Seedance 2.5: Make a One-Take Product Video from Three Reference Images
Use Seedance 2.5 on Pixocto to connect character, product, and setting references in one 15-second vertical product video, with an editable prompt.
Seedance 2.5: Make a One-Take Product Video from Three Reference Images
Use Seedance 2.5 to build a one-take product video from three references: one for the character, one for the product, and one for the setting. A single continuous action makes it easier to judge whether the same person, object, and room stay consistent as they move.
Our example uses an orange-and-white sunscreen tube. The shot plan and prompt below are a practice template, not the original prompt or a verified result from an existing video.
Open the video workbench and prepare three images →
Give each image one job
| Upload order | Reference | What to preserve |
|---|---|---|
| Image 1 | Front-facing portrait | Face, hair, skin tone, and outfit |
| Image 2 | Front-facing product photo | Tube shape, orange-and-white colors, cap, and label placement |
| Image 3 | Corner of a room | Window, table, light direction, and furnishings |
Choose a portrait with a visible face, a complete product photo with the front facing camera, and a room photo with space for the person and product. These are separate references, not three consecutive video frames.
In the Pixocto Seedance 2.5 workbench, choose reference images and upload them in the order above. Seedance 2.5 currently supports up to 30 seconds and 30 reference images; this exercise deliberately uses only three images and 15 seconds, giving the person, package, and room distinct roles. See the model page. Use the references workflow rather than the first-frame workflow.
Connect the references with one action
Place the person beside the reference window and the product on the table. Begin wide enough to see the person and space, follow one natural pickup, and end with the product facing the camera. Stay in the same room with the same person and camera position; avoid a second angle or a location cut.
Set 9:16 and 15 seconds. The English prompt describes the first, second, and third uploaded images by order; Seedance 2.5 receives them as reference assets, so no special image token is needed. If you change the product, replace the tube's shape, colors, and cap description with the real packaging.
Editable prompt
Create a 15-second vertical 9:16 single-take lifestyle product video. Use the three uploaded images as separate reference roles, not as storyboard frames to cut between.
REFERENCE ROLES
The first uploaded image is the character reference: one adult woman in a cream knit top. Keep the same face, hairstyle, skin tone, outfit, and body proportions throughout.
The second uploaded image is the product reference: one orange-and-white sunscreen tube. Preserve its silhouette, color blocks, white cap, proportions, and front-label layout. Do not invent brand text or product claims.
The third uploaded image is the setting reference: a warm apartment window corner with a small table. Keep the window shape, table position, late-afternoon sunlight, and background arrangement consistent.
ONE CONTINUOUS SHOT
00:00–00:04 — Medium shot of the woman beside the window. The tube is standing on the table. Natural handheld framing.
00:04–00:09 — Without a cut, she reaches for the same tube and lifts it. Her hand moves naturally around the package.
00:09–00:13 — She turns the front of the tube toward the camera once. The camera moves slightly closer while her face stays visible behind the product.
00:13–00:15 — Hold the final composition: the tube is sharp and upright in the foreground, the same woman and window remain behind it. End with a clean cut.
AUDIO AND FINISH
Quiet room tone and subtle handling sounds. No spoken claims, music, subtitles, on-screen text, or added logos. Natural skin and fabric texture.
AVOID
No jump cuts, extra people, duplicated tube, changed cap, warped fingers, floating objects, changing room, changing daylight, or newly invented packaging copy.
Upload the three references and make a first pass →
Check continuity before polish
Pause at the pickup and again when the product moves toward the lens. Did the face or outfit change? Is the tube in the hand the same one that stood on the table? Did the window or table move? If any answer is no, simplify the camera move and hand action before adding cuts or effects.
Check packaging text frame by frame before using the result in an actual ad. If the model invents lettering, replace that close-up with a verified product shot during editing rather than treating generated copy as accurate packaging.