OpenAI has released two new Sora videos on its YouTube channel. Each was produced by professional filmmakers and uses only videos generated by the Sora model.

Sora was first released earlier this year, but after months of waiting, it is still only available to a handful of filmmakers and creative professionals, and according to OpenAI, this is not likely to change anytime soon.

One of the new videos features old-style black-and-white documentaries featuring animals in unexpected roles. Another, in what can only be described as a neon dream world, features car washes, walking on clouds, and glowing coats.

These clips show that Sora is still superior to the current model, even after the massive upgrade to the Runway Gen-3 and the release of the Luma Dream Machine, but not by much; there are already several better alternatives to Sora, including the Luma Dream Machine, which is a more modern version of the Sora, and the Luma Dream Machine, which is a more modern version of the Sora.

Longer initial generations mean improved consistency, and the level of natural movement suggests that Sora is almost an open-world model. However, others are quickly catching up and may reach Sora levels before Sora is released to the public.

Tammy Lavin created the first video. She is a digital artist specializing in 3D and new technologies rather than filmmaking; all of the clips in the Sora video were generated by Sora without any additional VFX.

“What I like best about Sora is that I feel like I'm collaborating with Sora,” said Kurt. It feels like teamwork in the smoothest, most idealistic way.”

The video jumps from a neon car wash with crashing waves to a scene of a man walking through clouds and a woman lighting up the beach.

Lavin said that Sola triggered a new creative process, explaining that it “feels like magic” to show ideas on video that previously were only in his imagination.

“Ever since I was a kid, I would create montages or surreal visuals of certain things that I saw in real life. But I was never able to become a producer or director, so it never came true until now. So this is a dream come true.”

Like the previous video, this next production is by someone who is not traditionally a filmmaker. Benjamin Desai is a creative technologist and digital artist whose primary focus is augmented reality and immersive content.

He said in a statement that he is “excited to share an imaginative look into another past driven by Sora.” In the video, Desai “blends the aesthetics of early 20th century cinema with whimsical scenarios, placing animals in unexpected roles.” [At the beginning of the video, a bear is cycling and a gorilla is on a skateboard. Further along, a dancing panda, a man riding a dinosaur, and a woman riding a giant turtle appear. It is both impressive and disturbing at the same time.

“This work aims to spark a sense of wonder as well as demonstrate the possibilities of today's technology,” Desai explained.

“Creating with Sora is still an experimental process, requiring a lot of iteration and fine-tuning. It is a collaborative effort between human and AI rather than a magic button solution.” [Open AI stopped short of suggesting when Sora will be available to the public and instead talked about what they are doing to make Sora available to the public.

Earlier this year, CTO Mira Murati suggested that it might be released this summer, but that is not a possibility at this time. If a general release is made this year, it would be after the U.S. presidential election in November and probably in conjunction with a major update of ChatGPT.

The company is currently rolling out to a broader group of professionals, not just filmmakers. This includes VFX specialists, architects, choreographers, engineering artists, and other creatives.

This is “to help us understand the capabilities and limitations of the model and to help shape the next phase of research to create increasingly secure AI systems over time,” a spokesperson explained.

The video is impressive and continues to show the power of the Sora model, but other tools such as the Luma Labs Dream Machine, Runway Gen-3, and the Chinese Kling AI model all offer similar rendering quality.

Sora appears to motion down more accurately than any other model, but it is only a matter of time before it too is cracked.