Google Flow AI Tutorial | Create Satisfying Miniature Art Videos with Just 3 Prompts (step by step)

Summary

The tutorial demonstrates a Google Flow workflow for creating miniature painting videos from location images. The videos feature animated brushwork, realistic painting sounds, and a connected follow-up clip.

Workflow and features

1. Set up a Flow project: Choose Ingredients, set the aspect ratio to 9:16, select the Omni Flash model, set the duration to 10 seconds, and choose one output. The narrator says each clip costs 15 credits.
2. Create the first clip: Upload a clear location image, such as the Mount Fuji example, and include it in a prepared prompt. The prompt uses the reference image and a painting order that describes which elements should appear as the scene is painted.
3. Generate the painting order: Upload the reference image to ChatGPT and use a prepared prompt to generate the order. Then paste the result into the Google Flow prompt.
4. Continue the video: Capture a frame from the first clip using Flow’s frame-capture option. Switch from Ingredients to Frames, set the saved image as the starting frame, and submit a second prepared prompt. The continuation prompt can be adjusted to specify camera movements or other effects.
5. Reuse the workflow: Duplicate the prompts for another location, replace the reference image, and generate a new painting order. The tutorial cautions that unclear or difficult-to-interpret images may produce less faithful results.
6. Combine and export: Add both clips to the same scene in Flow, then download the combined video.

The tutorial emphasizes that image choice and prompt adjustments affect the results. It presents the workflow as reusable across different locations and images.

Main speaker and sources

• Main speaker: Narrator from the 3z Visuals channel
• Tools demonstrated: Google Flow for video generation and editing; ChatGPT for generating painting orders