Exploring Sora: OpenAI’s game-changing text-to-video tool
On December 5, OpenAI announced its “12 days of shipmas”, which will see the launch of new features, products, and demos, among which, after a lengthy beta trial that hasn’t been without issues, is its text-to-video tool Sora, after a long and complicated beta testing period.
So far, the best detailed analysis of Sora’s performance I’ve seen is Marquis Brownlee’s, which shows what it does well, badly, and not at all: very good for animation but not so hot for real images. It also has any number of features aimed at preventing it from being used to create deepfakes (for the moment, only a subset of OpenAI users will be able to generate videos using an uploaded photo or footage of a real person as a reference).
It’s not available in all countries, but it can be accessed — if you’re lucky and it’s not overloaded — via VPN. Sora’s interface allows you to upload videos and images, adapt prompts to make modifications to a series of videos created by other users or that appear as a demonstration, saving them for later use, with the possibility of making folders to…
