Draft:Video Rebirth
 | Review waiting, please be patient.
This may take 2–3 weeks or more, since drafts are reviewed in no specific order. There are 922 pending submissions waiting for review.
Where to get help
How to improve a draft
You can also browse Wikipedia:Featured articles and Wikipedia:Good articles to find examples of Wikipedia's best writing on topics similar to your proposed article. Improving your odds of a speedy review To improve your odds of a faster review, tag your draft with relevant WikiProject tags using the button below. This will let reviewers know a new draft has been submitted in their area of interest. For instance, if you wrote about a female astronomer, you would want to add the Biography, Astronomy, and Women scientists tags. Editor resources
Reviewer tools
|
Video Rebirth (Chinese: 重生视界; pinyin: Chóngshēng Shìjiè) is a Singapore-based artificial intelligence company that develops AI video generation models and world models. The company was founded in October 2024 by Wei Liu, a former Tencent distinguished scientist who served as chief scientist of Tencent's Hunyuan foundation model series.[1][2] As of March 2026, Video Rebirth had raised a total of US$80 million from investors including AMD Ventures, Hyundai Motor Group's ZER01NE, and Qiming Venture Partners.[2]
History
[edit]Video Rebirth was founded in Singapore in October 2024 by Wei Liu, who had left his position as a distinguished scientist at Tencent in September 2024.[1][2] Liu had spent more than eight years at Tencent, where he served as chief scientist of the Hunyuan foundation model series, which has since been integrated into products such as WeChat.[3][4] He holds a Ph.D. in electrical engineering and computer science from Columbia University and previously held research positions at IBM and DiDi, alongside teaching appointments at Rensselaer Polytechnic Institute and Stevens Institute of Technology.[2]
Liu co-founded the company with Dan Kong, a former director of the 42X Fund at Abu Dhabi-based AI company G42; Difu Li, a former senior director who led AI strategy at Tencent; and Peng Liu, a former quantitative developer at JPMorgan Chase.[2][5]
In November 2025, Video Rebirth announced a US$50 million seed round with investors including Qiming Venture Partners.[1] By March 2026, the seed round had been extended to US$80 million, with additional investors including AMD Ventures and ZER01NE, the venture capital arm of Hyundai Motor Group.[2] The company stated it was raising an additional funding round in July 2026 but declined to provide further details.[2]
Video Rebirth is headquartered in Singapore, where it has maintained its base of operations since inception, and also has an office in Hong Kong. The company has not conducted research and development in mainland China.[3][2]
Products
[edit]BACH engine
[edit]Video Rebirth launched its AI video generation engine, BACH, in May 2026.[3][4] The engine targets enterprise clients in advertising, entertainment, filmmaking, and gaming.[2] Upon its release, BACH debuted at sixth place on the Artificial Analysis text-to-video leaderboard, ranking behind models from Alibaba, ByteDance, Kuaishou Technology, and xAI.[2][3] It was the highest-ranking startup model on the leaderboard and had the lowest cost per minute of video generated among the top ten models.[2]
BACH can generate multi-shot videos of up to 45 seconds based on reference images and text prompts, produce individual clips of up to 8 seconds from text or images, and bind a static character to a reference video.[2] The engine supports native 1080p resolution at 24 frames per second.[4]
BACH is built on two proprietary technologies: Dual Diffusion Transformer (DDiT), which separates prompt adherence from visual generation into distinct processes, and Physics Native Attention (PNA), designed to maintain physical consistency across video shots.[2] The company also employs a technique called multi-step sampling loss (MSSL), which trains the model to anticipate and correct errors during the generation process, reducing the number of computational steps required to produce final video output.[2]
Video Rebirth reported that BACH's inference cost per clip was lower than that of other frontier models, though it declined to disclose exact figures.[2]
World model
[edit]Video Rebirth is developing a world model—a real-time interactive AI system designed to simulate the physical world—under the prototype name "Olympus."[3][2] According to Liu, the Olympus model would function similarly to Google's Genie 3 but would also generate environmental sounds, such as collision or footstep audio.[2]
The world model is being trained using the same proprietary architecture as BACH.[4] Its training data consists primarily of real-world video, supplemented by interactive and synthetic data.[4]
University of Adelaide researchers Zicheng Duan and Lingqiao Liu, who published a paper on generative video world models in March 2026, noted that improved video generation quality can correlate with world modeling ability but does not by itself prove that a model has acquired a robust understanding of the physical world.[4] Alec Wrubel, an associate partner at McKinsey, described world models as "somewhere between" hype and a game-changer, stating that they are "largely in early development" and "not yet at the level of fidelity or cost profile needed for broad deployment across industries."[2]
Partnerships
[edit]On May 15, 2026, Video Rebirth announced a partnership with the Digital Creator Africa Academy for Microdrama (DCAA) at the Marché du Film during the Cannes Film Festival.[6][7] The collaboration was described as Africa's first industrial AI filmmaking programme.[6][7] Under the agreement, Video Rebirth provided BACH platform access credits to 100 students enrolled in DCAA's AI Filmmaking stream, targeting experienced filmmakers who had worked on productions distributed through platforms such as Netflix and Prime Video.[6][7][8]
Business model
[edit]Video Rebirth targets enterprise and professional users rather than consumers.[3][4] Co-founder and chief operating officer Dan Kong described the company's positioning as becoming "the Anthropic of video generation," selling model capabilities to enterprise customers rather than offering end-user products.[4] Nikkei Asia reported that the company serves enterprise clients including microdrama studios in China.[9]
See also
[edit]References
[edit]- 1 2 3 Liu, John; Huang, Zheping (November 5, 2025). "Tencent AI Veteran Raises Funds for Rival to OpenAI Sora". Bloomberg News.
- 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Lee, Zinnia (June 24, 2026). "Singapore Video Startup Founded By Tencent's Former AI Head Bets Big On World Models". Forbes.
- 1 2 3 4 5 6 Chow, Vincent (May 22, 2026). "China is losing the LLM race but it can still win in AI, ex-Tencent AI lead says". South China Morning Post.
- 1 2 3 4 5 6 7 8 Kaonang, Glenn (May 26, 2026). "The world model ambition growing inside a SG AI video startup". Tech in Asia.
- ↑ Kaonang, Glenn (May 14, 2026). "Tracking the alumni networks powering SEA's AI startups". Tech in Asia.
- 1 2 3 Akutu, Geraldine (May 19, 2026). "Africa Steps Into The Future of Cinema With Landmark AI Filmmaking Partnership". Guardian Nigeria.
- 1 2 3 Ojugbana, Victoria (May 15, 2026). "DCAA, Video Rebirth Launch Africa's First Industrial AI Filmmaking Programme". Independent Newspaper Nigeria.
- ↑ "How Africa Is Positioning Itself As The Next Global Hub For AI Filmmaking". ThisDay Live. May 15, 2026.
- ↑ Chen, Lorretta; Zhou, Cissy; Suzuki, Wataru (July 25, 2026). "In China's 'microdrama' boom, AI moviemakers chase 2 minutes of fame". Nikkei Asia.