vidu.com/vidu-stream · Read 2026-11-01 · Not this button
Vidu S2: realtime avatar and editing, and the clip this site prices
Vidu S2 is Shengshu Technology's realtime model, described at vidu.com/vidu-stream. The form on this page runs Kling 2.6, which Nano AI has priced. It is not Vidu S2, and it is not the Vidu Q4 clip on /vidu-q4.
This generator runs Kling 2.6 for 55 credits a 5-second clip, with sound: false. It is Nano AI's priced model, not Vidu S2, and it does not call the Vidu S2 API.
Duration
Ratio
Generated Videos
No videos yet
Fill out the form and click Generate Video to create your first AI video.
Tips: Use detailed descriptions for better results
- Made by
- Shengshu Technology
- Run on Nano AI
- Kling 2.6 · from 55 credits per video
- Page updated
- November 1, 2026
Source: vidu.com
What Vidu S2 refers to
The page at vidu.com/vidu-stream, read on 2026-11-01, splits Vidu S2 into Vidu S2-Avatar and Vidu S2-Editing. It says Vidu S2-Avatar supports realtime voice, complex motion such as dancing, and reference images for objects, outfits, and backgrounds, including an offline mode that reads a script or a preset action. It says avatar output is 720p or higher. It says Vidu S2-Editing edits an incoming stream from style, outfit, subject, or background references, with camera, video, or image input. A FAQ answer says this is an upgrade from Vidu S1's 540p, and that voice instructions can drive dance. The same page points API readers to platform.vidu.com/vidu-stream/doc and lists vidus@shengshu.ai. This site did not open an account there.
The paper at arxiv.org/abs/2609.11638, read the same day, uses the same split and lists Tsinghua University and Shengshu Technology. It says Vidu S2-Avatar raises realtime generation from 540p to 720p while keeping 25~42 FPS, and that a reference image can change during a stream. It says Vidu S2-Editing can restyle a stream, change clothes, do character replacement, or replace a background. Those are the authors' claims. This page does not repeat their benchmark scores, and the form here does not replace a person in a clip you upload.
What this site prices is separate. /vidu-q4 runs Vidu Q4 as image-to-video or reference-to-video, 90 credits for 5 seconds, 180 for 10, and 270 for 15. That is not Vidu S2. /seedance-2-5 runs Seedance 2 at 720p without audio, from 150 credits for 5 seconds. The button on this page is Kling 2.6: 55 credits for 5 seconds, 110 for 10, and the request sets sound: false. Eligible new accounts get 64 credits, which covers one 5-second Kling clip and not a Vidu Q4 clip at 90. None of these is a realtime avatar session.
Vidu S2 at a glance
- Product page
- vidu.com/vidu-stream, read 2026-11-01
- Paper
- arxiv.org/abs/2609.11638, Tsinghua University and Shengshu
- Two modes they name
- Vidu S2-Avatar and Vidu S2-Editing
- Avatar, on the product page
- Voice, dance, reference objects, outfits, backgrounds, offline script
- Editing, on the product page
- Style, outfit, subject, background; camera, video, or image input
- Resolution they contrast
- 720p on S2, against 540p on Vidu S1
- Frame rate in the paper
- 25~42 FPS while moving from 540p to 720p
- API link they print
- platform.vidu.com/vidu-stream/doc
- Vidu Q4 on this site
- Image or reference to video, 90 / 180 / 270 credits. Not Vidu S2
- Seedance 2 on this site
- 720p, no audio, from 150 credits. Not Vidu S2
- Generator on this page
- Kling 2.6, 5 seconds 55 credits, 10 seconds 110, sound: false
- New account here
- 64 credits, one 5-second Kling clip, not a 90-credit Vidu Q4 clip
How to read a Vidu S2 claim
Avatar and editing are two modes
The product page separates Vidu S2-Avatar from Vidu S2-Editing. This form does not open a livestream, a camera input, or a realtime edit.
720p is their upgrade over 540p
The FAQ says S2 moves avatar quality from Vidu S1's 540p to 720p. The paper adds 25~42 FPS. This site has not measured either number.
Character replacement stays on their page
The paper lists character replacement as an editing result. The button here makes a new clip of an object or an empty place. It does not replace a person in footage.
