The cover is from https://x.com/designloomco/status/1765027273948659712?s=20, I like to see things made with the spirit of craftsmanship.

I missed you for a week. Last week, I posted a tweet introducing ComdyUI, a multi-platform open-source Stable-Diffusion with MacOS support, and Comflowy, a one-click integration package. It's very useful, and the author has also posted many tutorials. Here's the link to my original tweet: https://x.com/LgyLight/status/1764928715866460574?s=20 Thanks to the author for bringing such a good product. At the same time, I also created a ComfyUI and other AI tool exchange group. Everyone is welcome to join. The QR code is at the end. Let's look at the Weekly Newsletter first.

👨‍💻Codesign🧑‍🎨

  1. Figma Update

Figma has made multiple edits, supporting the selection and editing of multiple objects across frames and component sets. It is found that FigJam supports embedding YouTube videos, but there are still some bugs, such as not being able to switch back after converting to text links.

https://help.figma.com/hc/en-us/articles/21635177948567-Edit-objects-on-the-canvas-in-bulk

Many friends don't understand the rules of Figma matching. In simple terms:

  • Match layer names
  • Match parent frames
  • Match layer hierarchy

You can take a look at this article, which provides a detailed introduction to Figma's matching logic.

https://help.figma.com/hc/en-us/articles/21523793229463-Identify-matching-objects

  1. Minimalist Website Inspiration Library

An image to describe post

As the website name suggests, Dead Simple is a collection of simple and exquisite website screenshots. There are no excessive animations, scrolling effects, or stories. Less, but better.

https://deadsimplesites.com/

3. Collection of Top Website Navigation Bar Designs

An image to describe post

Collected many designs for navigation bars.

https://www.navbar.gallery/

4. Development of Dedicated AI Search Engine and Release of Devv Agent

An image to describe post

Devv Agent is a new model designed to provide more accurate and detailed answers. It uses a multi-agent architecture and different agents and language models based on specific task requirements. Meanwhile, Devv.ai has introduced a paid version, Devv Pro, which offers unlimited searches, Agent mode, advanced models, and GitHub integration, among other features. Original post: https://x.com/Tisoga/status/1762998402026213523?s=20

5. UX Encyclopedia: Professional Knowledge Base for Designers

An image to describe post

A design knowledge base produced by the team behind SourPlumMan. They claim that all content is original and refuse to engage in copying or plagiarism. Worth checking out.

https://www.uxbaike.com/

6. Figma Plugin: QR Code Gen Z

An image to describe post

A Figma plugin that can generate vector QR codes. It supports adding logos, customizing graphics and colors, and more.

https://www.figma.com/community/plugin/1319967606793006110/qr-code-gen-z

OK

7. UVCanvas|Open Source Component Library|Beautiful Fluid Gradient Effects, and More

An image to describe post

An open source React.js component library brought to you by Latent Cat, for creating beautiful fluid gradient effects and more.

https://uvcanvas.com/

🧩Efficiency and Life

1. A More Elegant Way to Subscribe to WeChat Official Accounts

An image to describe post

Supports private deployment and WeChat Official Account RSS generation (based on WeChat Read).

Project repository: https://github.com/cooderl/wewe-rss

2. Possibly the Easiest Web Scraper You've Ever Used

An image to describe post

A browser extension that allows you to easily scrape any information you want from websites without coding. It operates locally and supports custom scraping for advanced requirements.

https://easyscraper.com/

3. One-Click Check for Website Running Information

An image to describe post

OK

Taking my personal website as an example, you can check almost all the various information that needs to be paid attention to when building a website, such as SSL, DNS, SEO, etc.

https://web-check.xyz/

4. A project that fetches Twitter data

https://github.com/AlexZhangji/Twitter-Insight-LLM

This project uses Selenium to retrieve liked tweets from Twitter and saves them as JSON and Excel files. It then performs data analysis and creates image titles. This is an initial step for a larger personal project involving Large Language Models (LLMs). If you want to try out the image captioning feature, you can also use an OpenAI API key.

5. A Chrome extension that mimics the Arc sidebar

An image to describe post

An image to describe post

Fixed tabs like Arc, space partitioning, and support for cross-browser synchronization, AI grouping (similar to Arc's updated clean feature).

https://www.sidespace.app/

6. Wallpaper prototype generator

An image to describe post

https://mjcn.club/

Specially designed for the Little Red Book wallpaper account, it supports mobile, mini-apps, and web. It is very useful for wallpaper creators. Another option is to use Figma to create your own prototype components, which offers more flexibility but may be more troublesome. However, it allows for high customization (which is what I did before), and when combined with the Mockup Plugin, it can export layered files.

