How to Browse the Gemini Omni Prompt Library

Go to geminiomniprompts.org and use the category filters or the "View all (20)" link to reach the full library. Each prompt card shows its aspect ratio, duration, and verification status, and the prompt text is copy-paste ready. The library is organized for people who want a working prompt fast rather than a tutorial: you pick a category, scan the cards, and copy the one that matches your shot.

What you land on

The homepage opens with a featured row of hand-picked prompts and a link to the complete set. The site describes itself as an independent, unofficial prompt library adapted from Google's DeepMind prompt guide and community testing, and states it is not affiliated with Google.

The header summarizes the library at a glance:

Browsing by category

The library is split into 9 categories. Two are visible in the featured set:

• Cinematic — continuous shots, aerial reveals, physics tests, VFX triggers
• Product — commercial-style shots such as the cosmetics liquid splash
• Drone — FPV and aerial movement prompts

Use the category filter to narrow to the kind of shot you need, then open a card. If you don't know which category fits, start with the featured row — it mixes cinematic, product, and drone examples so you can see the range before filtering.

Reading a prompt card

Every card carries metadata before the prompt text itself:

• Aspect ratio — e.g. 16:9 for landscape, 9:16 for vertical/Reels
• Duration — e.g. 6s, 7s, 10s
• Verification status — cards marked VERIFIED were checked against actual output
• Tags — e.g. cinematic, continuous-shot, push-in, physics, shatter

The metadata matters because Gemini Omni responds to aspect ratio and duration when they appear in the opening line of a prompt. A card labeled 16:9 10s is telling you those values are already written into the prompt, so you don't have to add them.

Copying and using a prompt

1. Open the card you want.
2. Copy the prompt text as-is — it is written to be pasted directly.
3. Paste it into Gemini Omni without stripping the camera-motion or trigger language.
4. If you need a different format, change the aspect ratio and duration values in the opening line rather than deleting them.

The prompts are built around specific patterns rather than generic descriptions. For example, the aerial reveal prompt uses explicit camera-motion verbs like "pull-back" and "rotate revealing" instead of a generic "drone shot," and the VFX prompt uses "when X happens, Y changes" trigger syntax. Removing those phrases changes what the model produces, so edit them deliberately.

Checking where a prompt came from

The site states that every prompt cites its source, and it publishes source transparency tiers. Featured prompts name their origin — for example, the cinematic continuous-shot prompt is adapted from a PixVerse hands-on test, the physics shatter prompt from a Medium prompt playbook, and the VFX trigger prompt from DeepMind's official hero demo. If you want to know whether a prompt came from official documentation or community testing, read the citation on the card before using it.

Watching prompts tested

The site runs weekly video walkthroughs under "Watch each prompt tested live." Each video shows the prompt, the actual Gemini Omni output, and commentary on what holds up. This is the fastest way to judge a prompt before committing to it, since the card alone doesn't show the result.

Common snags

• Treating it as an official Google resource. The site says plainly it is unofficial and not affiliated with Google. Prompts are adapted, not published by Google.
• Assuming everything is verified. Only cards marked VERIFIED carry that label; the site distinguishes verified community testing from other sources through its transparency tiers.
• Editing out the technical phrasing. Camera-motion verbs, duration, and aspect ratio are functional parts of the prompt, not decoration.
• Skipping the video. If a prompt's output matters to you, the walkthrough shows the real result rather than the intended one.