Vidu Q4 is a different product on this site
/vidu-q4 is Vidu Q4 at 90 credits for 5 seconds. This page's Kling 2.6 clip is 55 credits and sets sound: false.
Vidu S2 and the video models this site prices
The first data column is vidu.com/vidu-stream and the paper, read on 2026-11-01. The other columns are models Nano AI has priced. This page has not called the Vidu S2 API.
| Vidu S2, read 2026-11-01 | Kling 2.6 here | Vidu Q4 here | |
|---|---|---|---|
| What starts the clip | Voice, a reference image, or an incoming stream | Text, or one image on the image tab | An image, or reference images |
| Realtime | The product page says yes, for avatar and editing | No | No |
| Resolution they state | 720p, up from 540p on Vidu S1 | Not sold as 720p here | 720p with audio, on this site's Vidu Q4 page |
| Frame rate | 25~42 FPS in the paper | Not stated | Not stated as S2 |
| Credits for 5 seconds | Not a Nano AI price | 55, and 110 for 10 seconds | 90, then 180 and 270 |
| Audio | Voice interaction on Vidu S2-Avatar | sound: false | The Vidu Q4 page prices audio on |
| Person replacement | Listed for Vidu S2-Editing | Not offered | Not offered |
How to try the priced Kling 2.6 clip
- 1
Start from a prompt on this page
Pick one of the four prompts. Each one is an object or an empty place. Do not upload a clip in order to replace the person in it.
- 2
Leave Kling 2.6 and 5 seconds selected
The form opens on text to video. A 5-second clip is 55 credits and a 10-second clip is 110. The request sets sound: false. It is not Vidu S2-Avatar and it is not Vidu Q4.
- 3
Generate and keep the file
Sign in. The button shows the cost before you start. Eligible new accounts get 64 credits, which covers one 5-second clip. The file is a Kling 2.6 video from this site, not a Vidu S2 session.
Prompts for the Kling 2.6 endpoint
Original prompts for a 5-second Kling 2.6 clip. They do not name a person or ask to replace one.
Paper kite
A paper kite lifts off an empty field. Slow rise, no people, no text, no logos.
Kettle
Steam rises from a kettle on an empty stove. Static camera, no people, no text, no logos.
Empty swing
An empty swing moves once in a light wind. Side view, no people, no text, no logos.
Candle
A single candle flame in a dark empty room. Slow push in, no people, no text, no logos.
FAQ
Vidu S2 FAQ
What vidu.com/vidu-stream and the paper said on 2026-11-01, and what this button calls.
Does this generator run Vidu S2?
No. It runs Kling 2.6 for 55 credits a 5-second clip, and the request sets sound: false. It does not call the Vidu S2 API.
What is the difference between Vidu S2-Avatar and Vidu S2-Editing?
The product page says Avatar is a realtime voice character, including dance and reference images, and Editing changes an incoming stream's style, outfit, subject, or background. This form does neither.
Is Vidu Q4 the same as Vidu S2?
No. /vidu-q4 is Vidu Q4, image-to-video or reference-to-video, at 90 credits for 5 seconds. Vidu S2, on Shengshu's page, is the realtime model. This page's button is Kling 2.6.
Does the paper's character replacement run here?
No. The paper lists character replacement as something Vidu S2-Editing can do. This site does not offer that. The prompts are objects and empty places.
Can 64 credits buy a Vidu S2 session?
This site does not sell a Vidu S2 session. 64 credits cover one 5-second Kling 2.6 clip at 55 credits. They do not cover Vidu Q4 at 90 credits, or Seedance 2 at 150.
Is Nano AI affiliated with Shengshu or Vidu?
No. Nano AI is not affiliated with Shengshu Technology or Vidu. Vidu S2 on this page is a reading of their public page and paper. The generator offers Kling 2.6, which this site has priced.
More from Nano AI
Related pages
Nano AI is an independent service and is not affiliated with Shengshu Technology or Vidu. Details are from vidu.com/vidu-stream and arxiv.org/abs/2609.11638 as read on 2026-11-01. The generator runs Kling 2.6 unless you pick another model this site sells. It sets sound: false, does not call the Vidu S2 API, and does not replace a person in a clip.