A great shot. One wrong detail. Should you edit it or regenerate everything? Google says Gemini Omni 1.2 Flash can create 15-second clips and refine them through follow-up requests. This short illustrated guide looks at a practical question: after an edit, what happens to the subject, camera and action? I use an original robot-and-umbrella miniature to explain a single-change workflow: keep the original, request one change, compare the result, then count every output needed to reach an accepted shot. The pricing example uses Google's published 720p video-output table: 5,792 tokens per second at $12 per million video tokens. That is approximately $1.04 for 15 seconds, or $3.13 for three such outputs. Input and other charges are extra. Prices were checked on 9 November 2026 and may change. These visuals are our own AI-assisted illustrations and local animation. They are not Gemini Omni-generated video, a model-quality benchmark, or a bill from our own API test. We have not tested how many attempts a particular Omni shot needs, or whether a follow-up request reduces its total cost. CHAPTERS 00:00 Edit or restart? 00:13 A single-change workflow 00:28 Compare what survived 00:41 Resolution and output pricing 00:58 Count the outputs 01:13 Attempts still count 01:25 Judge the accepted shot SOURCES Google announcement: https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemini-omni-1-2-flash/ Gemini API pricing: https://ai.google.dev/gemini-api/docs/pricing#gemini-omni-1.2-flash PRODUCTION Original AI-assisted robot illustrations, local 2.5D animation, graphics, music and sound design. Synthetic English narration: Supertonic-3 M1 (BigScience Open RAIL-M). Source illustrations are 1672×941; local compositions were rendered at 720p and enlarged for the 1080p exports. No affiliation with Google. #GeminiOmni #AIVideo #VideoEditing