Pick a photo that is easy to animate
Use one product, in focus, on a plain background with some room around it.
Crop out stray text, logos and watermarks first. Faint text in a photo can come back in the video.
Add it in the Studio
Open the Studio, add the photo to the prompt box (tap + Image, or drop or paste it there), then write the prompt. The video starts from that photo.
Photo to video works in Senfers 1.0, Kling 3.0, Seedance 2.5 and Wan 3.0. MiniMax H3 takes text only.
Describe only the motion
The photo already shows the product, so do not describe it again. Write what changes: a hand enters, the product turns, the camera pushes in.
Add one line about what stays the same: the shape, the colour, the label.
The product in the photo stays in the same place, with the same shape, colour and label. The camera pushes in slowly. A hand enters from the right, lifts the product, turns it once, and sets it back in the same place.
Pick the format for where it will run
Vertical (9:16) for TikTok and Reels, wide (16:9) for YouTube. Each model page lists its aspect ratios, so check it before you pick a model.
Check the result against your photo
Compare the shape, the colour and the label with the photo. Text on a pack can come out soft or changed. If the text must be readable, add it in your editor afterwards.
If the product changed, try again with less motion: one action and a slower camera.
Make a few versions
Ads are tested, not guessed. Make two or three versions with a different hand, room or camera move, and keep the one that works best.