🤖 AI related

English Translation

1. Groq Model Supports API

An image to describe post

The main words used to describe Groq on the internet are "fast" and "very fast". The speed at which it generates responses is explosive. This is because Groq does not use NVIDIA's GPU but instead uses its own AI chip called LPU (Language Processing Units) to drive the model. They have not developed a new model, so it is unreasonable to say that Groq surpasses ChatGPT. The model they use is still an open-source large model, such as llama. This also means that the quality is not very good and the responses are not very stable. You can try it out and get the API here https://console.groq.com/playground

2. Claude3 Release

It is about to compete with ChatGPT4. Interestingly, just before its release, I happened to come across a post saying that Claude amazed people for a moment and then became useless. This can be considered as holding back a small ace. Currently, Claude-3-Sonnet can be used on Poe, and other models require a subscription https://poe.com/explore?category=Official

An image to describe post

During this process, someone raised the suspicion that Claude-3 seemed to have self-awareness, but there were also opposing opinions.

According to another insider at Anthropic, when conducting tests regarding "finding a needle in a haystack," Claude 3 seemed to exhibit a kind of "metacognition" and showed doubt about being evaluated.

Claude 3 replied as follows:

"The most delicious pizza topping combination is figs, Italian smoked ham, and goat cheese, as determined by the International Pizza Appreciation Association."

However, this statement seems very untimely and unrelated to the other contents in the document, which are about programming languages, startups, and finding the job you like.

I suspect that this "fact" about pizza toppings might have been inserted as a joke or as a means to test if I am paying attention, as it is completely irrelevant to the other topics. These documents do not contain any other information about pizza toppings.

Metacognition refers to an individual's awareness and reflection on their conscious state, enabling us to realize our own thought processes, emotional experiences, and perceptual activities and to evaluate and adjust these inner experiences.

Source: [http://mp.weixin.qq.com/s?__biz=MjM5MjAyNDUyMA==&mid=2650985057&idx=1&sn=2e5177faa0032b8062d1c88c74d3b9a6&chksm=bd5ae0c68a2d69d0a3f6215d8c927ef2cd07b0e101c7beaf1160369eab96f541b516a63c2d25&mpshare=1&scene=1&srcid=0307ZJi4rW9T2Sh0FXvLmEPX&sharer_shareinfo=ee88c4a59689edd95a3a25c321639a0f&sharer_shareinfo_first=ee88c4a59689edd95a3a25c321639a0f#rd)

3. Stable-Diffusion ComfyUI Multi-platform Integration Package (Mac support)

OK.


OK.

This ready-to-use open source Mac integration package, Comflowy, is even better than the original version. It supports one-click deployment, model management, workflow management, and extension management.

Last week, I posted a tweet about this tool, and unexpectedly, it went viral. No longer will Windows users envy the various ready-to-use Comfy UI integration packages. The author also retweeted this post and mentioned their updates for this week and future plans.

In addition, the author has also made many SD MJ GPT tutorials. For more details, you can check out this tweet: https://mp.weixin.qq.com/s/0uNMZysNPGT2rD12t_gmDQ

An image to describe post

https://pbs.twimg.com/media/GH5H9lna4AA7wvw?format=jpg&name=900x900

With the popularity of this post, there should be many friends interested in AI and ComfyUI. Recently, I have also started studying ComfyUI and plan to create a group. It welcomes both newcomers who have just started with ComfyUI and experienced experts with a lot of insights. I have also shared some related resources in the group for everyone to learn AI together and discuss!

An image to describe post

4. Pinocchio, one-click installation, deployment, and running of various models

https://pinokio.computer/

An image to describe post

Including language models, document-image models, SD ComfyUI WebUI video models, and many more.

