Runway product swap vs a Sume prompted video edit
Runway's Product Swap Recipe takes a video plus product images. Sume's closest route is one edit call: a video_url and a prompt that names the swap.

Runway's Product Swap Recipe takes a reference video, an image of the original product and one to ten images of the new product, and returns the same ad with the product replaced. Sume has no recipe like that. The closest route is a Video Router edit on gemini-omni-flash-1.1: send the clip as video_url and describe the swap in the prompt. It takes no product images.
Runway facts are from its Product Swap page; Sume facts from the Video Router docs, both read 2026-10-01.
What does the Runway recipe take?
The page says it preserves the original camera motion, lighting and scene composition. It works best for like-for-like swaps, such as one water bottle for another, and struggles when the new product could not perform the action in the video.
| Runway input | Runway rule | Sume equivalent |
|---|---|---|
referenceVideo | 1.8 to 15 seconds | video_url, the edit source |
originalProductImage | Required | Name the old product in the prompt |
newProductImages | 1 to 10, optional view | Not accepted in an edit call |
resolution | 720p or 1080p | resolution, default 720p |
audio | Default true | Native audio always on |
How does the Sume edit call work?
In the video_to_video capability the prompt describes the edit, resolution is optional (default 720p), and aspect_ratio and duration are not sent. The docs' own example is "Replace the bottle with an apple. Keep everything else the same." Native audio is always on, and generate_audio: false is rejected.
const res = await fetch("https://api.sume.com/v1/video-router/generate", {
method: "POST",
headers: {
Authorization: "Bearer " + process.env.SUME_API_KEY,
"Content-Type": "application/json",
"Idempotency-Key": "swap-bottle-001",
},
body: JSON.stringify({
model: "gemini-omni-flash-1.1",
prompt: "Replace the glass bottle with a tall aluminum can. Keep everything else the same.",
video_url: "https://example.com/winning-ad.mp4",
resolution: "720p",
mode: "async",
}),
});
console.log(await res.json());What is the real difference?
video_url is the edit source, not a reference: it cannot be combined with image_url, end_image_url or reference_*_urls. So the new product can only be described in words. If the exact look of the new item matters, test a few prompts on a short clip first. The related post on Gemini Omni edit prompts covers wording.
Should I retry a swap with the same key?
Only for an exact retry. A replay with the same Idempotency-Key returns the original job instead of making a new one, so a changed prompt needs a new key. See idempotency keys for AI video APIs.
Sources
Related posts
- Edits Restyle with a prompt vs Sume video router edit mode
- One video ad into many variants with the same pacing, by API
- Gemini Omni video edit prompts that keep everything else the same
- Gemini Omni Flash video edit: 5 reference images vs Sume
- Idempotency keys for AI video APIs: retry without paying twice
More in Use cases
- AI presenter for a SaaS demo video, with or without a product image
For a SaaS demo, omit `product_image` for a productless avatar video, or add one. Each Sume avatar job runs 4 to 60 seconds, so longer demos are split.
- AI video background swap for a fashion lookbook with Seedance 2.5
Seedance 2.5 describes green screen editing that swaps a background and keeps the subject. On Sume, send the garment clip as a video reference, 4 to 30 seconds.
- Seedance 2.5 green screen editing: what Sume can do
ByteDance says Seedance 2.5 improves green screen editing. Sume's docs document an edit mode only for Gemini Omni Flash 1.1; Seedance takes video references.
- AI workout video generator: Seedance 2.5, 4 to 30 seconds
For a workout demo, one Seedance 2.5 request runs 4 to 30 seconds on Sume, with video and audio references; most other catalog models stop at 15 seconds.
Written by Sume