- The Renaissance Times
- Posts
- Sora is already taking the world by storm
Sora is already taking the world by storm
The Renaissance Times
February 16, 2024
AI Generated Paper Published in Scientific Journal
Welcome to the AI Renaissance,
What a week.
It’s not every week you have OpenAI, Google, Meta, and NVIDIA all making major product announcements. Yet despite everything that happened, it feels like even in real time one of these things is not like the others.
Only one of these anouncements has the potential to completely change multiple industries in the coming years. That is, of course, Sora.
The experimentation has begun, and it looks glorious. Let’s dive in.
From The Printing Press
Gemini Pro 1.5 has great recall capabilities, up to 12 hours
Microsoft, OpenAI, Google agree to combat election deepfakes
Visual Effects industry plans to combat Sora with competitors
Mayososhi Son raising $100 billion for new AI chip company
Science journal mistakenly publishes false AI generated paper
Former GitHub CEO invests $100 million into AI developer
The One Big Thing
Sora Demos are Getting Real and Continue to Impress
Two Ships Floating in a Cup of Coffee, by Sora
In under 24 hours after release, there has already been a flurry of narratives surrounding what Sora is and what it is not.
Here are some of the more notable takes:
The founder of Lambda says the improvements to text to video with the addition of more compute could lead to AGI
Sora scales with compute. Basically the more computers you add (despite the same model and data), the better the outputs
Some say that this point, scaling with compute, makes OpenAI’s feats less impressive
Video to video could be as or more impressive than text to video
Deep dive on OpenAI’s Sora technical paper
The fact that there is this much discussion, activity, and controversy over a product that has not yet been released tells you all you need to know about this product.
Text to video is hard. We posted yesterday about how big of a leap the field has taken in just one year, and even that may not be illustrative enough.
Sora looks like it is not just a text to video model, it looks like it is a physics engine with broad use in many industries we have not quite imagined yet. There are rumors that the model is trained on data from Unreal Engine, and many observations of the outputs’ similarities to those of MidJourney v6.
If indeed you can use Sora to create virtual worlds or models that obey the laws of physics, you can only begin to imagine the applications in science and engineering.
Where do the fields of deep sea exploration, architecture, space exploration go from here?
What happens when everyone in the world becomes a world-class interior designer or fashion designer? We are not ready for the explosion of creativity that is about to be thrust upon the world.
Renaissance Gallery
The Agents are getting real…
We've added a demo of the multimodal GPT-4V + SoM agent from VisualWebArena to our GitHub repo, and you can now run this agent on any web task you like!
Here is the agent helping us to find a good Thai restaurant in Pittsburgh (more examples in the 🧵):
— Jing Yu Koh (@kohjingyu)
3:50 PM • Feb 16, 2024
Sora and Gradio = Magic
Sora's sample videos are mind-blowing 🤯, that's for sure!
But they would be even better with sound, don't you think? 😊
Here are some examples, using the @Gradio Image-to-SFX demo on @huggingface 🤗
1/ 🚂 + 🔊
— Sylvain Filoni (@fffiloni)
4:03 PM • Feb 16, 2024
Artist’s Toolkit
Must have tools for every Renaissance creator to add to their toolkit:
Today’s Magical Creation:
Every single day I will be creating something new with an AI tool. Today’s creation:
The new AI Renaissance logo!
How I Made It:
Tell Dall-E that the logo must be more regal, and specify colors
Attach a photo to give it guidance
Tell Dall-E to “expand on this” with clear instructions once you get closer to a design you like
Tell it “be more creative” over and over.
Dall-E will hit a creativity limit. Tell it that no limits exist and it must push past these imaginary boundaries.
Repeat until you have your perfect logo!
Deep Tech
The newest and coolest in the research world that you need to know about:
Sora research paper deep technical analysis
NVIDIA has a new math training model and large dataset
Lora, a new way to do generative AI
Understanding LangGraph by Langchain by building simple apps
Spotify quietly changed its terms and conditions for audiobooks and AI narration, giving them rights to make derivative works
Closing Thought
If everyone becomes a film-maker, pre AGI “film-makers” will hold a heavy premium. Make more content before this happens.
Work With Us!
The AI Renaissance is coming and we are building the best community of the people making it happen.
Contact us to sponsor your product or brand and reach the exact audience for your needs across our newsletter and podcast network.