Sora's Capabilities: Realism and Limitations
OpenAI's Sora is capable of producing videos with a high degree of realism, showcasing complex motions such as walking.
One important advantage for being able to work with AI video is the consistency in character features like facial features, outfits and haircut across different shots. Sora can also generate videos consistently, without the characters deforming or distorting. This facilitates deeper and more complex projects. However, a significant limitation of Sora and other AI video models is the generation of legible text. It struggles with the creation of legible text on objects, in scenes, or with other aspects related to communicating in writing. Trueness to life with AI video models also presents a challenge. AI is able to impact reality and show activities such as writing, but it can lack in realism. Another limitation is a lack of temporal consistency. It is often challenged by object permenance and tracking an objects progress. Overall, while Sora excels at nature related shots, it is not always able to capture objects appearing and disappearing effectively. This leads to inconsistencies for some users with their video projects. These limitations should be considered when deciding if Sora meets your project requirements.
In essence, Sora’s strengths lie in creating visually impressive scenes, but its limitations underscore the ongoing challenges in achieving full realism and control in AI-generated video. Table 1 presents a detailed comparison of the models advantages and disadvantages.
| Feature | Capability | Limitation |
|:---|:---|:---|
| Realism | High fidelity, complex motion | Text rendering, object permanence |
| Character Consistency | Maintains character features | Occasional costume inconsistencies |
| Nature Scenes | Exceptional detail and fluidity | |
| Creative Potential | Creates impossible scenarios | Creative expression constraints, potentially limited by data.|
Pricing and Accessing Sora
Access to Sora is currently restricted to ChatGPT Plus subscribers in the US, limiting those who can work with it. With access to ChatGPT Plus, users are able to create up to 50 videos at 480p resolution and those videos last up to 5 seconds. Those paying for ChatGPT Pro enjoy unlimited generations and can create 20 second videos with 1080p resolution. This limited, exclusive, and short access comes at a higher price point for Sora than other comparable AI video generators. To analyze, a 5 second video equates to about $0.40. Many users and smaller businesses may need to carefully consider whether or not this subscription price and exclusive access is worth it. With similar AI video models costing a penny per generation, users need to evaluate the cost effectiveness of these models.