Turn a written scene, one still image, or a first and last frame pair into a short clip with synchronized dialogue and effects, using the Vidu Q3 routes this workspace actually runs.
Pick a generation mode and a Q3 tier, then set duration, resolution, aspect ratio, audio and optional off-peak scheduling before you review the point estimate.
This model does not support reference video input
Generated content will appear here
Facts reviewed:
The official model map describes a five-model Q3 series. This page runs one family with three tiers, so the two lists are not the same thing.
The official model map does not list a Pro Fast entry; the official pricing table does, and it lists image-to-video as its only route with 720p and 1080p output. This page follows the pricing table for Pro Fast and does not claim the model map contains it. The mix, drama and ad models are separate official routes that this page does not run. Vidu documents 1 to 16 seconds for the main Q3 routes while this workspace starts at 2 seconds, and Vidu documents movement amplitude and style controls as having no effect on Q3, so neither control appears here.
Vidu Q3 is the third generation of Vidu video models. The official model map groups five models under the Q3 label and describes the series as supporting simultaneous audio and video output, smart scene cuts, and more three-dimensional subject motion. Every model in the series renders at 24 frames per second, and the main routes reach 1080p.
The five models are not interchangeable. viduq3-pro takes text, one image, and a first and last frame pair, and Vidu positions it as the quality tier with strong dialogue and cinematic storytelling. viduq3-turbo runs the same four routes as the fastest option in the series. The remaining three are industry models: viduq3-mix, viduq3-drama and viduq3-ad are reference-to-video routes aimed at comic production and short advertising, and what they generate is driven by reference material rather than a first frame.
This page runs a single Vidu Q3 family with three tiers you can switch between: Pro for quality, Turbo for speed, and Pro Fast for image-to-video work. It does not open the reference-to-video, video extension, lip sync, upscale or template features that Vidu lists for its wider platform, and it does not expose the mix, drama or ad models. The duration range, aspect ratios, audio toggle, off-peak toggle and point estimate shown above are settings of this integration rather than a description of Vidu API billing.
Every item below maps to a control you can actually operate on this page, not to the broader Vidu platform catalogue.
Write the scene, dialogue, camera behaviour, lighting and sound cues in one prompt, then generate a clip from text alone.
Upload one image you are authorized to use and direct how the camera, subject, environment and audio should change from that starting frame.
Supply two visually compatible frames to define both the opening image and the destination of a single continuous transition.
Switch sound generation on to produce dialogue and effects in the same timeline as the picture. Vidu documents the audio parameter as available only on Q3 models.
The Q3 series is documented for scene segmentation, which lets a single prompt carry more than one shot without an explicit cut instruction.
Pro prioritizes quality and dialogue handling, Turbo trades some of that for speed, and Pro Fast is the image-to-video-only route with 720p and 1080p output.
Choose 16:9, 9:16, 3:4, 4:3 or 1:1. Vidu documents 3:4 and 4:3 as supported on Q2 and Q3 models specifically.
Queue the task at roughly half the normal point cost when turnaround is flexible. Vidu commits to generating off-peak tasks within 48 hours and refunds points for tasks that never finish.
Every item below maps to a control you can actually operate on this page, not to the broader Vidu platform catalogue.
Prototype short exchanges, reaction beats, shot changes and room ambience together, which is where synchronized audio saves the most revision rounds.
Animate an approved product still to test camera path, highlight travel, surface texture and commercial pacing before a paid shoot.
Build horizontal, square or vertical concepts from the same idea and compare them side by side for feed, story and pre-roll placements.
Connect two prepared key frames for transformations, time changes, entrances and reveals where the endpoint must be respected exactly.
Use Turbo to compare openings, product beats and vertical ad structures in volume, then re-render the winning direction at a higher tier.
Hold the subject and framing steady while varying light, palette and ambience, then keep the version that reads best before committing to a longer edit.
Each example loads its recommended generation mode and tier, so you can revise it in context rather than rewriting it from scratch.
A Pro text-to-video brief built around staging, spoken lines, reaction and ambient sound.
Medium two-shot in a quiet late-night café. A tired designer slides a folded sketch across the table and says softly, "I think this is the one." Her colleague pauses, smiles, and answers, "Then let us build it." Natural eye lines, restrained hand movement, warm practical lighting, rain against the window, low room ambience, synchronized dialogue, one continuous cinematic shot.
A Turbo prompt for a compact social video with clear motion and a sound cue that lands on the beat.
Vertical close-up of a pastry chef placing the final glossy berry on a small tart. Quick but stable push-in, crisp texture, soft morning window light, a light dusting of sugar falling in slow motion, subtle kitchen ambience with one clean plate tap, synchronized effect timing, no text or logos.
Upload a product still and use the image-only Pro Fast route for tightly controlled motion.
Preserve the exact product shape, material, label placement and background composition from the uploaded image. Add a smooth twenty-degree camera orbit, slow moving highlights across the surface, subtle condensation, and a soft shadow shift. Premium commercial pacing, realistic physics, quiet studio ambience.
