Google DeepMind has released Gemini Omni 1.2 Flash, and one of its showcase examples is a first-person-view (FPV) camera shot that moves through the Pantheon in Rome.
Google DeepMind · Gemini Omni 1.2 Flash
A new Flash model takes on a complex camera move through Rome's Colosseum
Google DeepMind's showcase includes a first-person camera shot moving through the Pantheon. A follow-up test uses the same kind of cinematic move at the Colosseum, comparing the new model with its predecessor on spatial continuity.
1
shared prompt used for both model versions
2
model versions compared: Omni 1.2 Flash and Omni 1.1 Flash
3
stages in the camera path: underground, then upward
The camera path the prompt asked for
A multi-stage move that tests spatial continuity from below ground to the upper structure
Stage 1
Hypogeum maze
Underground passages beneath the arena floor
→
Stage 2
Climb upward
Camera rises through the structure
→
Stage 3
Coherent layout
Geometry and perspective must stay consistent throughout
Omni 1.2 Flash · newer model
Received the same prompt as the baseline and is the version Google DeepMind highlighted for this release.
Omni 1.1 Flash · baseline
The predecessor, used as a reference point for judging how much the newer model has changed.
What the evidence does and does not show
The comparison is limited to a few curated examples. Broader quality claims remain unverified.
Available
The showcase example and one side-by-side test with the same prompt.
Not yet available
Independent benchmarks, pricing, latency and cost figures, and broader user reception.
Still to test
Other locations, prompt styles, and camera movements, to see how consistent results are.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article...