Run text, image and first/last frame tasks on the Pro tier of Vidu Q3, and see the official per-second rates next to the controls this route actually exposes.
Vidu Q3 Pro loads as a single tier: text, image or first/last frame input, 2 to 15 whole seconds, and 720p or 1080p output. Check the point estimate before submitting.
This model does not support reference video input
Generated content will appear here
Facts reviewed:
The official Q3 series contains five models. This page opens one of them as a dedicated route rather than a tier selector, so the two lists describe different things.
The official model map and the official pricing table are two separate documents, and this page follows both without merging them. The map lists viduq3-pro with image-to-video, first/last frame and text-to-video; the pricing table lists the same three routes with per-second credit costs. What neither document describes is this page: a single-tier entry point that fixes the tier and exposes three of the four routes Vidu documents for the Q3 line. Reference-to-video, the 540p resolution tier, the audio parameter and off-peak scheduling are all documented by Vidu and all absent from this route. That gap is stated rather than glossed: if you need any of those four, this is not the page to generate from.
Vidu Q3 is the third generation of Vidu video models, and the official model map groups five models under that label. Vidu Q3 Pro is the tier this page opens: the one Vidu routes through image-to-video, text-to-video and start-end2video, and the one the official pricing table lists at the highest per-second credit cost of the three consumer-facing Q3 tiers at 1080p.
That cost difference is the whole point of this page. Pro and Turbo are documented with an identical capability matrix on the official model map: both list image-to-video, first/last frame and text-to-video as available, both render at 24 frames per second, both reach 1080p, and both run from 1 to 16 seconds. What separates them is price. On the same official pricing table, the 1080p route costs 24 credits per second on Pro and 13 on Turbo, a gap of roughly 85 percent. Nothing in the documents this page relies on says that Pro produces better output than Turbo; what the table does say, precisely and checkably, is that Pro costs more per second at every resolution.
So the honest framing for this page is a cost question rather than a quality claim. Pro is the tier you pick when you have decided the extra per-second spend is worth it for the task in front of you, and Turbo is the tier you pick when it is not. The comparison section below puts both columns side by side using the same official rows, so you can make that decision on numbers rather than on marketing language. Everything this page states about limits, controls and unavailable routes is verified against the official documentation listed at the bottom, and anything the documents leave ambiguous is left out rather than filled in.
Each item below maps to a control or a documented route that applies to this specific tier, not to the wider Vidu platform catalogue.
Write subject, action, camera behaviour, lighting and timing into a single prompt and generate the clip from text alone. Vidu documents viduq3-pro with text2video on the official pricing table, so this is a route the tier is priced for rather than one inferred from the series.
Upload one image you are authorized to use, then direct how the subject, camera and environment should move away from that frame. This is the mode where the tier distinction matters least, because Pro, Turbo and Pro Fast all support it.
Supply two visually compatible images to define both the opening state and the destination of one continuous transition. The official model map lists first/last frame generation for viduq3-pro, and the pricing table prices it in the same start-end2video row as the other two routes.
Choose 16:9, 9:16, 3:4, 4:3 or 1:1. Vidu documents 3:4 and 4:3 as supported specifically on Q2 and Q3 models, which makes the two less common ratios legitimately available here rather than borrowed from an older generation.
Set any whole number of seconds from 2 to 15. The official floor is 1 second and the official ceiling is 16, so this route is narrower at both ends, and the page states that plainly instead of repeating the official range as if it were a local capability.
Pick the delivery surface first, then the resolution that matches it. The official table prices 1080p at 24 credits per second against 20 for 720p, which makes this the largest single cost lever on the page after total duration.
The workspace recalculates an estimate for the exact combination you selected, so changing duration or resolution updates the number before you commit. Treat it as the only current price signal for this route.
The specification block above separates what Vidu documents from what this route runs. That split is deliberate: it keeps an official capability from being read as a local feature, which is the most common way a model page misleads a buyer.
Each item below maps to a control or a documented route that applies to this specific tier, not to the wider Vidu platform catalogue.
The Q3 series is documented for synchronised audio and video output in one pass, which makes a two-person exchange with visible timing a natural fit. Because audio is a series-level capability rather than a Pro-exclusive one, test the same prompt on a cheaper tier before committing a long sequence to this one.
When one continuous movement has to start and end on specific images, the first/last frame route removes a variable that text prompts cannot pin down. It is the most predictable of the three routes on this page and the least sensitive to prompt wording.
Five to eight second spots are short enough that per-second pricing matters less than getting the framing right on the first or second attempt. A high-stakes short clip is the clearest case for paying the Pro rate rather than iterating on a cheaper route.
Repeated short renders from the same starting image let you compare identity drift frame by frame. Running that comparison at the same resolution and duration on each attempt keeps the only variable the thing you are actually testing.
Render a two or three second version first, confirm the composition and motion read the way you intended, and only then extend the duration. Extending an approved short clip costs far less than discarding a long one that missed the brief.
Vidu documents the movement amplitude parameter as having no effect on Q2 and Q3 models, so camera intent has to live in the prompt on this tier. That makes planning the shot language in writing a required step rather than an optional one.
Each starter lists the input route it assumes, and every prompt is written to work without any control that this tier does not expose.
A dialogue scene that tests how the tier handles two subjects, turn-taking and ambient sound in one continuous take.
Two colleagues stand at a rooftop railing at blue hour. The person on the left speaks first, gesturing toward the skyline; the person on the right listens, then answers with a short nod. Camera holds a medium two-shot for the whole clip, with a very slow push in. City ambience runs underneath the dialogue. Keep faces consistent, keep hands still, no text or logos.
An image-to-video test that measures how faithfully a supplied product frame is preserved while the camera moves.
