Sora 2 Finally Available to Everyone
OpenAI has opened the long-awaited Sora 2 model to all ChatGPT Plus subscribers. The original Sora model was introduced with a limited preview in February 2024 but was only available to selected users for a long time. Sora 2 takes a big step forward in both quality and accessibility.
New Features
Advanced physics engine: Sora 2's biggest improvement is in physical realism. Elements like water flow, smoke movement, fabric physics, and gravity are now much more convincing. Errors frequently seen in the previous version such as floating objects or feet passing through the ground have been largely eliminated.
Long video support: It is possible to generate videos up to 60 seconds in a single pass. Additionally, with storyboard mode, you can create a consistent narrative by sequentially defining multiple scenes. You can specify separate prompts, camera angles, and atmosphere for each scene.
Image-to-video conversion: You can upload an existing image and animate it. You can make a portrait photograph talk, add motion to a landscape photo, or convert a product image into a 360-degree rotating video.
Audio integration: Sora 2 can automatically add sound effects to generated videos. Environmental sounds like water, wind, and footsteps are automatically added to the video.
Pricing Model
Sora 2 access is included in the ChatGPT Plus subscription ($20/month). However, usage is limited; you can generate 50 720p videos or 25 1080p videos per month. For more usage, per-image payment is available through API access. API pricing is approximately $0.10 per minute.
ChatGPT Integration
Sora 2's most unique advantage is its direct integration with ChatGPT. You can make video requests in natural conversational language in the chat interface by simply saying create a video without needing to write a specific prompt. ChatGPT understands your request and converts it into an optimal video prompt. This feature provides great convenience for users without prompt engineering knowledge.
Limitations
While Sora 2 is impressive, it has some limitations. Text generation is still problematic; signs and text in videos are mostly unreadable. Human hands have improved but still show errors in complex hand movements. Loading times can be long; during peak hours, you may need to wait 5-10 minutes for a video.
Comparison with Competitors
When compared with Runway Gen-4 and Kling 3.0, Sora 2 leads in some areas while falling behind in others. In prompt understanding, Sora 2 is the clear leader; OpenAI's accumulated expertise in natural language processing makes the difference. In cinematic quality, Runway Gen-4 still has a slight advantage. In price-performance ratio, Kling 3.0 offers a more accessible option.
Conclusion
The opening of Sora 2 to everyone is a move that truly brings AI video generation to the mainstream. Thanks to ChatGPT integration, producing professional-looking videos without technical knowledge is now possible.