A Turbo brief with one fast scene change, a product detail beat, and sound that lands with the cut.
Fast 9:16 sports ad with two clean beats. First: a runner launches from wet pavement at dawn, low tracking camera, strong but realistic stride. Second: match cut to a close-up of the shoe landing, water droplets scattering in slow motion. Cool blue light with one coral accent, synchronized foot impact and breath, no logos or text.
A first-and-last-frame transition focused on spatial continuity rather than subject movement.
Connect the empty studio first frame to the furnished final frame in one continuous transformation. Preserve walls, windows, floor lines, camera angle and perspective. Furniture assembles in a physically believable sequence as daylight warms. Smooth pacing, subtle room ambience, no people, no camera cuts.
A Pro text-to-video prompt that asks for two shots and lets the documented scene segmentation carry the cut.
Night market sequence in two shots with consistent weather and screen direction. Shot one: wide establishing move down a lantern-lit alley, steam rising from food stalls, crowds crossing left to right. Shot two: medium close-up of a cook plating noodles, steam crossing the lens, warm backlight. Keep the same red lantern colour and the same drizzle throughout. Synchronized street ambience and sizzling wok, no captions.
All three tiers are read from the same official pages, so the differences below are directly checkable rather than inferred.
| Characteristic | Pro | Turbo | Pro Fast |
|---|---|---|---|
| Official routes | img2video, text2video, start-end2video | img2video, text2video, start-end2video | img2video |
| Routes here | Text, image and first/last frame | Text, image and first/last frame | Image only |
| Official duration | 1–16 seconds | 1–16 seconds | 1–16 seconds |
| Duration here | 2–16 whole seconds | 2–16 whole seconds | 2–16 whole seconds |
| Official resolution | 540p, 720p, 1080p | 540p, 720p, 1080p | 720p, 1080p |
| Resolution here | 540p, 720p, 1080p | 540p, 720p, 1080p | 720p, 1080p |
| Official rate at 1080P | $0.12 per second | $0.065 per second | $0.125 per second |
| Official rate at 720P | $0.10 per second | $0.055 per second | $0.10 per second |
| Official rate at 540P | $0.045 per second | $0.035 per second | Not offered |
| Off-peak rate | Half of the normal rate | Half of the normal rate | Half of the normal rate |
| Synchronized audio | Supported | Supported | Supported |
| Reference-to-video | Not in the official route list | Supported officially, not run here | Not in the official route list |
Cost on this page is expressed in platform points, and the point estimate updates as you change tier, duration and resolution. Vidu bills its own API per second with credits priced at $0.005 each, which is a separate schedule from the points shown here.
View platform pricingThe workspace shows a point estimate for the exact combination you selected. Changing tier, duration, resolution or the off-peak toggle changes that estimate, so compare settings before submitting rather than assuming an average cost.
Vidu prices its API by second at each resolution tier, and higher tiers cost more per second. Pro Fast, the image-to-video-only tier, has the highest per-second rate in the Q3 line at 1080P. Choosing the resolution that matches the delivery surface is the single largest cost lever on this page after duration.
Start from text, upload one image you are authorized to use, or provide a compatible first and last frame pair.
Pro and Turbo work across text, image and first/last frame modes. Pro Fast is available only when the mode is image-to-video.
Choose a whole number of seconds between 2 and 16, then a resolution that Pro Fast or the main tiers actually support.
Write actions in order and align dialogue, effects and ambience with them. With the audio toggle off you get a silent clip.
If turnaround is flexible, switch on off-peak for a lower point cost, then check the estimate and the selected settings before submitting.
Vidu Q3 can still produce unstable identities, unintended cuts, incorrect dialogue, inconsistent hands or objects, and imperfect audio synchronization. Watch the whole clip at normal speed and then check the first and last two seconds, transitions between shot sizes, and every contact point before you publish or hand it to a client.
Vidu documents the main Q3 routes as ranging from 1 to 16 seconds. This workspace runs from 2 to 16 whole seconds, and its point system, tier availability, resolution pool and off-peak option describe this integration rather than quoting Vidu API billing. Off-peak is a scheduling trade, not a quality setting: Vidu commits to generating off-peak tasks within 48 hours and automatically cancels and refunds tasks that never complete.
Three official Q3 controls have no effect on this model family and therefore do not appear here. Vidu documents movement amplitude, output style and background music as either ineffective or unavailable for Q3, so a prompt that relies on them will not behave the way it does on older Vidu generations. Put camera and motion intent in the prompt instead.
Upload only images you are authorized to use. Do not produce deceptive impersonation, privacy-invasive, or infringing material, and do not present generated footage as documentary evidence. Keep the prompt, source permissions, settings, tier and intended channel alongside the approved output so a later reviewer can reconstruct exactly what was generated.
Model capabilities, routes and rates can change. Official Vidu documentation and Nano Banana Pro platform information are listed separately below. All official pages listed here were reachable without a login when this page was reviewed.
Return to the workspace, choose the tier that matches your mode, and confirm points, resolution and timing before you generate.