# Vidu Q4 for product videos: what the preview can and can't do

> For product marketers and founders planning a launch video: what ShengShu's Vidu Q4 Preview does, what it costs per second, and which shots still need your real product filmed.

Published Oct 9, 2026 · uscale · https://uscale.ai/blog/vidu-q4-product-videos

## Vidu Q4 starts from a picture, not a prompt

Most AI video coverage opens with someone typing a sentence. Vidu Q4 Preview, which ShengShu launched on October 7, 2026, can't start there. Every shot begins with an image (image-to-video) or a set of references (reference-to-video). This release has no text-to-video.

That reads like a dealbreaker. For a product team, it costs less than it sounds, because you already own product shots, a logo and a founder photo. This reading comes from ShengShu's releases and Vidu's own pages, not from a test in uscale. The open question about Vidu Q4 for product videos is which launch shots those pictures can carry, and which they can't.

[Video: Cinematic for Compute Exchange](https://uscale.ai/examples/compute-exchange-cinematic.mp4) — A cinematic cut from our examples: moody opening footage, then the real app filmed. It was not made with Vidu Q4. It shows the mix of b-roll and real product a launch film needs.

## What Vidu Q4 Preview means for product videos

> Vidu Q4 Preview fits teams that already own images of what they sell, because every shot starts from a picture: product stills become launch b-roll, and reference images keep one product or presenter consistent across shots. Nothing in the announcement covers readable interface text, so a software product's real screens still get filmed.

## Vidu Q4 image-to-video and reference-to-video, shot by shot

Q4 Preview has two ways in, and each fits a different launch shot.

Image-to-video takes one image and a text prompt and returns 3 to 16 seconds. Unite.AI reports that the API accepts png, jpeg, jpg or webp up to 50MB and a prompt of up to 20,000 characters.

Reference-to-video takes 1 to 15 image references and 0 to 3 audio references and returns 1 to 16 seconds. ShengShu says references can define characters, wardrobe, props, products and environments. That keeps one product or presenter the same from shot to shot.

Output runs 540p, 720p, 1080p, 2K and 4K, with 10-bit color at 2K and 4K. Unite.AI reports that the output keeps the aspect ratio of the uploaded material, so choose the frame when you choose the image.

| Capability | What Vidu says | The launch shot it fits |
|---|---|---|
| Image-to-video | One image plus a prompt, 3 to 16 seconds | Hypothetical SaaS launch: a product still becomes a slow reveal |
| Reference-to-video | 1 to 15 image and 0 to 3 audio references | Hypothetical: a founder photo and a voice reference keep one presenter across three b-roll shots |
| Output | 540p to 4K, 10-bit color at 2K and 4K | A hero shot that holds up beside high-resolution footage |
| Aspect ratio | Kept from the upload, per Unite.AI | Crop the still to the channel's frame first |

## What a Vidu Q4 clip costs per second, and per shot

Budget each Vidu Q4 shot at the regular rate and treat the discount as a bonus. ShengShu's [GlobeNewswire release](https://www.globenewswire.com/news-release/2026/10/07/3376813/0/en/shengshu-technology-launches-vidu-q4-preview-a-next-generation-ai-video-model-built-for-lifelike-performances.html) of October 7, 2026 lists $0.045 per second at 540P, rising to $0.39 at 4K. [fal's image-to-video page](https://fal.ai/models/fal-ai/vidu/q4/image-to-video) matches those regular rates at 540p and 4K and lists the steps between (October 8, 2026).

![Vidu Q4 image-to-video, price per second on fal after the 30% discount ends November 30: 540p $0.045/s, 720p $0.095/s, 1080p $0.12/s, 2K $0.19/s, 4K $0.39/s](https://uscale.ai/blog-media/vidu-q4-product-videos/chart-vidu-q4-image-to-video-price-per-second--1200x456.svg "Sources: fal.ai / Vidu Q4 (Image to Video) API on fal")

Image-to-video and reference-to-video are 30% off through November 30, which puts 1080p on fal at $0.084 a second. Here is the cost of one maximum-length clip at 1080p:

- Regular: 16 seconds × $0.12 = $1.92
- Discounted: 16 seconds × $0.084 = $1.344

Price the shot you will keep, not the cheapest line in a press release. A launch that slips past November 30 pays the regular rate.

### Figures that don't line up

A separate ShengShu release on [PR Newswire](https://www.prnewswire.com/news-releases/vidu-launches-q4-preview-making-flagship-ai-video-creation-accessible-to-everyone-302901266.html), also dated October 7, 2026, says promotional pricing starts at $0.014 per second and may vary by plan and region. Neither release explains the gap with $0.045. [Vidu's API platform](https://platform.vidu.com/docs/pricing) sells credits at $0.005 each (October 8, 2026), but we found no figure for how many credits a Q4 second uses. ShengShu also claims up to five times the output for the same budget. That is its claim, not a measurement. These rates cost the shots Q4 is built for, not every shot a launch needs.

## Can Vidu Q4 stand in for your real product screens?

Fair question: if reference-to-video can hold a product steady across shots, why not let Vidu Q4 make the demo too? For a physical product or a presenter, the announcement supports that reading.

A software product is different, because the screens are the proof. [ShengShu's release](https://www.globenewswire.com/news-release/2026/10/07/3376813/0/en/shengshu-technology-launches-vidu-q4-preview-a-next-generation-ai-video-model-built-for-lifelike-performances.html) says Q4 Preview is built for short drama production companies, advertising agencies, social content creators and film production teams. Software demos aren't on that list. Nothing in the announcement covers readable interface text or camera control, so we claim neither way. A generated imitation of your dashboard can show buttons that don't exist.

