It was mid-February this year when OpenAI previewed Sora, their text-to-video AI engine. The examples were jaw-dropping, but there were unresolved safety issues. Only a week after OpenAI’s announcement, Google decided to pull back their text-to-image AI engine after it created historically inaccurate images, spurring a public outcry. AI companies got gunshy.
OpenAI spent ten months fine-tuning Sora with a group of test users.
As part of its 12 Days of OpenAI, the $150 billion AI behemoth finally made the video tool available to the public this week—and it’s pretty good. Below is the first video I created. It took only a couple of minutes to generate this clip using the prompt: “Create a video of a space alien talking into a microphone in a radio studio.”
Is it perfect? Not yet.
In my sample video, there are too many microphones, but it’s an acceptable first pass.
In another example posted on X, you can see one of the most challenging AI videos to create—an athlete in action. A still shot taken from a Sora-created gymnastics video shows the technology isn’t close to being ready to re-create a sporting event — yet.
Is it better than what text-to-video platforms were creating only six months ago? Absolutely.
This technology’s advance is remarkable. It’s hard to believe that none of this existed pre-2020.
The advancement of this technology is remarkable. It’s hard to believe that none of this existed before 2020.
It’s becoming clear that these tools will soon be widely used to assist creators and advertisers in enhancing their content across all digital platforms. Need a shot of a streaming cup of cocoa for a client or a video of a morning show DJ skiing down the Matterhorn?
It will soon be as easy as typing a few sentences into an AI engine.
12 Days of OpenAI Releases So Far
Day 1 – ChatGPT o1 and ChatGPT Pro – a new model for advanced reasoning and a premium tier for power users
Day 2 – Reinforcement Fine-Tuning Preview – the ability to fine tune data for power-users
Day 3 – Sora AI – The text-to-video AI engine goes public for OpenAI paid users
Day 4 – Canvas Mode launches – A new multi-screen layout to assist with writing text and coding
Day 5 – ChatGPT integration with Siri in iOS 18 on the latest iPhones
Day 6 – Advanced Mode – ChatGPT can interface with users in real-time with both audio and video
Day 7 – Projects in ChatGPT – the ability to create, organize, and manage long-term tasks or multi-step workflows, plus saving and collaboration functionality
[details]
Next week we’ll review the final five days’ of OpenAI releases, new Google AI releases, and a Santa AI tool!
Along with his duties at Jacobs Media, Chris develops revenue opportunities for jācapps’ mobile apps clients.
- 2024 Wrap-Up - December 20, 2024
- OpenAI’s Text-to-Video Generator is Now Public - December 13, 2024
- OpenAI Gives Twelve Days of Treats This December - December 6, 2024
Leave a Reply