‘Sora’ is the text to video AI model and it is surely turning eyes and ears! AI is ruling the digital world and it is on everyone’s digital lips today. Everyone is keenly following to see what new AI innovation has occurred each day.
Keeping that in mind, OpenAI – the company responsible for creating ‘ChatGPT’ has created the next innovation called ‘Sora’ which creates videos from text using AI. Here are some highlights of Sora:
- It can create videos from text. Maybe when ‘Sora’ is open to all, I can tell it to create a video teaching the difficult concepts in Math and Physics for my daughter! 🙂
- Jokes aside, ‘Sora’ can create videos from text even modeling how things might exist in a real world without the user explicitly specifying it in their prompt(like the physical landscape) It is still in the research phase and is not open for public use.
- The videos can be a minute long and they are supposed to stick to the prompt
- As an example, the following is a video generated from the Sora’s OpenAI site using this prompt:
“Prompt: Drone view of waves crashing against the rugged cliffs along Big Sur’s garay point beach. The crashing blue waters create white-tipped waves, while the golden light of the setting sun illuminates the rocky shore. A small island with a lighthouse sits in the distance, and green shrubbery covers the cliff’s edge. The steep drop from the road down to the beach is a dramatic feat, with the cliff’s edges jutting out over the sea. This is a view that captures the raw beauty of the coast and the rugged landscape of the Pacific Coast Highway”
(Prompt and Video credit: https://cdn.openai.com/sora/videos/big-sur.mp4)
Now this is an AI generated video and isn’t this amazing? 🙂 I am blown over by the power of AI…aren’t you? 🙂
Before we start gushing more about AI, let us continue to see some other the capabilities of ‘Sora’.
4. Sora is a diffusion model and does not just stitch up pictures to form a video
5. It does have its limitations and the current model cannot understand ’cause and effect’ and may not be able to model it accurately. Sora might also get confused with “left” and “right” and spatial details resulting in incorrect videos.
6.However, it can also merge two videos to create yet another video(Cool, isn’t it? :))
7. Sam Altman(CEO of OpenAI) has posted a number of ‘Sora’ videos on Twitter or X. Do check out his videos here:
8. OpenAI is taking steps to ensure safety and working to limit misinformation, hateful content and bias when anyone works with this model. In the future, there will be tools that will detect when a video is generated by ‘Sora’. OpenAI is working with educators, policymakers and artists and others to understand different safety concerns regarding this new AI model.
9. When it is integrated into an OpenAI product, it will be whetted for prompts against “extreme violence, sexual content, hateful imagery, celebrity likeness, or the IP of others”
10. Every frame of the generated video will be reviewed.
9. ‘Sora’ is still a WIP(Work in Progress). AI is still gently moving our tech lives around(out of fascination or necessity) but nobody knows the real challenges that will be faced in the future but authenticity will be lost for sure.
Who has access to ‘Sora’ now?
As of today, ‘Sora’ has been given access to
a. Red teamers to understand risks
b. Visual artists, Designers, Film makers
After the ‘ChatGPT’ phase, ‘Sora’ and AI are again going to rock our tech lives. Whatever, be the future of ‘Sora’, I know there will be a flood of “good” prompts that will be published by many, many authors once it becomes public! 🙂