Jewelry ring close-up video at 1080p: Omni on Sume vs Veo Lite

For a ring close-up at 1080p, Gemini Omni on Sume lets you choose any length from 3 to 10 seconds, so a 5-second sparkle costs $0.94. Google's Veo 3.1 page says 1080p supports only 8-second clips (read 2026-10-07), so a short Veo 1080p is not an option. Sume does not list Veo 3.1; this is a comparison of how the two limits shape a small jewelry shot.

Jewelry is one of the harder subjects: reflections, prongs and stone facets change from frame to frame.

The limits side by side

The Veo numbers are Google's published list per second times the clip length; the Omni numbers are Sume's list times 1.25, rounded up to the cent per job.

The prompt

Use a sharp photo of the ring as the first frame; this keeps the stone shape and metal tone. Describe light motion, not the ring: the ring does not need to move, the light does.

{
  "model": "gemini-omni-flash-1.1",
  "prompt": "Macro close-up of the ring on dark velvet. A soft light sweeps slowly across the stone, producing small moving highlights on the facets. Locked-off camera, shallow depth of field, single continuous shot. Audio: very faint shimmer tone. No dialogue.",
  "duration": 5,
  "resolution": "1080p",
  "aspect_ratio": "16:9",
  "frame_images": [
    { "type": "image_url", "image_url": { "url": "https://example.com/ring.jpg" }, "frame_type": "first_frame" }
  ]
}

Why 1080p matters less than the take

Google describes Omni's 1080p and 4K outputs as upscales (read 2026-10-07), so a higher setting makes a bigger file, not necessarily more facet detail. For jewelry the quality gap between takes is larger than the gap between resolutions. Draft at 360p ($0.19 for 5 seconds), pick the take with the cleanest prongs, then re-run at 1080p.

Honest limits

A generated close-up is advertising, not documentation. Do not use it as proof of the stone's size, cut or color; show the real photo for that. The stone may change shape mid-clip, so compare the first and last second to the photo. On Sume each re-run is a new take with no seed, and a failed generation is refunded under the reserve-then-capture billing described in the docs, but a take that is merely ugly is still charged.

A small-brand workflow

For a shop with ten rings, the economical path is to generate one 5-second clip per ring at 360p, review all ten, and re-run only the best six at 1080p. At the rounded per-job prices that is $1.90 for the drafts and $5.64 for six finals, about $7.54 before any re-roll. Keep the photos consistent in angle and background so the clips look like one collection, and keep the original product photos beside the clips on the product page so buyers see the true item.