To the Gemini Development Team,
I am writing to report a recurring issue with the Gemini Omni Flash video generation model. When attempting simple video-to-video edits—specifically, trying to replace a character in a video using a static image reference—the model frequently fails or rejects the prompt entirely.
The core problem is that the system does not provide any specific violation reason or error log explaining why it failed. It simply stops working or gives a generic failure message. This makes it impossible as a user to understand if I am hitting a safety filter, a technical limitation of Omni Flash, or just a random error. This overly strict and unexplained rejection process creates a very frustrating user experience, especially when other video generation models handle these exact same prompts and complex actions smoothly. Please review the safety filters and error reporting for video editing tasks so users can get clear, actionable feedback when a generation fails.
Thank you.