12 November 2026 EN ES
The AI Side Project

Weekend projects that quietly became businesses

Costs

Add a 15-Second Video Feature for About $1 Per Clip

Turn a weekend video feature into a controlled experiment: draft cheap, fix a detail, upscale only the winner, and check the bill before it surprises you.

Illustration: Add a 15-Second Video Feature for About $1 Per Clip

You have a weekend, a side project, and a feature that would be easier to explain with a short clip. The question is whether video can stay inside a budget that does not punish a retry.

Treat the clip as a small-budget test. A 15-second 720p clip costs about $1.05, so the feature can stay a weekend experiment. The rest is keeping the scope small.

The clip is the unit of work

Google made Gemini Omni 1.2 Flash generally available in the Gemini API and Gemini for business. That gives a side project a practical path to short video.

You can choose clip lengths from 3 to 15 seconds in one-second increments, up from the 10-second maximum in Gemini Omni 1.1. That extra room matters when a product demo needs a setup, a payoff, and a call to action.

The 720p price is $0.07 per second, a 30% reduction from Gemini Omni 1.1. For a weekend builder, the unit cost is low enough to test a few directions.

When the clip is the unit, the pricing question gets easier. You can compare a teaser, a demo, and a final cut by counting clips, not by guessing at render time. The bill is a count of clips.

Before you write code, decide how many clips you will make. A small number keeps the experiment honest. If the feature is a product teaser, a teaser clip may be enough. If it is a demo, a few clips may be needed. Stop before the weekend turns into a render queue.

Build the weekend pipeline before the invoice

Keep the workflow in a few moves. Each one should be small enough to finish before the next coffee goes cold.

  • Drafting at 360p or 720p is part of the workflow. The rough cut is good enough to judge motion, pacing, and whether the idea survives contact with real footage.
  • A follow-up edit fixes details. It can correct a logo, a hand, a caption, or a timing slip without throwing away the whole generation.
  • Upscaling to 1080p or 4K is for the selected clip. The final file should be polished enough for a landing page, a demo, or a short social post.
  • Check model routing before billing, because beginning December 3, 2026, requests to Gemini Omni 1.1 are routed to Gemini Omni 1.2 and billed at Gemini Omni 1.2 rates. A small request that confirms the active model prevents a set of clips from spending the weekend.

Write the cost check into the app, not just the notes. A tiny log line can live next to the video request. It records the model, the length, and the resolution, so the bill is easy to read. If the number looks wrong, you can pause the feature before it spends the weekend.

Keep the feature small enough to ship

Start with the smallest clip that proves the feature. A short explanatory moment is enough. Keep the first cut disposable. If the draft feels weak, do not add another scene. The low-resolution draft is the place to argue about the idea. Change the prompt, the angle, or the ending, then draft again at low resolution.

Reserve the polished version for the moment you are ready to show it. Upscaling belongs at the end, after the idea is settled. If you are still deciding what the video should say, keep the file small and cheap. When the idea is clear, the final clip can carry the project without turning a weekend experiment into an unexpected cost.

Ship the feature behind a simple switch. A switch also makes it easy to remove the feature if the bill or the quality is not right. If the video is not ready, the page still works. If the video is ready, it can appear where the user expects it. That keeps the side project usable while the clip is still being tested.

Advertisement