Create an AI Character Video Series with Google Flow

Take a guided tour of Google Flow, beginning with how to access the full workspace. You’ll learn how projects and collections are organized, where to find your images, videos, and characters, and how to create collections to keep your media organized.

Starting with a blank Google Flow project can be confusing when the video options and project media you expect to see haven’t appeared yet. In this lecture, I’ll take you from a completely new project through creating and naming your characters, adding reference images, choosing voices, selecting the video format and aspect ratio, and submitting your first prompt.

You’ll then see the resulting 10-second video appear inside the project, making the video and media areas much easier to understand and use. This practical walkthrough will help you begin your own project confidently before moving on to the more detailed prompting and production techniques covered later in the course.

Explore the Google Flow prompt area and learn the difference between an individual prompt, project-wide Agent Instructions, and Agent Settings. You’ll also see how to set generation defaults, require confirmation before spending credits, and estimate the video length needed for your dialogue.

Review how to select the correct project, video model, aspect ratio, and clip length before generating a simple test video. You’ll also see how to open the Agent sidebar, approve or reject a generation before spending credits, and discover what happens when the prompt is too vague.

Learn how to divide a short story into connected scenes, time your dialogue, choose an appropriate clip length, and keep characters, clothing, locations, and sound consistent. Using a real Tabbys episode, I’ll show how planning reduces unwanted AI additions, unnecessary generations, and wasted credits.

Discover how vague, overloaded, or contradictory instructions can send an AI-generated video in an unexpected direction—and why a longer prompt is not necessarily a better prompt.

Learn the essential components of an effective video prompt, including the characters, location, action, dialogue, camera framing, sound, and any important exclusions.

Learn which technical choices are selected in Google Flow and how to organize recurring continuity requirements and scene-specific instructions within the written prompt. Plus, download "DEMO showing Preamble and Scene Request" plus "SCENES TWO and THREE" .txt files.

Learn how to create reusable character references with consistent physical features, personalities, and voices. You’ll also see a successful George generation, an unsuccessful Maggie attempt, and why patience and experimentation are essential when working with rapidly developing AI technology.

Follow the complete workflow as we use ChatGPT to develop a simple three-scene story, complete the downloadable scene template, combine the recurring preamble with each individual scene, and generate and evaluate all three clips. You’ll also see what happened when I accidentally omitted the character references—and how the detailed character-consistency instructions in the preamble still helped Flow create recognizable versions of George and Maggie.

Learn how to evaluate imperfect AI-generated footage and decide whether to keep it, regenerate it, or work around the problem during editing. Using real examples—including misplaced characters, impossible cars, and a steering wheel on the wrong side—you’ll see how quality and purpose influence the decision.

Learn how to create a reliable folder structure for your Camtasia projects, use consistent filenames, and keep source clips, project files, audio, captions, notes, and final exports organized and safely connected.

Learn how to download your completed AI video clips from Google Flow, organize them in your project folder, and import them into Camtasia—or another video editor. You’ll also see how easy it is to place clips on the timeline and trim away footage you don’t want to use.

Learn how to export your finished video’s audio from Camtasia, use Whisper Transcription to create a VTT subtitle file, and export the completed video. You’ll also learn the difference between VTT and SRT subtitles and how to check caption timing before uploading. A reusable VTT Timing Check for ChatGPT prompt is included in the lecture resources.

Learn how to make AI-generated footage work for your project without immediately regenerating it. You’ll see how to trim unwanted footage, split and rearrange clips, mirror a scene to improve its direction or continuity, and rescue the usable parts of an imperfect generation. The techniques are demonstrated in Camtasia but can be applied in many video-editing programs.

See how AI-generated dialogue can be replaced with your own voice using Camtasia’s Voice Narration tool. Although the new narration could not be recorded live during the screen recording, you’ll see the setup, earlier timing attempts, how the replacement audio is aligned with the original clip, and the finished overdub matched to the character’s mouth movements.

Follow along as three AI-generated scenes are arranged, trimmed, and refined into one finished video. You’ll learn how to choose the best parts, remove awkward moments, improve continuity, and align the clips, music, and outro on the timeline.

Learn how to reuse a completed Camtasia project as the template for your next episode. You’ll replace the old scenes with new clips while retaining the existing branding, sizing, track arrangement, and other finishing elements that keep your series consistent.

Discover the most valuable lessons I learned while creating The Tabbys, including planning for continuity, improving prompts through experience, accepting wasted credits, choosing "usable" over "perfect," and knowing when to repair a clip or start again.

George and Maggie return with a special final message to thank you for joining the course and encourage you to have fun creating your own AI character series.

Stay current with newly discovered Google Flow features, practical workarounds, troubleshooting solutions, and helpful tips added as the platform continues to evolve.

Learn how to fix character drift between multi-clip scenes by using a screenshot from a successful clip as a visual reference. This workflow also helps preserve existing clips and maintain consistent faces, clothing, and character appearance.