Google Ads built-in image and video generation vs a Sume pipeline

Google Ads can generate images from a prompt inside Demand Gen, with up to 20 selectable results per prompt and a choice of aspect ratio, and can auto-generate video in horizontal, square and vertical from a logo, two images, two generic texts and a business name. A separate Sume pipeline gives you per-request model and ratio control and Sume-hosted result files, at the cost of one more step before upload.

What Google's tools take as input

The Google Ads Help page on generative AI in Google Ads describes two features. Facts below were read on 2026-10-03.

Where a Sume pipeline differs

Both routes can produce usable assets. The differences are in what you control and where the files live. The Sume facts below come from the Image API and video docs.

When each makes sense

Use the built-in tools when speed inside the account matters more than control, such as a quick test of a Demand Gen group from existing brand images. Use a separate pipeline when you need the same character or product across ratios, a fixed model for comparison tests, or the same assets reused on other channels.

A caveat for sume/auto: the docs state that responses echo sume/auto and the family that ran is never disclosed, so pin a model id when a test needs one fixed model.

A minimal Sume side

A video request for a vertical placement looks like this. It follows the documented POST /v1/videos shape, with a retry-safe header.

curl -X POST https://api.sume.com/v1/videos \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: dg-hero-001" \
  -d '{
    "model": "sume/auto",
    "prompt": "A vertical product clip on a desk, natural light, slow push-in",
    "aspect_ratio": "9:16",
    "duration": 5
  }'

What this does not decide

Neither route settles ad policy, disclosure, or whether a synthetic asset suits your brand. Those checks apply to the finished creative either way, so run the same review on both.