Vidu Q3 Image to Video Prompting Guide

Vidu Q3 Image to Video is Vidu's image-to-video model on Versely. This page is its structured prompting reference: the 5 parameters its schema actually exposes, the i2v technique that applies to it, copy-ready templates, and real prompts from production workflows that shipped on it.

Everything here is grounded in the same sources Versely's agent reads — the model's input schema, and prompts quoted verbatim from shipped workflows. Where a line is general craft advice rather than a documented fact about Vidu Q3 Image to Video, the page says so.

What Vidu Q3 Image to Video wants

The exact input surface, from the same schema the Versely agent fetches with get_model_input_schema before every generation.

Technique that applies here

Image-to-video: animating a start frame; motion description relative to the input image; UGC hook pool applies

• Aspect ratio usually isn't a settable parameter in image-to-video mode — it's inherited from your source image. Vidu Q3's runpod schema note is explicit: 'No style/aspect_ratio fields in I2V — aspect ratio is inferred from the input image.' Pixverse 5.6 I2V's note says the same thing: 'I2V infers aspect_ratio from image (no aspect_ratio field).' Wan's i2v schemas (2.7, and the folded Wan 2.5 variants) all omit an aspect_ratio param entirely; two of its provider variants confirm this directly in their own schema notes — kie's says 'inferred from image', replicate's says 'aspect is derived from the input image.' Crop or compose your source image to the aspect you want before you upload it; asking for '16:9 widescreen' in the prompt text does nothing on these schemas.
• The prompt describes motion relative to what the image already shows, not the scene itself. Vidu's own param description for its i2v prompt field is literally 'Text description of the desired motion and action' — contrast a text-to-video prompt, which has to establish the whole scene from nothing. General craft, not a schema fact: name what changes — a hand reaches, a head turns, hair moves in wind — rather than re-describing the subject or setting the photo has already fixed, which just competes with the image for the model's attention instead of directing it.
• Several i2v schemas quietly accept a second, optional 'end' image alongside the required start image — Vidu Q3's fal variant (end_image_url, described as generating a transition), Kling Video V3 Standard Image to Video's fal variant (end_image_url), LTX 2.3 Image to Video Pro (a model folded into the LTX 2 Pro page; end_image_url), and Wan 2.7 Image to Video (last_frame_url). If you supply one, write the prompt as the path between the two states, not as 'what happens next' from a single photo — you're now effectively writing a first-last-frame prompt inside an i2v call.

Copy-ready templates

Replace the bracketed slots; each template says when it's the right shape.

[WHAT MOVES — e.g. 'she turns her head toward camera and smiles'], [SECONDARY MOTION DETAIL — e.g. 'hair drifts in a light breeze']. Camera: [static / slow push-in / slight handheld].

[STATE A DESCRIPTION] transitions into [STATE B DESCRIPTION] as [WHAT CHANGES IN BETWEEN].

Real prompts that shipped on Vidu Q3 Image to Video

Quoted verbatim from Versely's production workflow library — each one generated a scene in a shipped workflow.

A man sitting at his desk during the day, working on his laptop, quietly struggling with a blocked nose. He looks at the screen and types calmly, then tries to take a breath in through his nose — it's blocked, so no air passes. His brow furrows, he pauses typing, parts his lips and breathes in through his mouth instead, gives a small uncomfortable sniff and goes back to the screen, a little bothered. The faint red glow over his nostrils pulses gently each time he tries and fails to breathe through his nose. Subtle realistic chest movement, small natural head movements. Camera: slow steady push-in toward his face. Cool grey-blue daytime light, calm and ordinary, a quiet hint of discomfort. Avoid: no tissues, no extra people, no red nose, no fast motion, no face morphing, keep him the same person (Mark), the red glow only pulses faintly.

The HIVEWELL honey jar character looks at the camera with a playfully frustrated expression, gives a tiny head shake on "Please stop using me," gestures toward the steaming tea mug beside her with one hand. On "I'm actual medicine," her expression shifts to confident and slightly serious. Mouth perfectly lip-synced. No overacting. Speaks in one smooth continuous flow with zero pauses, zero breath gaps, zero sighs, zero hesitations between or within sentences: "Please stop using me just to sweeten your tea. I'm actual medicine." Voice: warm friendly American female, late 20s, mid-pitched, casual with a playful sassy tone and smile in voice. Completely human, real, natural — not AI, not a voice actor. Ambient: dead silent. No music, no song, no tune, no melody, no ambient sound, no kitchen sounds. Only her voice. Background: steam rises gently from the tea mug in visual silence, warm sunlight glows softly through the window. Static medium shot. Warm amber golden hour lighting. Pixar-style 3D, cinematic.

Camera slowly pushes in toward the man hunched at his desk. He keeps typing slowly, shoulders slumped, head tilted forward toward the laptop. Subtle tired breathing, a slow blink, one small weary shift of his neck. Warm lamp light flickers gently, soft screen glow on his face. Calm, quiet, slightly sad mood. Static cozy room, slow cinematic motion. 4 seconds.

Mistakes that waste generations

• Asking for a specific aspect ratio in the prompt when the schema doesn't expose an aspect_ratio param — it's silently ignored; on Vidu, Pixverse, and Wan's i2v variants the frame shape comes entirely from the source image you upload.
• Re-describing the subject or scene the source image already shows instead of focusing on the motion — wastes prompt budget and can conflict with the photo (a different outfit or background than what's actually in frame).
• Treating every 'image-to-video' model as the same shape: sending a HeyGen-style voice/talking_style prompt to Vidu, or a plain motion-description prompt to HeyGen, targets a control surface that model doesn't expose.

Frequently asked questions

Related prompting guides