Guides

    Pressure cookers: steam is Vidu; the panel type is overlay

    Do not Veo the buttons.

    Versely Team4 min read

    Do not Veo the buttons. Steam is atmosphere on a pot you already photographed. The panel is type you already know: minutes, pressure, a mode name. A cinematic generate that restypes those digits is a cousin of the cooker with a panel you cannot ship.

    Vidu Q3 Image to Video is the steam pass: image required, native audio on, 1080p ceiling, durations only 5s, 10s, and 15s. Listed at 16 credits and marked discounted. Features: image-to-video, motion generation, high fidelity. Add a text overlay is the panel pass: one authored line, burned in, not invented by a video model.

    Steam is a motion pass on a locked pot

    Photograph the real cooker: lid seated, valve where it lives, housing colour the SKU actually is. Then ask Vidu for steam, a lid wiggle, a slow push. Prompt the motion only. Re-describing the pot is how the model treats the pot as optional.

    Q3 is image-to-video only. A paragraph about "luxury cooker, chrome, golden hour" is the wrong family. Lock the still first. Image-to-video is the door.

    You pick 5, 10, or 15. There is no 6s, no 8s, no 12s. A 6-second steam idea is a 5-second clip that gets cut or a 10-second clip that pads. Sixteen credits is the listed figure; billing still depends on duration and resolution. Discounted is not a reason to start at 15s. A cheap miss still arrives with a soundtrack, because audio is true.

    Write the mix in the 5s prompt if you want room tone and steam hiss. A silent brief still gets a score. If a host must speak, this remains image-to-video, not lipsync. There is no aspect-ratio list on this slug. Crop the still to 9:16 or 16:9 before upload.

    Buttons are a typesetting job

    Video models render panel text as visual pattern. They garble digits. A "12" that becomes "18" is a cook time. A PSI that swims is a claim. Do not bake the panel into the generate.

    Generate the pot without load-bearing type. Then overlay the line you control. add_video_captions burns one fixed string for the whole clip — top, center, or bottom. That tool does not listen. If you needed a transcript of a host, you opened the wrong door. If minutes swap at 0s, 2s, and 4s, use timed overlays. The persistent badge — "6 minutes," "high pressure" — is the fixed overlay.

    Lock the picture. Then print the line. Cook times change. Overlaying a sample is typesetting you will throw away.

    Q3's three lengths, not a Veo panel

    Veo is a cinematic short-clip family with native audio and a 4K tier. It is a good row for a scene. It is a bad row for a control panel. Asking Veo to "keep the buttons exact" is how you pay for a soundtrack under digits the model invented. Do not Veo the buttons. Overlay them.

    requires_image is true. High fidelity is a promise about the plate, not a licence to skip it. Iterate still-plus-steam at a lower listed quality if identity is still in argument; spend 1080p on the 5s take you will post. Qualities include 360p, 540p, 720p, and 1080p. The ceiling is 1080p, not 4K. Caption after. Five seconds of native audio is not a subtitle file.

    Burn the time after the steam holds

    Would you print the still at 1080p? If no, stop. Steam will not rescue a melted lid. If yes, write a five-second steam action and a mix note. Generate 5s. Only go to 10s or 15s if the action is actually that long. Then overlay the one line you would put on the box.

    If you needed 3s or 8s, pick a different image-to-video row. Do not pad Q3 to fake 8s, and do not Veo a panel to get 4K digits. The SKU is the still. The minutes are type.

    FAQ

    Can I generate a 3-second Vidu Q3 steam clip?

    Not on this ladder. Durations are 5s, 10s, and 15s. The shortest test is five seconds. If you need a three-second hit, pick a model that lists 3s.

    Should the cook time be in the video prompt?

    No. Models treat panel digits as texture. Type the real minutes as an overlay after picture lock. Generate a clean pot.

    Does Vidu Q3 require an image?

    Yes. Image-to-video, image required. "Transform static images into dynamic videos" is the description. A text-only generate is a mislabel.

    Can the overlay change mid-clip as the timer counts down?

    Not on the fixed overlay. One string, whole file. Timed lines are add_timestamped_captions. Either way, you write the digits. The model does not.