Sora Has Exponentially Improved AI Video Generation
Open AI’s new Sora video model has just released, and boy is it impressive. Lets take a brief look at what it is, what it is capable of, and its future.
Sora Has Exponentially Improved Ai Video Generation
Sora is OpenAI’s (makers of GPT/DALL-E)newest creation and has been making huge waves in the AI community. What it is capable of is leaps and bounds ahead of previous text-to-video models. Let’s examine what Sora is, and what we can expect to see from it shortly.
What is Sora?
According to OpenAI “Sora is an AI model that can create realistic and imaginative scenes from text instructions.” and boy are they being modest. Sora has capabilities far beyond any previous text-to-video model by miles. Sora was just recently announced and is being showcased on OpenAI’s Website, and already AI enthusiasts are buzzing about it.
Some Fantastic Sora Examples
These examples are all taken from OpenAI’s website, and there are MANY more, I encourage you to visit and take a look for yourself at all of them.
Knowing It’s Weakness
The website also shows some of the limitations and weaknesses in the model at present. There are some things it simply struggles to do. They stated the following:
The current model has weaknesses. It may struggle with accurately simulating the physics of a complex scene, and may not understand specific instances of cause and effect. For example, a person might take a bite out of a cookie, but afterward, the cookie may not have a bite mark.
The model may also confuse spatial details of a prompt, for example, mixing up left and right, and may struggle with precise descriptions of events that take place over time, like following a specific camera trajectory.
Weakness: Animals or people can spontaneously appear, especially in scenes containing many entities.
Weakness: In this example, Sora fails to model the chair as a rigid object, leading to inaccurate physical interactions.
Sora’s Reception
Video generated with OpenAI’s Sora. Hard to wrap your mind around this.
byu/gantork insingularity
This was the initial video I and many others saw posted to Reddit that captivated me. Truly the reflection is impressive, and the overall composition is mature. This goes well beyond anything seen in the short amount of time that text-to-video has existed.
Introducing Sora, our text-to-video model.
— OpenAI (@OpenAI) February 15, 2024
Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions. https://t.co/7j2JN27M3W
Prompt: “Beautiful, snowy… pic.twitter.com/ruTEWn87vf
Twitter is also lively with the discussion on Sora.
And Just Because
That is it, I simply wanted to show what I found to be quite impressive and am excited to see more of in the future. AI has come a long way in a short time, and these examples are sure to get many believing in their multitude of multimedia possibilities. OpenAI hasn’t done it all themselves, remember:
That's just too good. Haha https://t.co/BEpo1GsDkf
— Wolf Merrik (@WolfMerrik) February 16, 2024
