The world of video generation has been quite turbulent since the beginning of 2024, particularly after the launch of Sora, a product designed to push the limits of creativity using AIMany creators anticipated this technological breakthrough with excitement and eager debateOn February 16, 2024, the initial unveiling of a demo left audiences breathless as they contemplated the implications of such technology on their workflows and creativityA year later, Sora finally opened its doors to users, but the journey has been anything but smooth.

Chen Wei, a rising filmmaker and screenwriter who graduated from the Beijing Film Academy, represents the voice of a generation of creators navigating this evolving landscapeThe excitement he felt a year ago was palpable: "I registered on the very day of its launch, despite the steep price tag of $200 for the pro mode." Sora is noteworthy because it is among the first of its kind to offer direct outputs from its capabilities, a feature known as DiT or "Dramatic Interaction Technology." However, the initial thrill soon transitioned into skepticism as users began to test out the product extensively.

On the night Sora opened for registrations, traffic surged to the landing page, resulting in the site temporarily shutting down registrations due to overwhelming demand

Alongside Chen Wei, popular figures in AI content creation such as Han Qing, the head of AI Talk, and the AI vlogger known as @Qiuzhi 2046, quickly secured their spotsWith subscriptions nearing 1500 RMB, they were eager to discover the possible outcomes of investing in this revolutionary software.

Nonetheless, the anticipation waned as creators grappled with the underwhelming reality of Sora's performanceA wave of new competitors, both domestic and international, had launched their offerings throughout the past year, and their capabilities set a high bar for expectationsChinese products like Keli and Jiemeng, along with platforms such as Runway and Luma, had already established themselves as formidable contendersDespite Sora's initial promise, it faced significant criticism for not living up to many creators' lofty expectations — especially regarding stability and realism in generated videos

• Learning from the Historical Lessons of Protectionism
• Are Robots on the Horizon?
• Stable Trends in the Global Commodity Market
• Flying Car Development Soars with Automaker Backin
• Micron Shares Plunge 16% on Weak Forecast

Problems like disappearing characters and disordered scenes still plagued Sora's outputs, and much like other AI video generation products, it required extensive retrying, informally termed drawing cards.

Professor Ni Bingbing from Shanghai Jiao Tong University took a critical look at Sora, sharing insights from his expert analysisHe and his team compared Sora to several leading domestic video generation tools and concluded that Sora did not outperform its rivals, lacking in certain aspects such as consistencyThis made it clear that while the industry strove for innovation, Sora did not manage to carve out a significant competitive advantage.

As the year progressed, the rush in video generation technology felt like an arms race, with both local and international players refining their products continuouslyGoogle even released an upgraded version known as Veo2, which some users believed surpassed Sora in the fidelity of simulated physical elements

Speculation and debate about who would dominate the market remained inconclusive.

The anticipation surrounding Sora's launch was significant, with early progress being described as a "return of the king", but such sentiments quickly shifted towards dissatisfaction once the product actually became available for widespread useUsers began to voice their disappointment over high generation failure rates and the software's inconsistencies compared to other options they had been using over the past months.

Chen Wei noted, "When we put Sora through its paces, we found its failure rate alarmingly high, living up to expectations far less than initially believed." Elements like the storyboard features allowed for some nuanced control over generated videos, yet the instability with image generation and model performance created a disconnect between the pitch and the productMany in the industry found that generating usable segments often required numerous attempts—an issue that seriously dampened the buzz initially surrounding Sora's promise.

The sentiment echoed by fellow creators was one of incredulity as they recalled their earlier feelings of thrill and wonder

As they engaged with Sora, they found the supposed advances in understanding physicality and semantic coherence lacking, failing to differentiate itself against competing models that had already set a high bar in these areas.

The discontent was palpable; some users took to social media platforms to lament their investment with comments like, "I regret spending $200 on this." Many lamented the high subscription price, given that it did not appear to offer markedly superior outputs compared to existing alternatives.

However, some advanced users regarded the price as acceptable given that high-end video models elsewhere also demanded a premiumFor instance, Runway's subscription costs around $95, while domestic competitors like Keli also cost around 666 RMBIf you had a specific need for such technology in professional settings, the view changed: many creators stated it was a cost they were willing to absorb for the potential returns it could yield in production.

Sora's introduction did not solely focus on pricing structures

Instead, it offered features — such as diegetic editing capabilities and a vivid storyboard function — aimed at enticing potential usersWhile the storyboard allowed for a more precise and delightful experience controlling video sequences, creators like Han Qing found the tools promising yet imperfect, raising the question of whether these capabilities could justify the investment in a faltering model competency.

Despite these challenges, many professionals believed there was still merit in Sora's designHan Qing, for instance, discovered an appealing aspect of Sora was its generation speed, delivering videos quickly even in its slower modeThis contrasted with other platforms, which could take significantly longer to produce outputsSpeed, he noted, was crucial — particularly for those in the fast-paced business of content production where inspiration could be fleeting.

Looking ahead, the landscape of video generation software remains competitive and uncertain

Artists and content creators now find themselves at a crossroads, wondering which tools to trust and rely upon for their creative visions amidst a profusion of offersWhile creators strive for high-fidelity outputs, they remain cautious of returning to the drawing board after a less-than-satisfying experience with a product that promised so much.

As new developments emerge, the prevailing sentiment is one of cautious optimism, with discussions regarding AI's future in video generation being ongoingWith many solutions grounded in the DiT architecture — which has yet to fully grasp the intricacies of physical interactions — it remains to be seen which model will ultimately rise to prominenceWill the innovations of 2024 lead to lasting change in the video generation realm, or will they be mere stops along the road to further exploration and refinement? Only time will tell in this ever-evolving narrative.