Following ChatGPT, OpenAI has launched another product that shocked the tech circle—SORA! Let's take a closer look at it today!
What is SORA?
SORA is OpenAI's first**generative model,Through text instructions,It can directly output up to 60 seconds**,And contains highly detailed backgrounds、Complex multi-angle shots,And emotionally rich multiple characters,Debut is the king bomb!
SORA builds on past research on the DALL·E and GPT models. It uses Dall·e 3's restatement prompt technology to generate highly descriptive annotations for visual training data, so it is able to follow the user's text instructions more faithfully.
What's so disruptive about SORA?
The length of time that ** was generated. Prior to SORA, due to the limitations of physical and spatiotemporal reasoning from AI text to ** generation, the maximum length of a single coherence published in the entire industry was currently 16 seconds, and the general level in the industry was between 2-4 seconds. However, the maximum support length of SORA this time is 60 seconds, and the direct leapfrog upgrade is 15 times.
Camera cuts and smoothness. Previously, AI-generated ** products were all single-shot single-generation, and at the effect level, it was more like making a ** move. However, the ** generated by SORA can achieve seamless switching of multi-angle shots on the premise of maintaining the consistency of the subject, and the whole picture is clean and smooth, so as to achieve a real blockbuster effect.
The underlying model architecture used. Most of the previous AI generation ** is based on the "large model + big data" architecture, so it is difficult to break through the AI illusion problem, and Runway also officially proposed to build a general world model in December last year. However, this time SORA already has the prototype of the world model, and its generation is based on the understanding of the real physical world, so the content is more logical and convincing.
Let's take a look at some of the official releases and feel the charm of SORA!
Hint: Tokyo is bustling after the snow. The camera travels through the busy streets, following a few people enjoying the beautiful snow and shopping at the nearby stalls. Beautiful cherry blossom petals flutter in the wind along with snowflakes.
Tip: A sea bike race** that lets various animals ride bikes as athletes, using a drone shooting perspective.
Tip: A grandmother's influencer hosts a homemade dumplings cooking class set against the backdrop of a rustic Tuscan country kitchen with cinematic lighting.
Is it possible to try it out now?
At present, SORA is not open to the public, and only a very small number of test accounts are provided to professionals, and the threshold is very high. But there are also relatively simple ways to test, it is recommended that interested partners refer to the official example, write a good **prompt word, open a certain push, go to @openai founder Sam Altman, and wait for a reply