![Real product screens filmed from the live app](https://uscale.ai/blog-media/vidu-q4-product-videos/compute-exchange-dolly-4_0s-1280x720.jpg "Real product screens, filmed from the live app: the part of a launch video the Vidu Q4 announcement makes no claims about.")

Two more limits shape the plan. Q4 Preview has no text-to-video, so someone has to make or own every starting image. And Vidu is [releasing the preview ahead of the full Q4 model](https://www.prnewswire.com/news-releases/vidu-launches-q4-preview-making-flagship-ai-video-creation-accessible-to-everyone-302901266.html), so behaviour and prices may change.

Say you're launching a [developer tool](https://uscale.ai/industries/software-products/developer-tools) from its GitHub repo. The viewer needs to see the real CLI output and the real UI, not a plausible copy. Give Q4 the stills and the presenter, and film the screens.

## Using Vidu Q4 in a product launch video

**Split your shot list in two.** Your quick win: write it in a doc this hour. One column is generated b-roll from images you own: a product still turned into a slow reveal, one presenter kept consistent from a founder photo, an establishing shot of your brand world. The other is filmed product: the screens that prove what the launch claims. Done when every shot sits in one column.

**Film the proof screens.** The screen recorder you already have will do. If you'd rather not record, uscale films your live site or app with no screen recording on your side. It also writes a storyboard with every narration line for your approval before anything is filmed. Vidu Q4 has not been tested in uscale.

![The director in the uscale chat proposing a storyboard, beat by beat, with the narration line for each](https://uscale.ai/blog-media/how-to/chat-storyboard-800x1694.webp "Every narration line is approved before anything is filmed, so generated b-roll and real screens are planned in one place.")

**Price it before you start.** On [uscale's pricing page](https://uscale.ai/pricing), a walkthrough video is about 10 credits. Cinematic cuts with AI film clips cost more, and every price is shown before you press anything.

If you do only one thing, write the two-column list. Agencies can send it to the client for sign-off first; see [videos for marketing agencies](https://uscale.ai/industries/agencies-studios-consultants/marketing-agencies). Success is one launch video on your page that pairs owned-image b-roll with real screens. For more, see [product launch videos](https://uscale.ai/use-cases/product-launch-video) and [video for software products](https://uscale.ai/industries/software-products).

**Try it:** [make a product launch video](https://uscale.ai/use-cases/product-launch-video): real screens filmed from your live site, every narration line approved first.

## Start with the pictures you own

Vidu Q4 starts from a picture, not a prompt. So the first step is gathering the images you already own.

- Collect product stills and a presenter photo.
- Mark which shots must be the real product.
- Cost the generated seconds before the 30% discount ends November 30.

When you're ready to film the real screens, [our guide to turning a URL into a product video](https://uscale.ai/blog/url-to-video) walks through it, and your first video is free. [The example films](https://uscale.ai/examples) show what a finished cut looks like. We'll update this post when Q4 leaves preview, adds text-to-video or changes price. Until then, the shots worth generating are the ones you already have pictures for.

## Questions

### What is Vidu Q4 Preview?

Vidu Q4 Preview is an AI video generation model from ShengShu Technology, launched on October 7, 2026, as an early release ahead of the full Q4 model. It makes clips of up to 16 seconds from a single image and a text prompt, or from up to 15 image references and 3 audio references, with output from 540p up to 4K and 10-bit color at 2K and 4K. For product videos, that means product stills and reference images can become launch b-roll, while a software product's real screens still get filmed.

### Does Vidu Q4 support text-to-video?

No. Vidu Q4 Preview does not support text-to-video. Every clip starts from an image (image-to-video) or from image and audio references (reference-to-video), so a prompt alone is not enough. Nothing in ShengShu's announcements says whether text-to-video will arrive with the full Q4 model. For a product team, product stills, a logo and a founder photo can serve as the starting images.

### How does Vidu Q4 reference-to-video work, and how many references can I use?

Vidu Q4 reference-to-video builds a clip from 1 to 15 image references and 0 to 3 audio references, returning 1 to 16 seconds of video. ShengShu says the references can define characters, wardrobe, props, products and environments, which keeps a subject and its voice consistent from shot to shot. For a launch, a founder photo plus a voice reference can hold one presenter steady across several b-roll shots.

### How much does Vidu Q4 cost per second?

Vidu Q4 Preview's regular rate starts at $0.045 per second at 540p and rises to $0.39 at 4K, with 1080p at $0.12 on fal. Image-to-video and reference-to-video are 30% off through November 30, which puts 1080p at $0.084. A separate ShengShu release cites promotional pricing from $0.014 per second, varying by plan and region, so budget at the regular rate. A 16-second 1080p clip costs $1.92 at the regular rate.

### What does a product launch video with real app screens cost?

Per uscale's pricing page, a product launch video with real app screens, made in uscale, starts with one free first video: 10 credits on sign-up, 20 with a work email, no card needed. After that, a walkthrough video is about 10 credits, and credit packs start at $49 for 50 credits. Cinematic cuts with AI film clips cost more, and every price is shown before you press anything. There is no subscription.

---
Source: https://uscale.ai/blog/vidu-q4-product-videos