This video teaches how to automate image and video generation inside Google Flow (agent mode) using a provided prompt file. It covers three automation types: text-to-image with character reference, text-to-video without reference, and image-to-video for consistent characters. The process includes handling failed generations by reusing prompts.
A short editorial from the VEONIB team on why this content matters.
A practical method for bulk AI image/video generation using Google Flow agent, with character consistency and failure retry.
Distinctive for using a prompt file and sequential scene prompts, enabling scalable automation without extensions.
Content creators should adopt this to automate storyboard visuals and reduce manual effort.
Google's AI agent platform for automating tasks like image and video generation.
A setting in Google Flow that enables multi-step automated workflows.
Maintaining the same character appearance across multiple generated scenes.
Generating images from text prompts using a reference image for consistency.
Generating videos directly from text prompts without any image reference.
Generating videos from a sequence of uploaded images to ensure character consistency.
A downloadable file containing pre-written automation prompts for different generation types.
What is required to automate image generation in Google Flow?
A character reference image, prompt file, and agent mode enabled.
How do I ensure character consistency across scenes?
Upload a reference image and add it to prompt before pasting scene prompts.
Can I generate videos without any image reference?
Yes, use text-to-video automation with same prompts but no reference image.
How do I handle failed video generations?
Click 'reuse prompt' below the failed video and regenerate, or wait for agent auto-retry.
What model should I use for video generation?
Google VO 3.1 light, with output count set to one.
How many scenes can I generate at once?
You can generate 5, 10, 20 or more scenes by pasting all prompts.
Do I need to number scene prompts?
Yes, each prompt must include its scene number (e.g., Scene one, Scene two).
What if generation fails due to high demand?
Manually reuse the prompt for that specific video to restart.
How do I upload images for image-to-video?
Upload all scene images in correct sequence, then add them to prompt one by one in order.
Is the prompt file provided?
Yes, in video description and Telegram channel.