Sora It is able to generate realistic videos based on text prompts. The model can create videos up to one minute long and at a resolution of up to 1080p. It performs well in handling reflections and shadows.
It is not yet open to ordinary users because OpenAI Red team testing is being conducted in collaboration with experts to assess the model’s possible biases, risks, and harms.
This article is divided into two parts. The first is Sora common sense, experience website, and effect display. The second is how to register Sora.
1. Common sense
1.What kind of company is OpenAI?
A: OpenAI is a top company in the AI industry, focusing on developing large language models.
There are five major schools of thought in the world of large language models:
"Nandi" (Google Gemini),
"The Beggar of the North" (Amazon Claude),
"Eastern Evil" (Grok),
"Meta Llama"
"Zhongshentong" OpenAI's ChatGPT!
OpenAI was originally a non-profit artificial intelligence research laboratory founded by Musk and Altman.
There are too many company stories, so I will write a chronicle of OpenAI's major events for you later.
OpenAI's main products include the GPT series, Vincent DALL-E3
In short, looking at the global AI community, OpenAI ranks first in algorithms, computing power, and product reputation.
2. What is Sora?
Answer: Sora is a "vintage video" AI model released by OpenAI on February 16, 2024.
Create realistic and imaginative scene videos based on text instructions.
In simple terms, imagine if you could tell a really smart computer, "Tell me a superhero story," and
The computer can create a whole video with superheroes flying around, saving people, and even special effects and backgrounds.
This is what OpenAI's SORA can do.
Sora is a very advanced tool. You just need to give it some text, such as describing the content of the video you want to watch, and it will
Can create a video based on what you say.
The above gif is a screenshot from Sora's homepage. Please note the small text below: "All videos on this page are directly generated by Sora.
Unmodified"
3.What is “Vincent Video”?
Answer: Creating video from text is called text-generated video.
It's like having a small movie studio, but everything is done automatically by computer, without the need for real actors or real cameras.
Sora has seen a lot of videos and pictures, and can generate new videos based on the user's prompt words.
Example: Night shot of a hermit crab using an incandescent light bulb as its shell
4.What are the Wensheng video software?
A: Before Sora, the mainstream ones were Pika, Runway, moonvalley, DomoAI, Leonard, etc.
Users do not need to understand specific technologies. If you use a refrigerator, do you need to study refrigeration technology?
The advantage of Sora over other video software is that it has better algorithms and more powerful computing power. Other video software can only generate videos of a few seconds or at most ten seconds. Sora can generate 60 seconds at a time.
OpenAI was able to train the Diffusion Transformer on a wider range of visual data than was previously possible, including different durations, resolutions, and aspect ratios. More faithfully following the user’s text instructions in the generated videos
Simply put, other software infer videos from pictures, while Sora automatically generates videos similar to 3D modeling after understanding!
5. Sora is positioned as a "world simulator"
Answer: To make Sora a Vincent video model is to underestimate OpenAI’s ambition.
Judging from the leaked videos, Sora has the ability to simulate real-world people, animals, and environments to a certain extent.
There is no need for any specific assumptions about three-dimensional space or objects; it is purely a natural phenomenon after scaling.
I even think Sora has a certain intelligence, and can infer the surrounding scene based on the scene.
Sora's advancement lies in its ability to infer and generate new things based on existing knowledge, which is the prototype of self-awareness.
For example, you have never seen a Mobike hit a tank, but you know from your experience that it is definitely like an egg hitting a rock.
A horrific scene will automatically appear in my mind.
Sora also has the same deduction ability as you.
Sora demonstrates more than just the ability to make videos. It also shows that when large models understand and simulate the real world, they can bring new results and breakthroughs.
6. Who can use Sora
A: Currently only the "red team" can use Sora
The red team is composed of the earliest customers and film and television professionals.
They are currently testing Sora to make sure it produces cool videos but also safe and without any bad content.
Currently, there are only two sources for various Sora videos circulating on the Internet, the official demo and the red team evaluation results.
7.Sora generates video effects
A: Sora can not only simulate real videos, but also generate special effects videos, and can also show different
Lenses
Example 1: Input prompt "POV shot of ants moving inside an ant nest"
You'll get a special effects shot of the animal world
Example 2: "Macro photo of a leaf showing tiny trains moving through its veins"
Sora will generate the following video for you:
8.Does using Sora require programming skills?
A: No, just use natural language prompts.
The so-called natural language is human language, commonly known as "speaking human language", see the prompts above.
9.Does Sora support Chinese?
A: Judging from the style of OpenAI, it should support directly inputting Chinese prompt words.
10. When is Sora expected to be officially released?
Answer: Expected to be before the end of March!
According to the path of OpenAI's release of DELL-E (drawing AI)
It should be divided into two usage paths, the first is an independently available version, and the second is a version combined with GPT4 or GPT5!
11. How to use Sora?
A: There are two ways, official website version and API version.
The official web version is generated directly on the OpenAI website and does not require users to install it on their local machines.
The API version uses a third party to call official server resources.
There may be an APP version later
12.Do I need to pay to use Sora?
A: Yes! Currently, OpenAI PLUS membership is $20/month
After Sora is launched, PLUS members should be able to use Sora with certain restrictions (such as duration)
The API is charged separately based on traffic!
13. Can I use Sora now?
Answer: It is currently available to a small number of users and is not open to the public!
ChatGPT Plus members should be the next batch of users!