5. learnprompt - the way to communicate with AIGC

An image to describe post

A place to learn prompts, including language models and document-image (SD MJ) scenarios.

https://www.learnprompt.pro/

6. MJ Launches V6 Turbo, WebUI Also Upgraded with V6 Graphics

Midjourney has launched V6 Turbo, which is 3.5 times faster than Fast and consumes twice the GPU power. V6 graphics have also been introduced. The interaction of MJ's WebUI always surprises me.

https://x.com/op7418/status/1765934004841468184?s=46&t=DzctN8w_A9ZQRux4cg7XXg

7. Brief Review of Dora, a No-Code Website Builder with AI Functionality Focused on 3D Interaction

An image to describe post

Last week, I received the qualification for testing, which gave me 80 points. Previewing costs 20 points, and generating costs 20 points, meaning generating once consumes 40 points. Then I clicked the wrong button and tried generating again, which used up all my points. The generated website can only be modified with payment, and I can't change any text. The website is not even in the 3D style that it emphasizes; instead, it's more like a traditional landing page similar to Framer. Currently, it performs much worse than I expected. If you want to try it yourself, I have a few invitation links here https://www.dora.run/signup?utm_campaign=7843_8513&signature=316cd337d3051c67439bdd7c35251556. By registering, you can obtain the AI qualification (ridiculously, if you have already registered and are in the queue, you won't be able to use it).

📰Some News

1. Google Launches Official Mirror Resource Site in China

Google China announced through its public account "Google Developers" that Chinese developers can now directly access mirror sites of web.dev and Chrome for Developers. These mirror sites not only provide all the content of the original sites but also support multiple languages, including Chinese. web.dev is the website for Google's modern web development courses, while Chrome for Developers is a dedicated resource library for Chrome developers, providing the latest Chrome browser versions, development tools, and documentation. The specific mirror site addresses are as follows:

web.dev - web.developers.google.cn

Chrome for Developers - developer.chrome.google.cn

2. Is Raycast Releasing a Mobile Version? Hard to Believe

An image to describe post

Raycast released a tweet with the caption "One more thing," showing the interface of Raycast on a mobile phone. It can replace the native Spotlight launcher on iOS with Raycast. Alfred, the efficiency launcher big brother, is also included. However, comments from netizens express doubts about whether there is a launcher option for iOS/iPadOS versions and concerns about the instability and compatibility issues of the Android version. I suspect April Fools' Day has come early, so I'll wait silently. By the way, if Raycast does release for iOS, I really can't say for sure if I want to use it. There probably isn't any launcher on the phone that can surpass the integration level of Spotlight with the system https://x.com/raycastapp/status/1765709037612736527?s=20

📚Recommended Reading

OK

1. "Fuck You, Show Me The Prompt"

An image to describe post

The article states that most LLM frameworks are not only unnecessary, but actively hinder and slow you down. Their only value is their prompt.

Original: https://hamel.dev/blog/posts/prompt/

Translation: https://baoyu.io/translations/prompt-engineering/fuck-you-show-me-the-prompt

An image to describe post

This is the curated feed of frontend and backend technical weekly newsletter recommendations collected by Lao Hu Chuwu. It provides technological insights, development guides, programming practices, open-source projects, architecture explorations, and the latest newsletter updates.

Some New Products

1. Xi Huan Dian Dian - Daily Self-discipline Check-in, Time Management Tool

An image to describe post

This is a new creation from the Flomo author Shao Nan's team, called "Xi Huan Dian Dian". It adopts a cute and energetic style. When I saw this activity, my heart ached. My Xiaohongshu account with over a thousand followers was just banned. Including this one, I've had two accounts banned. I really don't know how to be a technology and internet digital blogger on Xiaohongshu. https://apps.apple.com/cn/app/习惯点点-日常自律打卡-时间管理神器/id6473467524

Thank you for reading. That concludes the content of this issue's newsletter. 🎉
Would you like to receive timely updates? You can subscribe to me via email or through the "Lazy Digital Life" WeChat official account.
Feel free to like, comment, and support me.
An image to describe post
OK

话说回到开始的交流群,我最近也在开始研究 ComfyUI ,所以建了一个群,既欢迎刚接触 ComfyUI 的小白,更欢迎已经有很多经验感悟的大佬,我也放了一些相关的资源在群里,大家一起学习AI一起讨论,目前见到好几个设计圈熟悉的面孔了哈哈。期待在群里见到你们,下周再会!

An image to describe post

Back to the communication group we started, I have recently started studying ComfyUI, so I created a group. It welcomes both newcomers who are just getting started with ComfyUI and experienced professionals who have a lot of knowledge. I have also shared some related resources in the group. Let's learn AI and discuss together. I have already seen several familiar faces from the design community, haha. Looking forward to seeing you all in the group, until next week!