Turn your photos into 3D

PLUS: Apple acquires AI startup

Good morning human brains, welcome back to your daily munch of AI news.

Here’s what’s up:

  • Turn any 2D picture into 3D 🗿

  • The FTC is looking for monopolies in the AI industry 🕵️

  • New research uses AI to detect skin cancer ☀️

  • Apple acquires an AI video-compression startup 🍏


Make-It-3D: turn any 2D image into a 3D model.

Imagine you have a picture of a cute bunny rabbit that you want to turn into a sculpture:

How would you do it?

It’s really hard to turn a 2D picture into a 3D model without knowing what the entire object looks like.

Well, that’s exactly the problem Make-It-3D solves, a new research paper that uses AI to hallucinate a 3D model from a 2D image.

You’re a human brain, so you’ve probably got a good idea of what the back half of a bunny looks like even if you haven’t seen the back of this particular bunny.

Make-It-3D can create high-fidelity 3D content from only a single image.

Make-It-3D’s “brain” works in a similar way. It uses a large dataset of information it already knows about other 2D images to improve the quality of the 3D model it creates.

3D-aware texture modification such as tattoo drawing is possible.

Possible Applications:

1/ Tattoo-previews. Take a picture of your arm and see exactly what your 2D tattoo art will look like once you’re inked.

2/ Immersive photos. Imagine if you could step into any photo from your camera real. Take a photo of a Parisian street and this tech would let you walk back down it whenever you want using VR

3/ Video game modeling. We’ve written about the possibilities of 3D model generation for video games before. Imagine how much easier it will get to create a video game when all you need to create photo-realistic video game objects is to take a picture of it in real life.


Use AI to extract gold from your videos 🏆

Maximize the ROI of your podcasts, interviews, or webinars with short clips optimized for social media.

Munch uses generative AI to extract the most engaging, trending, and impactful clips from your long-form videos.

They maximize the “trendability” of each clip with marketing and trend insights, ensuring the success and effectiveness of the content.

Munch can also provide you with pre-written social posts based on video content using GPT, and will instantly generate accurate, editable subtitles.



Think Piece

AI’s iPhone Moment — An Interview with Nvidia CEO Jensen Huang.

Startup News

Apple acquired AI startup WaveOne which uses AI to compress videos.

The FTC is monitoring the AI industry for emerging monopolies.

Zoom is partnering with OpenAI to add AI capabilities to its video platform.


Text2Video-Zero: a technique to turn text into a video that’s more consistent than previous techniques. The background and character stay the same from frame to frame instead of constantly changing, a common problem with AI-generated video.

Dermatologist-AI: using AI to identify and diagnose skin diseases like melanoma.

LightPainter: change the color and direction of light in your photos by drawing on them using an AI-powered image editing tool.


APO: an AI Chatbot designed to help older adults get comfortable with technology.

Summary.Legal: AI-powered transcripts & summaries to help you organize and search video calls, testimonies, depositions, and other important meetings.

Tourist Itinerary Bot: a Telegram bot to get tourist info for any city in the world based on ChatGPT.

Consensus: ChatGPT for researchers.


The two best tools for cheating on your homework of all time have combined forces. Bing (powered by GPT-4) can now make queries to Wolfram Alpha, unlocking new abilities in the maths and sciences.

Sorry professor.


“Explorers working in the Quantum Realm.” - @brockwebb

Until next time 🤖😋🧠

What'd you think of today's newsletter?

Login or Subscribe to participate in polls.