What is Sora by OpenAI and why is it important?
OpenAI, the company behind ChatGPT just released public information on their new AI video generator called Sora. What is Sora? Sora is an...
OpenAI, the company behind ChatGPT just released public information on their new AI video generator called Sora. What is Sora? Sora is an...
OpenAI, the company behind ChatGPT just released public information on their new AI video generator called Sora
Sora is an AI video generator capable of producing incredibly realistic videos. OpenAI describes it as "an AI model that can create realistic and imaginative scenes from text instructions". Sora is not yet available to be used by the public. We believe part of the reason for this is that it is too good at what it does and would therefore be at risk of misuse. However, there is significant interest in Sora because of the examples which OpenAI posted as part of the release.
Open AI has traditionally been transparent about the limitations of their AI models. While there are videos showing some limitations, it appears to be a significant advancement in AI generated video. What is most impressive is the video examples which are incredibly realistic. Many have reacted saying that this is a sign that AI generated movies, video ads, training course videos and more could be coming sooner than we think.
The first example released is of a woman in a red dress with a black leather jacket walking down the streets of Tokyo at night time. It is extremely realistic. The reflections of the night city landscape on the wet Road, as well as the reflections from the woman's sunglasses are very convincing to the human eye.
Another example shows a movie trailer featuring a 30 year old Spaceman which again is just insanely realistic. It's hard to fathom that these people are not real because the videos of them do not look fake.
There are many other amazing examples some of which are scenic art or cartoonish. One of the examples includes pirate ships sailing through a cup of coffee which although is not a realistic scene per se, the visual details are still incredibly impressive.
Sora is certainly not perfect and OpenAI is transparent about this. Probably the worst example of its imperfections is where archaeologists discover a plastic chair. The chair starts morphing and flying in the air which is very unexpected. Some other examples include a grandma blowing out candles but the candles don't actually get blown out and the people celebrating with her seem to have forgotten how to clap.
Another example shows a pack of wolf puppies overlap with each other and end up merging into one or appearing out of nowhere which is obviously not meant to happen. There is an example of a person running on a treadmill however the treadmill is moving backwards so the man is running in the wrong direction.
Sora shows strong capabilities to generate realistic videos and understand the details of each prompt.
There is a clear opportunity to improve the efficiency and quality of content creation with generative AI.
Sora will likely enable AI generation of training videos of any kind. Currently, we are limited to generating AI videos of people speaking to camera with tools like Heygen and Synthesia. Think of video tutorials showing particular actions for training purposes such as software tutorials and hands on labour tasks eg construction and hospitality. There are still challenges creating high quality presentation videos that show more than just a person speaking to camera but additionally demonstrating or going through presentation slides. Sora appears to be moving towards a future where this is possible to generate with AI. It also enables the possibility to customize and personalize training videos for each individual learner based on attributes like their learning style, existing knowledge and skills, and background.
It is unclear when Sora will be released to the public. Open AI is responsibly taking safety precautions considering the strong potential for misuse in areas such as politics. Open AI is testing the model and its downfalls in a small closed off group. Considering the apparent concerns, it is predicted that several phases of testing will occur before Sora is released to entire public.
It is likely that OpenAI is planning to offer access to Sora via ChatGPT, similar to how DALL-E images can be generated with ChatGPT Plus. Importantly to software developers, Sora will very likely become available through Open AI's APIs.
Considering the impressive videos that Sora is producing and the considerable work that has gone into this video generation model, it is very unlikely to be available for free. Users will need to upgrade to ChatGPT plus in order to access the model. We also have to consider that OpenAI may make an update to their pricing model. For example, this could potentially warrant a more expensive plan in addition to Plus in order to access Sora video generation. It is unclear how difficult it is to scale the AI video generated videos with Sora and the expenses to Open AI for doing so.
When comparing Sora by OpenAI with other AI video generators, several key points emerge from various sources:
Sora by OpenAI emerges as a cutting-edge AI video generator with exceptional realism and advancements that position it as a significant player in the realm of AI-generated video content creation. Its innovative capabilities and unique features set it apart from traditional models, paving the way for transformative applications across various industries.
For those intrigued by Sora and its capabilities in the world of artificial intelligence, we recommend visiting OpenAI's official website for more information, or follow them on social platforms for continuous updates and community engagements. Prefer an explanation in Spanish? Read one here.