OpenAI launched ‘Sora’ a new AI video creation tool

OpenAI, an Artificial Intelligence (AI) company, launched Sora, its new AI model that generates videos based on user prompts. These videos could have lengths from a few seconds up to a minute.

How it works:

Sora is similar to ChatGPT, OpenAI’s text-to-text AI chatbot. Instead of getting a text-only reply, this model delivers a video presentation based on the user prompt.

Here’s a prompt example: “A stylish woman walks down a Tokyo street filled with warm glowing neon and animated city signage. She wears a black leather jacket, a long red dress, and black boots, and carries a black purse. She wears sunglasses and red lipstick. She walks confidently and casually. The street is damp and reflective, creating a mirror effect of the colorful lights. Many pedestrians walk about.”

OpenAI Sora Prompt Example (Credit: OpenAI)

OpenAI has shared the video output of this prompt and both the quality and the resolution are quite convincing. However, a close look at the video would allow you to identify the artificial element in the scene.

Capabilities:

There are two key capabilities that Sora has.

It can understand the time, place, and appearances (of humans, buildings, animals, or any other objects (non-living) and background) required to build a scene to relate to the prompt.

• Here’s a prompt example to understand this capability “footage of California during the gold rush”.

In this video not only the scene is relative but the camera quality is also blurry and low-quality to create an old footage effect.

OpenAI Sora Prompt Example (Credit: OpenAI)

Second is the capability to generative dense scenes with multiple objects and prompt specific characters or camera movement.

(Credit – OpenAI)

Early Bugs:

OpenAI says that Sora is still in the early stage of development and has some “weaknesses”. For example, a person taking a bite out of a cookie but the cookie may not show it.

Other bugs reported in the early version include collision detection, particle misplacement, unexpected object shapes, abnormal movements, and more. These issues mostly appear when a scene is dense and has many properties. These bugs may be addressed in future releases.

Safety and Security:

The AI firm says it will have security measures for Sora and it will not create anything that violates its usage policies. OpenAI launched Sora in a limited testing environment and a wide release is not yet announced at the moment.

(source – OpenAI)