The takeaway: Google’s latest AI-video release offers more room to tell a short story. The useful experiment is to see whether a clip communicates your idea while keeping the important details intact.
Google announced Gemini Omni 1.2 Flash on 5 November 2026. Its release post describes clips up to 15 seconds, compared with 10 seconds in Omni 1.1, plus follow-up editing through conversation.
Google also claims stronger reference following and visual consistency. Those are provider claims, rather than results from a ThinkAI test. A product that stays recognisable in a demonstration may still change in your own clip.
Give those extra seconds a job
Imagine a Melbourne maker exploring a short video for a handmade ceramic mug. This is a hypothetical creative brief, not a tested workflow.
A longer clip could show three connected moments: the mug on a table, a closer look at its surface, then the mug beside a notebook. That gives the viewer a small story about using the object.
The extra time is useful only if the sequence becomes clearer. If the handle changes shape between shots, a longer video gives the mistake more time on screen.
Before generating anything, decide what you want to learn. Does a slow camera movement reveal the texture? Does the scene feel welcoming? Would a simple sequence of real photographs communicate the same idea more accurately?
Try a brief you can judge
Use a photograph you own and have permission to upload. For this fictional experiment, a starting brief might be:
“Create a quiet tabletop scene using this mug as the visual reference. Begin with a wide view, move closer to the surface, and finish beside a plain notebook. Keep the mug’s shape, handle and colour unchanged. Do not add a logo, writing or product features.”
That is an example brief, not a promise the model will follow it.
Watch the whole result once for the story and again for the object. Look at the handle, rim and surface pattern as the camera moves. A convincing opening frame is not enough.
Then ask for one revision: perhaps warmer lighting or a slower camera move. Compare the revised clip with the first version. Did it fix the requested detail while keeping the rest?
Google’s model overview describes reference-based creation and editing. Our suggested comparison is a way to examine those advertised capabilities on a brief you understand.
Exploration and product evidence serve different purposes
An imagined scene can help you explain a visual direction to a designer or filmmaker. It should not become evidence that a product has a particular finish, size or feature.
For a sales listing, keep real product photography available as the factual reference. If the generated scene changes what a buyer would receive, use it as an internal concept or choose another asset.
Check access before planning around it
Google says the release is generally available through the Gemini API and Gemini for business. That does not establish access or identical features for every Australian account.
Check the model offered in your chosen service, its current charges and the terms for uploaded material before starting. This article does not confirm an Australian subscription entitlement or quote an Australian-dollar price.
A worthwhile first result may simply be a clearer creative brief. You do not need to publish the generated clip for the experiment to have been useful.