I cloned myself with Gemini Omni in 15 minutes (and it's terrifyingly good)

TL;DR

• 1

  A usable AI avatar came together fast: Claire scanned her face with Google Flow, generated about seven scenes, stitched them in Flow's browser editor, and had a full one-minute promo in roughly 15 minutes.

• 2

  The best moments were genuinely convincing: Claire says some shots looked 90% right, especially side profiles and one serious close-up that even preserved small details like sun damage instead of smoothing her face away.

• 3

  The failures were funny, not subtle: The avatar invented blue nail polish, glasses she does not wear, long wavy hair she recently cut, and several uncanny expressions that made the whole thing feel equal parts impressive and jump-scare.

• 4

  Flow acts like a creative partner, not just a generator: Instead of only making clips, the tool helped storyboard a seven-shot podcast hype video with scene ideas like keyboard close-ups, a chair spin reveal, a HUD overlay, and a call to action.

• 5

  Consistency is still the weak point: Backgrounds, lighting, props, bookshelves, and even room colors changed from shot to shot, which Claire says could likely improve with tighter prompting and more reference images.

• 6

  This widened who can make video: Claire's main takeaway is that someone with little video production knowledge can now brainstorm, direct, generate, and edit a promo solo, which she frames as a real creative unlock for non-video people.