Walmart holiday food ad: product in and out of its pack, $1.10 on Sume
Walmart Connect advises food ads that show the product in and out of its pack, 6-15 s. Two 4-second Omni clips plus a Timeline stitch cost $1.10 on Sume.

Walmart Connect's video guide says to show food and beverage products both in and out of their packaging, with bright colors, high contrast, and eye-catching animation, and recommends 6 to 15 seconds for the category (Walmart Connect video guide, read 2026-10-05). Two 4-second Gemini Omni Flash 1.1 clips at 720p, one on the pack and one on the plated product, plus a Timeline stitch, make an 8-second spot for $1.10.
Two shots, two jobs
Shot one is the pack: a clean packshot, first frame pinned with frame_images and frame_type: "first_frame". Shot two is the product out of the pack, served: the recipe idea. Walmart's holiday guidance says to pair product images with preparation ideas in food so the result looks appetizing and festive (Walmart Connect holiday guidance, read 2026-10-05).
Omni takes 3 to 10 seconds per clip, so 4 seconds is inside its range. At 720p it lists at $0.10 per second, billed at list times 1.25 = $0.125 per second, which is $0.50 per 4-second clip.
| Item | Walmart guidance | Sume job | Price |
|---|---|---|---|
| Shot 1: pack | Product in packaging | Omni 720p, 4 s | $0.50 |
| Shot 2: served | Out of packaging; recipe idea | Omni 720p, 4 s | $0.50 |
| Stitch | Whole spot 6-15 s | Timeline, first minute | $0.10 |
| Total | 8 s | $1.10 |
What to put in the prompt
Keep the two clips consistent: same surface, same light, same angle family, so the cut between them reads as one scene. Reuse the first prompt sentence in both and change only the subject state. If the two shots drift, regenerate the second one only; at $0.50 it is the cheaper fix than redoing the pair.
Say what the viewer should see change. For the pack shot, ask for a slow push-in on the box with the lid or seal opening in the last second. For the served shot, ask for the finished dish with steam or a pour, in bright, high-contrast light, since the guide asks for bright colors and contrast. Add holiday cues the page names: sparkle, a decorated setting.
Do not ask the model to render a recipe's text. Add the recipe name as a caption cue ($0.20) so the words are exact.
Keep it truthful
A generated plate is an illustration, not a photograph of your product. Keep the packshot as the first frame so the package is the real one, and avoid claims the shot cannot back up, such as portion size. Walmart also says its videos launch on mute, so rely on the picture and a caption rather than sound.
Sources
Related posts
More in Use cases
- Wan 3.0's 20 reference assets: what an ad request can carry on Sume
Alibaba lists 20 reference assets for Wan 3.0, including documents and webpages. Which kinds can a Sume wan-3.0 request send, and what do you do with the rest?
- Wan 3.0 auto scene splitting vs explicit scene slots in Sume Timeline
Wan 3.0 advertises auto scene splitting inside one generation. If you want control of each scene, build separate clips and join them with Timeline 1.0.
- Wan 3.0: one 30-second prompt or three 10-second clips on Sume
Alibaba says Wan 3.0 splits scenes itself. On Sume, one 30 s request and three 10 s clips cost $3.75 at 720p each way; with three clips a redo costs $1.25.
- Twelve sequential images then animate: Wan 3.0 storyboard cost
Alibaba says Wan 3.0 makes up to 12 sequential images. On Sume, make stills with an image model, then animate each: 12 two-second 480p clips cost $1.50.
Written by Sume