I’m trying to implement VEO 3.1 extend video capability for my project. Here’s the step by step workflow:
A process is responsible of generating videos. After it generates a video using VEO 3.1, it saves it to GCS.
When using extend video capability, another process downloads the generated file from GCS and uploads it using files API for model to use. When I use the uri of the uploaded file as video parameter for extend video API, I’m getting an error response:
VideoGenExtend Error: {“error”:{“code”:400,“message”:“Input video must be a video that was generated by VEO that has been processed.”,“status”:“INVALID_ARGUMENT”}}
Can you confirm if the error is related to using a different uri (from files api) that is not from video generation result ? If so, it doesn’t make sense because the generated videos have 2 days retention period.
Thanks for reaching out to us. Could you please confirm whether you’re passing the original VEO-generated video reference into the extend video API, or a URI from a video that was downloaded and re-uploaded?
Thank you for your response. To extend videos previously generated using Veo, please note that the Gemini API supports video extensions only for Veo-generated videos. The video must come from a prior generation, such as operation.response.generated_videos[0].video.
For more details, please refer to the documentation.
Hi @Nireeksha_K_A
Yes, I know it. The issue I want to resolve is:
As it stated in the documentation:
”Generated videos are stored on the server for 2 days, after which they are removed. To save a local copy, you must download your video within 2 days of generation. Extended videos are treated as newly generated videos.”
If I want to use extend video on a veo generated video, there’s no way to do it after 2 days. Is that correct ?