Gemini Omni Flash camera words: smartphone zoom, film, whip pan

The camera and shot vocabulary Google's Omni guide lists, from locked off to dolly zoom and webcam style, with a Sume test grid to compare them cheaply at 360p.

5 min readSume
All posts

Google's Omni prompt guide lists camera wording in four groups: continuity ("one continuous shot", "oner"), angle ("static", "locked off", "fixed"), movement ("push in", "punch in", "dolly zoom") and camera type ("natural smartphone zoom", "film camera", "webcam style"). Use one term from a group per clip and test the rest at 360p before paying for a final render.

The vocabulary below is from Google DeepMind's Omni prompt guide, read on 2026-10-03. Google's launch post for Omni 1.1 Flash describes 360p as a draft mode that is up to 60% faster and one-third the cost of 720p, which is what makes a test grid affordable.

Camera words by job

The guide's own example combines a low-angle close-up with a whip-pan transition, so movement words can be paired with shot sizes. It also names wide-angle, medium and close-up shot sizes, and asks you to describe how the camera moves, for example gliding gently or rushing suddenly. Treat the table as a menu, not a recipe: Google does not say which combinations are reliable.

Camera wording from Google's Omni guide (read 2026-10-03)
JobWords in the guideTry it for
Shot sizewide-angle, medium, close-upProduct hero versus context
Fixed camerastatic, locked off, fixedPours, unboxings, desk shots
Move closerpush in, punch inReveal a detail
Perspective shiftdolly zoomOne dramatic beat
Phone-shot feelnatural smartphone zoomUGC-style ads
Other camera looksfilm camera, webcam styleRetro or call-style framing
Transition inside a shotwhip-panHard turn to a second subject

A cheap test grid

Keep the subject, light and action identical and change only the camera phrase. Run the same prompt six times with six camera phrases at 360p and 3 seconds. Google's page lists 360p as the draft resolution, and Sume's Video Router accepts 3 to 10 seconds at 360p through 4K. At Google's $0.10 per second at 720p and a third of that for 360p, six 3 second drafts are 18 seconds of output; the exact Sume rate by resolution is in the live catalog, so read it there before you plan a budget.

Use a distinct Idempotency-Key for each variant, otherwise Sume will treat a repeat key as the same request. Store each job id next to the camera phrase it used, so you can match the clips to the words afterward.

Run one variant

Change only the camera phrase between requests. Native audio is always on, so each draft also tells you whether the sound suits that shot.

curl -X POST https://api.sume.com/v1/video-router/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: camera-grid-smartphone-001" \
  -d '{
    "model": "gemini-omni-flash-1.1",
    "prompt": "Natural smartphone zoom. A street vendor flips a flatbread on a hot griddle, steam rising, evening neon.",
    "resolution": "360p",
    "duration": 3,
    "aspect_ratio": "9:16",
    "mode": "async"
  }'

Pick a winner and re-render

After the grid, keep the phrase that gave the camera behavior you wanted and rerun it once at 720p or higher. Google describes 1080p and 4K as upscaled outputs, so the camera behavior you see at 720p is what the upscale will keep. See the Video Router docs for the resolution and duration limits per model.

Sources

Related posts

More in Models

All Models posts

Written by Sume