Coworking day-pass promo: a price card stacked over a room video

A coworking day-pass promo with a price card on top and a room video below costs about $0.87 on Sume: a 6-second Gemini Omni Flash 1.1 clip at 720p ($0.75), one Nano Banana 2.1 card ($0.10) and one Timeline compose job ($0.02). Compose puts a still and a video on screen at the same time and returns one MP4.

This is the cheap way to get an offer on screen without asking a video model to spell words. You make the card as a still, the room as a video, and let compose stack them. Prices come from the Sume pricing code and the Timeline compose page (read 2026-10-10).

What is the endpoint chain?

Three jobs, two of which can run at the same time because neither depends on the other.

• POST /v1/videos with gemini-omni-flash-1.1, duration: 6, resolution: 720p, aspect_ratio: 9:16: a slow pan across desks, no people.
• POST /v1/images with google/nano-banana-2.1: a plain card background with a clear lower half. Do not ask for words; leave the area empty.
• POST /v1/timeline-1.0/compose with operation: stack, the card as image.url, the clip as video.url, layout: {split: horizontal, image_region: top, ratio: 0.35} and output: {width: 720, height: 1280}.
• Optional: POST /v1/video-captions with cues for 'Day pass $25' if you want the price as live text, at $0.20.

What does it cost?

Omni Flash bills output seconds by resolution: $0.0375 at 360p, $0.125 at 720p, $0.1875 at 1080p and $0.375 at 4K. It accepts 3 to 10 seconds.

What are the limits?

Compose needs the still to probe as a still and the clip as video, both on media.sume.com in your workspace. The output length comes from the video layer, with a 300-second ceiling, and the still stays up for the whole clip. If you ask for more video than the source has, the job clamps it and says so in a warning.

Omni Flash is 16:9 or 9:16 only, so a square promo needs another model or a crop through Video filter. It also takes 3 to 10 seconds, which matters because some of the Sume video docs say every non-Wan model tops out at 15; for Omni the real ceiling is 10.

Put the compose output into a Timeline render if you want a music bed or a fade out; that adds a render at $0.10 per started minute. The Timeline compose page shows the stack and overlay layouts.

When is a different shot better?

If the room has to look like your room, give Omni a reference image of it instead of a pure text prompt; the model takes image and video references, no audio. If you want to test a hook first, a 360p draft is $0.225 for 6 seconds, as in the 480p proof loop, and the 720p is only worth paying for once the card layout is approved.

Pre-publish checklist

Look at the card on a phone-sized preview. The stack ratio of 0.35 leaves the card the top third; if your price is long, raise it to 0.4 and let the room shrink.

Compose output is an ordinary MP4 in your workspace, so it can go into a Timeline render with a music bed from the free BGM catalog without any new upload.

• Ask the image model for an empty lower half and never for words.
• Use audio only through Timeline; the compose job itself does not mix sound.
• Re-run compose alone if the price changes and you used a caption overlay for it.