The Verge - OpenAI’s Sora is an impressive first step, but it’s still plagued with oddities and inconsistencies
The iOSaur video generation tool has been released recently and has garnered mixed reviews. While it offers impressive features such as fast video generation, consistent fur and textile patterns, and impressive lighting, shadow, and mirror effects, it also has significant drawbacks. Users have reported visual errors in simple prompts, such as additional limbs and distorted backgrounds, and more complex prompts result in even more issues, particularly with human motion and eating actions. The tool has guardrails to prevent the simulation of politicians and celebrities, but it can still produce concerning content, like toddlers in swimsuits or wielding guns. Despite its capabilities, iOSaur does not yet feel like it is democratizing content creation, as it requires a $200 monthly subscription and significant skill, time, and energy to produce usable footage. Comparatively, its results are better than Runway AI and on par with Adobe's Firefly video model.
Key Points:
- iOSaur offers fast video generation with consistent patterns and impressive effects.
- Visual errors are common, especially in complex prompts, affecting usability.
- Guardrails exist but concerning content can still be generated.
- High cost ($200/month) and skill requirement limit accessibility.
- Results are better than Runway AI and comparable to Adobe's Firefly.