4th issue: 🤯 AI can now make prompts for images

🌯 Wrap for the week

A close look at all the main headlines in AI

The table has been flipped, Midjourney v5 can now generate prompts from an image.

Midjourney v5 is pretty new and now we have yet another new feature from the awesome team.

Midjourney has launched a new "/describe" feature that allows users to convert images into words, improving accessibility and search functionality for internet users.

Alt text elements used to describe images can be time-consuming to create manually, but the “/describe” feature will overcome this hurdle. The detailed captions provided by the feature can also offer more clarity to viewers.

With the /describe tool, users can upload an image and generate four different text descriptions of the image, which can be used to improve search results. Additionally, users can generate new variations based on a selected description, and remix the uploaded image using the new text prompt.

The feature also creates an interesting feedback loop with Midjourney's text-to-image system, potentially making it easier to develop alternate and more fruitful descriptions for the text-to-image generator.

The tool shows promise for creators, even in its early stage.

Now here comes example time 👇️ 

I must say I am pretty happy with the results. As you can see, the generated prompts provided quite good descriptions of the images presented. It can’t capture all details, but the spirit of the image is definitely there.

I can certainly see myself using more of this feature for sure. Are you excited about this?

News credit:
Midjourney Flips the Formula with New Image-to-Text Generator from PetaPixel

Have you met SAM from Meta?

Another one joins the AI race in the headline!

Meta has launched an image segmentation model called the Segment Anything Model (SAM), which can identify individual objects in images and videos, including those not encountered during training.

SAM can respond to text prompts or user clicks to isolate specific objects within an image, making the image easier to analyze or process. Meta believes the technology will be useful for understanding webpage content, augmented reality applications, image editing, and aiding scientific study by automatically localizing animals or objects to track on the video.

SAM democratizes the segmentation model creation process, reducing the need for specialized training and expertise, and making it easier for more people to conduct research into computer vision.

To see more of its capability, be sure to check this site from Meta:

News credit:
New AI model can “cut out” any object within an image—and Meta is sharing the code from Ars Technica

Other top AI headlines for the week

Hugging Face hosts ‘Woodstock of AI,’ emerges as leading voice for open-source AI development from Venture Beats

I personally have not used Hugging Face a lot but I have heard their name for quite some time while immersing myself in the AI world. It is good to see an organisation that is leading the open-source of AI development so that more people can try to get some hands-on experience with AI.

Also, I have heard a lot of murmurs about HuggingGPT, which can use multiple large language models (LLMs) to do the job. Truly intriguing stuff.

AI image generators' latest challenge? Depicting unicycles. from Business Insider

First we have hands, now we have unicycles.

The text-to-image prompts area never ceases to amaze me with the things that it fails at. I wonder what could be the next and greatest challenge 🤔 

Google Will 'Absolutely' Bring AI Chat to Your Searches from CNET

We won’t have a complete week without hearing something from Google.

I personally having the AI chat experience integrated into search is good for the users as that is going to shorten the time for the users to find out what they are looking for. However, this could impact Google’s Ad business, so I am keen to see what the final outcome could look like.

🧰 Toolbox

Check out this week’s tools and resources that give you more superpowers with prompt crafting!

1. 🔧 SiteGPT
If you got a site and want to provide your visitors with the latest AI chat experience, this is the tool to check out.

2. 🔧 AutoGPT
If you have heard of the experiment to run a business by ChatGPT and want to try it yourself, you should check this tool. AutoGPT allows you to see the AI reasons, plans and executes the tasks. Here is a demo for it 👇️ 

3. 📜 BloombergGPT
BloombergGPT learns and trains from the financial data in the world. So, does it possibly the most valuable model? 🤔 I guess so, but I think at least it should be pretty good with the finance questions now. If you don’t want to go through the 65-page paper, here is a good summarising video.

👋 Your next Twitter friend

Connect with great people in the industry to supercharge your growth

🖼️ Image of the week

I have seen a lot of people creating people competing in eating festivals with some weird stuff. Well, here is my take on this trend.

Didn’t quite get the pho bit but the eating contest feel is certainly there. Also, looking at how people are using their hands, maybe the next challenge for AI image generation is matching the right utensil for the right food. 😂 

💕 Share the love and get some cool avatars!

I hope you have been enjoying the Prompt Craftsman newsletter so far. If you do, why not share your love of the newsletter with your friends as well? 😉 

Obviously, if you are sharing the love, I should do something in return to share my gratitude. That is why, I have decided that I will randomly select 3 subscribers each week who have got at least 3 of his/her friends to join the newsletter that week and provide them with 3 cool avatars.

For the lucky chosen subscribers, I will create these avatars using the portrait photo(s) you give me in 3 different styles of your nominations. You can suggest styles either using some image(s) you like and/or some descriptive text. If you have heard Lensa before, it is basically the same concept, but free of charge with the Prompt Craftsman! 😄 

I won’t promise to get the same level of detail, but the end result could be similar to the ones below.

Alright, if you are in, here is your link to share your love of the Prompt Craftsman with your friends. You are the best! ❤️