14.How do Sora and ChatGPT combine?
A: There are two ways to combine
The first is interface integration, similar to the current DALL-E
The second is the combination of functions. For example, after you finish the Chatgpt conversation, you can directly summon Sara and say "Based on XXX, generate a
Similar videos"
15. If I am not satisfied with the generated video, how can I ask Sora to modify it?
A: Enter a new prompt directly, as follows:
Sora can not only generate AI videos from text, it can also change the style and environment of uploaded videos.
For example, after uploading a racing video, only the prompt words were modified, and 12 videos with different styles and environments were generated.
16. Who owns the copyright of Sora used in the film and television industry?
Answer: The producer. According to current cases, as long as it is not a naked copy, it is legally recognized, especially
Especially countries like Japan, which encourage the development of AI. I checked and found that the legal standards of each country are different. The laws of most countries are that as long as you don't completely plagiarize,
It is allowed to borrow styles and reorganize the original content, such as Japan.
17. What is Sora’s underlying technology?
A: Sora’s core technology is derived from the Diffusion Transformers (DiT) model.
This is a model proposed by two researchers from Berkeley and New York University in December 2022. Currently, one of them is at Meta AI and the other is at OpenAI.
All based on Google's open papers
Google's own paper, but in practice, OpenAI is the best. Google is full of talents but has big company problems.
Industry insider: OpenAI is crossing the river by feeling Google's way, and everyone is crossing the river by feeling OpenAI's way
18.How can I use Sora in the country?
A: Wait for the API development version, you should be able to use it directly
You can follow this official account and we will push you the latest resources as soon as they are available!
19.Does Sora have an app?
A: Not in the early stage, but it is expected to be integrated into GPT APP later.
20.Can Sora only generate videos?
A: You can also generate pictures
Sora is positioned as a real-world simulator, and making videos is just a side job
Just like the singer civilization, making a video is just a conventional weapon of "two-dimensional foil", while GPT5 is a killer weapon of dimensionality reduction
21. How to make Sora generate high-quality videos?
Answer: High-quality prompt words,
This requires first aesthetic sense, second imagination, and third photography and videography experience.
Case in point: A white and orange tabby cat was seen scurrying through a back alley in the heavy rain, seeking shelter..." (Chad Nelson)
22.How to make Sora have sound?
A: There is no official dubbing at the moment, but it will be available soon.
In addition, ElevenLabs is about to launch a semi-automatic AI dubbing test, which is expected to be in the form of prompt
Now the test list needs to apply https://form.typeform.com/to/gg0xzZW4
23. Will Sora allow film and television personnel to work in the industry?
A: No. Sora will be a tool for film and television people.
Sora can "seamlessly" mix two videos. Video 1 is a Sora-generated Minecraft video.
Then I mixed it with a video of riding a motorcycle to create the second video.
It can be foreseen that this function has huge creative potential in the future.
24.Why should I use Sora?
Answer: Throughout human history, every improvement in energy and production tools has brought about social changes.
More than 20 years ago, PCs became popular, the Internet emerged, and everyone was talking about informatization; more than 10 years ago, mobile phones emerged and mobile office was discussed everywhere.
However, people nowadays no longer talk about informatization and mobile office. This is because IT and mobile office have become basic resources, like water, which are everywhere. The same will be true for AI in the future.
The current application of AI is just the beginning. Not only abroad, but also domestic AI products are changing with each passing day.
The tools are similar. Once you master ChatGPT, it is easy to use other AIs. The sooner you master it, the sooner you can get on board.
25.What jobs will Sora replace?
Answer: I am conservative and my work scenes are relatively fixed. I work in video-related positions.
26. Who is suitable to learn Sora?
A: Many people think they are programmers, but they are not. Traditional programmers have a fixed mindset.
Film and Television major, Literature + IT are very suitable for young people!
27.How should education develop?
A: I suggest that domestic film and television majors open relevant elective courses as soon as possible, so that everyone can make good property planning and knowledge reserves in advance.
2. Sora Experience Website and Registration
Opportunities are reserved for those who are prepared. Although Sora has not been officially released yet, we can make good preparations and enter
Enter the Godfather's "sleeping mattress" battle state and welcome the day it is released
1.Sora Resources
Official website: https://openai.com/sora
Official website technical report:
Original: https://openai.com/research/video-generation-models-as-world-simulators
Translation: https://baoyu.io/translations/openai/video-generation-models-as-world-simulators
2. How to register Sora
Answer: According to OpenAI practice
Sora should be given priority to Plus users
In order to use Sora first, you need to register as a GPT member, then upgrade to a PLUS member and wait for Sora to be released.
As usual, after Sora is released, there will be an additional sub-column on the ChatGPT interface.
3. How to register Sora's API
The API interface has not been released yet, but all OpenAI APIs are under one account
Conclusion:
This article is the first official article in Sara's series. There will be more wonderful articles, cases, and tools later.
Opportunities are reserved for those who are prepared. You can pretend that nothing has happened, or you can choose to actively embrace changes. The spirit of innovation is rooted in the traditional culture of the Chinese nation. The "Great Learning" says "If you are constantly improving, you will be improving day by day, and you will be improving day by day"; the "Book of Changes" says
The spiritual origin of is also innovation, Yi means change, as the saying goes, "Heaven moves vigorously, and the gentleman strives to improve himself constantly."
Keep up with the pace of world development, grasp the changes, and find certainty for yourself in this complex world lacking in certainty.