Starting from the provided image, rotate the camera slowly around the product by about 30 degrees, keeping the label facing the lens and the background pattern unchanged. Soft studio light stays fixed. Movement should be even and continuous, with no change to the product shape, colour or proportions.
A first/last frame task that uses two stills to define both ends of one continuous camera move.
Start on the first frame and end on the second. Move the camera forward along the corridor at a steady walking pace, letting the ceiling lights pass overhead in rhythm. Keep the geometry of the walls straight and the floor reflections consistent between the two endpoints.
A close-up built to check lip movement and audio alignment separately from the wider composition.
Close-up of a single subject facing the lens, speaking one short sentence. Mouth movement should match the words; shoulders stay still. Background is soft and out of focus. Light comes from the left at a low angle. Hold the framing steady for the entire clip with no cut.
A scene where the only intended change is environmental, which makes it a clean test of how much motion the tier adds on its own.
A wide shot of a stone bridge over a river stays fixed. Over the clip, rain intensifies from a light drizzle to a steady fall, ripples spread across the water surface, and mist builds in the far trees. No camera movement. No new objects enter the frame.
An image-to-video task designed to expose unwanted changes in the subject while the camera tracks.
Using the provided image as the opening frame, have the character walk forward along the path at a calm pace while the camera tracks backward at the same speed. Clothing, hair length and facial features must stay unchanged throughout. Background depth should expand naturally as the character advances.
Every row below is read from the same official model map and the same official pricing table, so the two columns are directly comparable rather than assembled from different sources.
| Item | Vidu Q3 Pro | Vidu Q3 Turbo |
|---|---|---|
| Official route list | img2video, text2video, start-end2video | img2video, text2video, start-end2video |
| Official resolution | 540p, 720p, 1080p | 540p, 720p, 1080p |
| Official duration | 1 to 16 seconds | 1 to 16 seconds |
| Official frame rate | 24 fps | 24 fps |
| Resolution on this page | 720p, 1080p | 720p, 1080p |
| Duration on this page | 2 to 15 whole seconds | 2 to 15 whole seconds |
| Official rate at 1080P | 24 credits/s, $0.12 | 13 credits/s, $0.065 |
| Official rate at 720P | 20 credits/s, $0.10 | 11 credits/s, $0.055 |
| Official rate at 540P | 9 credits/s, $0.045 | 7 credits/s, $0.035 |
| Off-peak rate at 1080P | 12 credits/s, $0.06 | 7 credits/s, $0.035 |
| Reference-to-video | Not in the official route list | Listed officially, not opened on this platform |
| Difference at 1080P | Defines the higher rate | 47 percent lower than Pro |
Vidu bills its own API per second, with credits priced at $0.005 each on the official pricing page. The point estimate in the workspace is a separate schedule for this platform, so use the official figures below to understand the tier gap and the workspace estimate to understand what this task will cost you here.
View platform pricingThe estimate recalculates for the exact combination you selected. Duration and resolution both move it, and because the official Pro rate is roughly 85 percent above the Turbo rate at 1080p, switching tier is the largest single change available. Compare the estimate across settings before submitting rather than assuming an average.
Per-second rates on the official table step up at every resolution tier: 9 credits per second at 540p, 20 at 720p and 24 at 1080p on Pro. Because cost scales with both resolution and duration, a 1080p clip of twice the length costs roughly four times a short low-resolution test. That is the argument for validating composition at the lower setting first.
Compare the official Pro and Turbo rates for your resolution before you start. If the answer is that the cheaper tier is good enough, that decision costs you nothing to make now and a full render to make later.
Text, one image, or a compatible first and last frame pair. The first/last frame route is the most deterministic of the three, so prefer it whenever the shot has to land on a specific final image.
Whole seconds from 2 to 15, then 720p or 1080p. Remember that 540p exists on the official table but not on this route, and that the official 16 second ceiling is not reachable here.
Describe subject, action, shot size, camera direction, lighting and pacing in that order. Camera movement has to be stated in words on this tier, because there is no movement control to fall back on.
Check the point estimate and every selected setting before submitting. Once the task starts, the settings are the contract: nothing in the workspace changes them retroactively.
Output is not deterministic. Vidu Q3 video can still produce unstable identities across shots, unintended cuts, mismatched dialogue, inconsistent hands or props, and imperfect audio synchronisation. Watch the whole clip at normal speed before judging it, then re-watch the first and last two seconds and every transition between shot sizes, because that is where the failures concentrate.
This page runs a narrower range than the official documents describe in four separate ways, and each of them is a real limitation rather than a wording choice. The resolution pool omits 540p. Duration starts at 2 seconds and stops at 15. The audio parameter that Vidu documents for Q3 models has no control here. Off-peak scheduling, which the official pricing table prices at half rate, is not available on this route. If a project depends on any one of those four, generate from a route that exposes it instead of planning around this page.
The per-second figures on this page describe how Vidu prices its own API. They are not this platform's point prices, and they are not a quotation. Vidu can change its table without notice, and this page does not repeat any promotional or discount offer because the official Q3 section of the pricing page contains none. Verify the current table before you use any of these numbers in a budget.
Upload only images you are authorized to use. Do not generate deceptive impersonation, privacy-invasive content, or material that infringes someone else's rights, and do not present generated footage as documentary evidence. Keep the prompt, the source permissions, the selected settings and the intended channel together with the approved output so that a reviewer can reconstruct exactly what was produced and why it was considered acceptable.
Capabilities, routes and rates can change. Official Vidu documentation and platform information are listed separately below. Every official page listed here was reachable without a login when this page was last reviewed.
Return to the workspace, confirm the duration, resolution and point estimate, and decide whether this task justifies the Pro rate before you generate.