Blog
AI Weekly 003

AI Weekly #003

🆕 What's new


Our team has developed a more user-friendly version of ComfyUI. You can view the product introduction here. If you are interested in our product, feel free to download (opens in a new tab) it for a trial. Also, join us on Discord (opens in a new tab) to get the latest news.

New tutorials added last week:

🤩 Weekly‘s AI highlights

📄 Noteworthy papers and technic

PhotoMaker: Utilizes multiple photos as identity IDs to extract individual characteristics and then generates a new, personalized character image. It can create person images that match descriptions and blend features from different individuals to create a completely new character. Additionally, it can modify the gender and age of photo subjects and generate various styles of photos, all with fast, realistic, and natural results.

It can convert noisy audio into clear and powerful speech, removing various noises and disturbances from the audio while preserving the clear speech. Not only can it eliminate noise, but it can also restore audio distortion and extend audio bandwidth. It makes the original sound clearer and more natural. For example, it can turn old cassette recordings into high-definition audio, making them sound more comfortable to listen to. When it comes to the restoration and recovery of historical recordings or archived audio, Resemble Enhance provides an effective solution, breathing new life into these audio recordings.

Adobe Premiere Pro has introduced text-based AI video editing features, harnessing the power of AI to make video editing faster and smarter. This includes features such as automatic video editing, content rearrangement, and support for voice transcription. It can automatically generate transcripts, simplifying the video editing process, allowing you to edit videos much like editing text documents. Additionally, it offers automatic color balancing and matching, automatic volume adjustment, noise reduction, and voice enhancement, as well as automatic subtitles, transitions, and aspect ratio adjustments.

Byrdhouse AI: A tool that can provide real-time translation of over 100 languages during video calls. It offers voice translation with AI-driven real-time interpretation, allowing you to convert your language into another during meetings or chats. You can even choose whether the translation voice is male or female. It automatically detects languages and translates, and you can edit the translated subtitles. It can also automatically generate meeting/call summaries in different languages.

🛠️ Products you should try

Cutout.pro is an AI-driven image enhancement product created by the LibAI team in Hangzhou. It offers features such as portrait cropping, style transfer, restoration, reverse image search, and more. Cutout Pro has risen to the global top ten in the category of image editing tools, particularly excelling in generating passport photos. On December 5, 2023, at the Baiwhale Annual Meeting in Shenzhen, its founding team, Li Bai AI Lab, was awarded the "2023 Enterprise Innovation" prize, thanks to its two major international projects, PromeAI (image generation from sketches to final images, concept images to design images) and Cutout.pro.


  • Puzzle (opens in a new tab):This is an AI financial company dedicated to revolutionizing the accounting industry from scratch using AI. For instance, it has rebuilt the core accounting ledger to make it smarter, more efficient, and adaptable to today's digitally-driven business needs. Its platform seamlessly integrates real-time data from various essential fintech tools, transforming static accounting data into real-time financial insights. If you are an accountant or an AI technology expert interested in financial services, you can take inspiration from Puzzle's approach to create an intelligent AI for accounting, similar to an upgraded version of domestic software like Yonyou or Kingdee Financials with AI capabilities. Of course, Yonyou and Kingdee won't stand still either, so you should be prepared to compete with these industry giants, using speed to overcome their scale advantages.

Bittensor is a trading marketplace built for algorithmic models, where the traded assets are the algorithmic models themselves, representing machine intelligence. Due to its combination of AI and Web3 concepts, it has achieved a ranking of 60th in the cryptocurrency market cap. Training artificial intelligence models currently requires significant amounts of data and computing power, making it a costly endeavor. As a result, once large companies and research institutions make breakthroughs, they often establish closed ecosystems, hindering the collaborative development of artificial intelligence.

Therefore, there is a need for decentralized, open-source solutions. Among the three core elements of AI: data, algorithms, and computing power, Bittensor focuses on the algorithmic aspect. It does this by utilizing blockchain networks and incentive mechanisms to schedule and select different algorithms, creating a marketplace where AI can thrive through free competition and knowledge sharing.


Botto is a generative AI artist born in October 2021, belonging to the Botto DAO (Decentralized Autonomous Organization). It was collectively trained by community members and is the world's first decentralized autonomous AI artist. Its NFT artworks have been auctioned on the crypto art platform SuperRare, generating sales revenue in the tens of millions of Chinese yuan.While it has become easy to prompt AI to generate art using keyword prompts, determining what art is valuable remains a complex question. AI itself cannot make such value judgments. Therefore, Botto creatively introduced a DAO mechanism where DAO members vote to decide what artworks are valuable ("crowdsourcing wisdom"). This data can then be used to train the AI artist to continuously generate new, high-quality works of art.



Subscribe for free to receive new posts and support my work. Or join our Discord.