpenAI's DALL-E3 to generate simplified cartoon-style images, ultimately producing high-contrast scaled images.
🔧 Pixelbot3000 divides the AI-generated image and samples the color of the center pixel of each square to get a better mosaic pattern.
10. Researchers Teach AI to Recognize Human Line Sketch Sketches
This article introduces a new method developed by the University of Surrey and Stanford University research team to teach AI to understand the importance of human line sketches and the results achieved. By combining sketches and text descriptions, AI demonstrates an understanding ability close to that of humans, accurately identifying and labeling objects in complex scenes. This research brings new possibilities for human-computer interaction and design workflows.
AiBase Summary:
🧠 AI learns to understand the importance of sketches, demonstrating performance close to that of humans
🌳 AI can identify and label objects such as kites, trees, and giraffes with 85% accuracy, outperforming other models
🎨 The new method is not only applicable to sketches drawn by non-artists but also to sketches drawn by objects without explicit training
Details: https://arxiv.org/abs/2312.12463
11. Research: AI-Generated Images Fail to Accurately Represent the Nuances of Islamic Architectural Culture
AI has brought revolutionary changes to the field of architectural design, but in culturally sensitive areas such as Islamic architecture, AI-generated images fail to correctly represent historical elements. Research points out that AI generators lack historical knowledge and recommends cautious use. The author believes that AI can be a valuable tool but needs to be combined with human expertise and cultural sensitivity.
AiBase Summary:
🏗️ AI brings revolutionary changes to architectural design but faces challenges in areas like Islamic architecture.
🕌 AI generators lack historical knowledge, failing to accurately represent Islamic architectural cultural details.
🤖 AI should be used as a tool to enhance human creativity, combining professional knowledge and cultural sensitivity.
