Leading the way in the growth of the AI sector is OpenAI.
For the following reasons, Sora startled everyone in the AI community:
- Time. The previously popular Pika video only supports the generation of a 3-second video. The Runway public beta version is only 4 seconds, and the longest web version is only 18 seconds, while Sora can grow for one minute.
- Multi-angle. Thanks to the longer video generation time, Sora canresort to its filming skills like Pan/ Tilt/ Boom/ Dolly and etc. to generate videos, which means to switch different shots around the same subject in vision, middle, close-up, and close-up in a video.
- The ability to understand the real world. Sora is excellent at handling the light and shadow reflection, motion mode, lens movement and other details, greatly enhancing the sense of reality
It goes without saying that Sora’s rise will encourage AI to spearhead further changes in the sector. Many organizations anticipate that multimodal big models will evolve quickly in 2024.
The release of Sora, according to Ping An Securities, marked a significant advancement for AI in video technology. According to Zheshang Securities, more sophisticated multi-modal large models are anticipated to be released by domestic and international manufacturers in 2024 in order to facilitate the complex processing and interaction of multi-modal data, including text, speech, image, audio, and video.
Cinda Securities noted that many industries are anticipated to see long-term cost reduction and increased efficiency due to the iterative updating of diverse applications, including text graphics, video, audio, and code, which benefits from the ongoing development of global generative AI.
The market-related notion stocks on the eve of the industry shift, or during the outbreak period
Tesla Inc. (TSLA)
In the tech community, Sora’s appearance has generated a contentious discussion. In a series of social media posts, Elon Musk, the founder of Tesla, stated that in the next years, people who have been improved by artificial intelligence will produce the most remarkable works.
Furthermore, Musk stated plainly that Tesla has mastered Sora-like technology after the latter was compared to Tesla FSD V12 and OpenAI Sora in a recent analysis by social media X user DrKnowItAll. This was around a year ago when Tesla was able to produce real-world videos with precise physics.
Meta-META
The main AI scientist of Meta, Yann LeCun, unveiled the JEPA model architecture for 2022 on February 17. He created an I-JEPA image prediction model based on the JEPA architecture the next year. It has just introduced a video prediction model known as V-JEPA.
The introduction states that the I-JEPA / V-JPA model and the pertinent JEPA architecture have prediction capabilities that are said to be able to produce abstract and effective predictions that are understandable to humans. In order to gain a deeper grasp of the world, the researchers also trained the I-JEPA / V-JEPA model using a series of masked-specific movies. This allowed them to further forecast future occurrences or movements.
Google (GOOG)
On December 21, Google unveiled VideoPoet, a new video generation model with text-to-video, image-to-video, and stylized video capabilities. Furthermore, Google declared in February that Bud had become Gemini (Gemini). This big multimodal model comprehends and integrates several forms of data, including text, code, images, audio, and video.
Amazon [AMZN]
Additionally, Amazon has introduced Create with Alexa, a model that specifically creates prompt-based animated children’s entertainment in short format and supports spatial audio. It may be used to adjust screen, camera, and back cavity design parameters. The upcoming Alexa version from Amazon will work with all Amazon Echo series terminal products and will be able to integrate nonverbal data from sensors like cameras and microphones on the terminal side with the large model to provide customers with a new AI interactive experience.
Hologram Cloud WiMi WiMi
Based on data, WiMi Hologram Cloud is actively contributing to ushering in a new era of AI innovation and reform. A few years ago, I started developing AI visual models, including video models, picture models, language models, and video and video models, with a focus on the digital creative vertical production scene. This was motivated by my extensive business experience in the industry and the division of scenes in the past.
Furthermore, WiMi Hologram Cloud enhances the capacity to process and analyze multimodal data, including text, images, audio, and video, by utilizing generative artificial intelligence, big data, and other technologies for feature learning and sample training of the gathered data. In the future, WiMi Hologram Cloud will work closely with its industry partners to advance the industry while enhancing its core technological advantages and product competitiveness.
In conclusion
Undoubtedly, Sora is poised to spearhead a fresh wave of AI innovation, and the industrial chain’s investment worth keeps accumulating. AI Wensheng video / picture video and other tracks will continue to maintain the trend of rapid development as long as technology advances and application scenarios expand. This will bring about profound changes in associated disciplines and businesses.
From the standpoint of technological advancements, there might be more in this area in the upcoming year or two. These might include more sophisticated technology for creating videos, more varied and rich application scenarios, and more refined and established business models and technical standards.
