A text-to-video model gives you one short clip per request. The prompts that work describe that single shot like a camera brief: who is in it, what moves, where, how the camera behaves, how it is lit, and what it should sound like.
Capabilities reviewed 2026-09-26
Vizify returns one short video per generation, not an edited sequence. Write down the single moment you need, such as a product turning on a plinth or a cyclist crossing a bridge at dusk, and leave the rest of the story for separate clips.
Describe the subject concretely (material, color, age, clothing, brand-neutral details) and give it one motion beat: pours, turns, walks toward camera, lifts off. Several competing actions in a few seconds is the most common reason a clip looks confused.
Add the setting, time of day, weather, and light direction. "Soft window light from the left, late morning" gives the model far more to work with than "nice lighting", and it keeps color and shadows consistent across the clip.
State the framing and the movement separately: a medium close-up with a slow push-in, a locked-off wide shot, a handheld follow at walking pace, a slow orbit. If the camera should not move, say it is locked off.
Name the visual treatment (natural documentary color, 35mm film grain, clean studio product look) and then say what the audio should contain, or ask for a silent clip. Audio support depends on the model, so check it before you rely on sound.
Duration, resolution, aspect ratio, reference images, and audio are set per model in Vizify. Kling 3, the current default, runs 3 to 15 seconds with audio off unless you turn it on; Veo 3.1 is a fixed 8 seconds with audio always on; Seedance 2 is always silent.
Watch the full clip, not one frame. Faces, hands, logos, and small text drift first. When you revise, change one element of the brief at a time so you can tell which change fixed the problem.
Slow 180-degree orbit around a matte white ceramic pour-over coffee set on a dark walnut counter. Soft morning window light from the left, gentle steam rising from the cup, shallow depth of field. Clean studio product look, no text on screen. Silent clip, 16:9.Why it works: One subject, one camera move, one light source, and an explicit request for no audio or on-screen text, so nothing competes with the product.
Night market street food stall: a vendor flips skewers over a charcoal grill as sparks lift into the air. Handheld medium shot at chest height, slight natural sway. Warm tungsten stall lights against a blue dusk sky. Audio: sizzling grill, low crowd murmur, no music, no dialogue.Why it works: The audio line names the sounds you want and the ones you do not. Use it with a model that generates audio, such as Veo 3.1 or Kling 3 with audio turned on.
Vertical 9:16. A trail runner in a red jacket crests a rocky ridge toward the camera at sunrise, breath visible in cold air. Low-angle tracking shot that slows as she reaches the top. Crisp natural color, light lens flare, 8 seconds.Why it works: The aspect ratio, the direction of movement relative to the camera, and the duration are all stated, which keeps the subject framed for a phone screen.
Use the attached photo as the reference. Keep the bottle shape, label artwork, and color exactly as shown. The bottle rotates slowly on a turntable while condensation beads on the glass; locked-off camera, dark gradient background, soft rim light.Why it works: It tells the model what must not change before describing the motion. Upload a PNG, JPEG, or WebP image in Vizify; a video Vizify generated earlier has to be downloaded and uploaded again to be used as a reference.
Wide coastal grassland under a moving storm front. Locked-off camera on a tripod, long grass rippling in strong wind, cloud shadows sweeping across the hills. Muted desaturated color, overcast diffuse light. 10 seconds.Why it works: With no character to keep consistent, the prompt spends its detail on motion in the environment and on a static camera, which gives stable, usable footage.
Text-to-video models are trained to render a continuous shot, and Vizify’s video capability returns exactly one clip per request. A prompt that describes a sequence (“she opens the door, walks to the window, then turns and smiles”) asks the model to compress three shots into a few seconds, and it will usually blur them together. Write the shot you need now, generate it, and write the next shot as its own prompt.
A reliable order for the brief is subject, action, setting, camera, light, style, audio. You do not need headings or a template, but covering each of those in plain sentences removes most of the guessing the model would otherwise do.
Because every Vizify video model exposes its own parameter set, the same brief behaves differently depending on where it runs:
If a detail in your prompt conflicts with the model’s limits, such as a 20-second request to a model capped at 15 or “silent” on a model that always returns audio, the model’s contract wins. Check the model page before you write the timing and audio lines.
Watch the clip at full size from start to finish. The first things to drift are faces, hands, product geometry, logos, and any small text. When a revision is needed, keep everything else in the prompt the same and change one line, whether that is the camera move, the light, or the action, so you learn what actually made the difference. Each revision is a new generation and uses credits the same way the first one did.
Long enough to cover subject, action, setting, camera, light, style, and audio, which is usually two to five sentences. Stacking extra adjectives beyond that tends to add competing instructions rather than more control.
It depends on the model. Veo 3.1 and Grok Imagine Video 1.5 always include audio, Kling 3 and Seedance 2.5 let you turn audio on or off, and Seedance 2, Wan 2.7, Runway, and Hailuo 2.3 are silent. Check the model page before you rely on sound.
Each model has its own range in Vizify. Kling 3 runs 3 to 15 seconds, Seedance 2.5 and Wan 3.0 run up to 30 seconds, and Veo 3.1 is fixed at 8 seconds. For anything longer, generate separate clips and assemble them in your own editor.
Yes. Name the shot size and the movement separately, for example a medium shot with a slow push-in, or a locked-off wide shot. If the camera should stay still, say so; otherwise the model may add movement you did not ask for.
Vizify does not expose a separate negative-prompt field for video. Put exclusions in the brief itself, such as no on-screen text, no music, or no people in frame, and keep them short.
Yes, with a model that accepts reference images. Upload the image with your prompt and say what must stay unchanged. The image-to-video tools on the Site prepare that request for you.
Yes. Video generation requires Vizify Pro or a credit pack. You can still prepare and refine the prompt on the Site before you sign in.
Motion models can drift in identity and fine detail, especially with several actions or a moving camera. Reduce the clip to one action, keep the camera simpler, and add a reference image when identity matters.
Prepare one text-led clip and hand it to Vizify with your chosen framing.
OpenStart from a still image when the subject has to stay recognizable.
OpenUse this when the clip must come back with native sound.
OpenCompare duration, resolution, audio, and reference limits for every model.
OpenSee how the default model and a fixed-length audio model differ.
Open