Aerial establishing shot with Gemini Omni Flash: 1080p or 4K cost

An aerial establishing shot on Gemini Omni Flash 1.1 costs $1.50 for 8 seconds at 1080p and $3.00 at 4K on Sume. The model makes 3 to 10 second clips in 16:9 or 9:16, so a drone-style opener fits in one request and needs no source footage.

Cost by length

The rates are fal's list price times 1.25. The 4K tier is double the 1080p rate per second.

Know what 4K means here

Google's launch post describes upscaling to 1080p or 4K as a step after you have a clip you like. The model generates, then you upscale; treat the 4K tier as an upscaled result. For an establishing shot that will play on a phone, the 1080p tier is enough, and 4K earns its price only if the clip will be cropped or shown on a large screen. The longer note is in what the 4K option means.

A prompt that reads as drone footage

State the altitude, the movement and the destination of the shot in that order.

• Altitude and angle: high, looking down at about 45 degrees.
• Movement: a slow forward glide, then a gentle descent toward the harbor.
• Subject: the town, the water line, the one landmark that must be recognizable.
• Light and time: early morning, low sun, haze over the water.
• Ending: where the camera stops, so the clip has a last beat.

Vertical or horizontal

Sume lists 16:9 and 9:16 for this model and nothing else. A square or 4:5 crop has to be made afterwards. If the shot goes into a vertical short, request 9:16 directly; cropping a 16:9 aerial down the middle loses the coastline or skyline that made it an establishing shot.

Native audio is always on and cannot be switched off. An aerial clip will arrive with generated wind or ambience, so if you plan a music bed, plan the mix: the timeline's `soundtrack` accepts a bed with duck_db, but ducking needs a real audio spine, not silence.

Write the ending so the clip has a last beat

Models given an open-ended movement tend to just stop. A closing instruction such as "the camera slows and settles over the harbor wall" gives the 8 seconds a shape, and it makes the clip easier to cut against a title card or a voice-over. Keep the whole prompt as three or four sentences; the limit is 20,000 characters, but nothing is gained by using it.

If you need more than 10 seconds of aerial footage, make two clips and join them. A hard cut between two different angles of the same place reads as editing; a failed continuation does not.

Draft first

Eight seconds at 360p is $0.30. Check the framing and the movement there, since a wrong altitude is obvious at any resolution, then repeat the winning prompt at 1080p. See the 360p draft cost for the full ladder.