Avatar video with a product image: the premium is 1-3 cents a second

A product image on a Sume Avatar 1.0 video adds $0.010 per second on the standard tier, $0.013 on plus and $0.030 on max. For a 30-second holiday ad that is 30 cents, 39 cents or 90 cents on top of the base price.

That is small enough that the real question is not cost. It is whether the clip is better with the product in frame. This post gives the arithmetic and the request, so you can decide per ad.

The rates

The avatar video docs say to omit product_image for a productless video. When you send it, the rate moves to the product column of the pricing package. The charged rates are per second of output, by quality tier.

A 12-second ad, three ways

Take a 12-second ad: a 3-second hook, a 4-second demo beat and a 5-second call to action, as in the multi-scene example in the docs. The plan is estimated at its planned length, so the arithmetic is simple.

On plus, 12 seconds without a product is $2.94 and with a product is $3.096, about $3.10. On standard it is $2.208 against $2.328. On max it is $6.60 against $6.96. Across ten hook variants on plus, the product adds about $1.56 in total, which is less than one extra variant would cost.

Treat these as estimates from the rate table. The job result is the billing record, and a request with the wrong duration or tier will not match your arithmetic.

The request

Here is the request shape. The fields that matter for the premium are product_image and quality.

curl -X POST https://api.sume.com/v1/avatar-1.0/talking-video \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: avatar-product-001" \
  -d '{
    "avatar_handle": "product_host",
    "script": "Gift season is here. This one ships in two days.",
    "product_image": "https://example.com/product.png",
    "quality": "plus",
    "aspect_ratio": "9:16"
  }'

When it earns its cent

product_image is a public HTTPS image URL, as the media inputs page describes. Media fields must be fetchable, and Sume rejects localhost, private-network and non-HTTPS URLs before it submits anything.

Use the product image when the product is the point: a gadget, a bottle, a gift box. Skip it when the clip is a talking-head offer announcement where the product appears in the end card. And if you are unsure, preview first. An avatar video preview renders first-frame stills so you can see how the product sits before the full render, and you can then pick a tier at generate time.

Tier first, product second

Tiers matter more than the product line. The standard tier is the fastest Sume execution path, plus is the default balanced path, and max is the highest quality and slower. Moving a 12-second ad from plus to max costs $3.66 more, which is more than ten times the product premium on plus.

So the order of decisions for a holiday batch is: choose the tier by how long the clip will live, then decide on the product image by whether the product must be recognizable. A flash-sale clip that runs for 48 hours can sit on standard. A hero clip that carries the whole campaign deserves max.

Mistakes to avoid

Three things go wrong with product-image avatar ads, and none of them is cost.

• Sending a cluttered product photo. Use a clean image with the product centered, since that is the reference the render has to reproduce.
• Changing the script without a new preview. Structural fields such as the script, avatar handle, scene and aspect ratio need a new preview, so approve the stills only after the script is final.
• Forgetting the 4 to 60 second window. Sume rejects plans that estimate outside it, so cut long scripts or split them into two jobs.

The batch view

At ten variants of a 12-second ad on plus, the whole batch with the product image is about $31.0, and without it about $29.4. If the product is what you are selling, that is a rounding error against the media spend that will run behind the clip. If the clip is a pure offer announcement, drop the image and keep the cents.

When the premium is worth paying

How to decide whether the premium is worth it. Take the 10-second clip you plan to run. On plus, a no-product clip is $2.45 and a product clip is $2.58, a difference of $0.13. If the product is visible in the shot and it is the point of the ad, that is a small price for a model that is built to keep the item in the frame.

If the product never appears, do not pay it. Use the no-product rate. If you are unsure, render one clip each way at the standard tier, which is $1.84 and $1.94, and compare them before you commit a batch of thirty. The gap across thirty clips is $3.00 on standard, $3.90 on plus and $9.00 on max.

Remember the quality tier moves the bill much more than the product premium does. Moving from standard to max on a 10-second clip adds $3.66, which is far more than the one-to-three-cent difference per second. Pick the tier first, and then decide on the product input.