Product Page Video From One Packshot for Under $2.50

The short answer

A 15-second product-page video from a single packshot costs about $2.30 on Sume list prices: background removal ($0.0225), three scene stills with Nano Banana 2.1 at 1K ($0.30), three 5-second Omni Flash clips at 720p ($1.875) and one timeline render ($0.10). Figures come from the catalog read on 2026-10-08.

The chain

Start with the product photo. rmbg_create (background removal) takes it to a clean cutout. Use that cutout as a reference in image_urls for three scene prompts: on a kitchen counter, in a hand, on a shelf. Then send each accepted still to Gemini Omni Flash 1.1 for a 5-second image-to-video clip, 9:16 or 16:9. Finally assemble the three clips with a Timeline 1.0 render; the audio can be a short music bed or a silence spine.

Google's pricing page lists Nano Banana 2.1 as a current model (read 2026-10-09), which is why it is a natural default for the stills; the numbers below are Sume's, not Google's.

Keep the product real

Image models can change a label, a logo or a proportion. Inspect each still against the packshot before you spend on video, because a wrong still becomes a wrong clip. Video costs 4 to 5 times as much per item as a still here, so reject at the still stage.

Native Omni audio is always on, so the raw clips carry generated sound. Put your own spine in the timeline and the clip sound is replaced by it.

• Run generation_admission_preview or dry_run before a batch of product videos.
• Stills at 2K cost $0.15 each; the three-still line would be $0.45.
• If the product page needs 4:3 or square, check supported_aspect_ratios first: Omni takes only 16:9 and 9:16.

Where to spend more

The cheapest change that helps most is a better still. If a scene still fails review, a second 1K try costs $0.10, but a clip made from the failed still costs $0.625 for 5 seconds at 720p. Only the accepted still should reach the video step.

A hero product page might also use a 1080p clip: Omni Flash at 1080p is $0.1875 per second, so one 5-second hero clip is $0.9375 instead of $0.625. Do that for one clip, not all three.

Keep every accepted still and clip as a Sume media URL. The same three scene stills can feed a 4:5 social image set at no further generation cost beyond the crop, and the packshot cutout can be reused for the next colorway. That is where the first run's spend gets paid back: only the video line is new work.