This editorial demo used one image of two fictional adults as character references, Standard's Seedance 2.0 Mini model, an 8-second request and a vertical 720p frame. The published MP4 contains video and generated audio. An exact lyric request did not guarantee exact wording.
Updated
The image that guided the duo
This is the actual joint reference recorded for the render. Both performers are fictional adults, so the input and output can be published together without exposing a customer’s photos.
The image already has the orange stage. That makes it a useful demonstration of this pair, but it does not measure how reliably unrelated real-world selfies will transfer to a new setting.
The generated output
Listen to the original generated vocals and instrumental. The web copy was compressed for playback; it was not made by pasting faces into the Quavo and Takeoff recording.
What the files and generation record show
This is one observed render, not an average, a speed comparison or a delivery promise. The record does not isolate time in the queue from rendering and download polling.
At today's public settings, Standard 720p for 8 seconds uses 7 credits. This demo was generated through an editorial API workflow; that retail quote describes a new studio render, not an invoice for this historical demo.
| Item | Recorded or measured value |
|---|---|
| Requested model | Standard / Seedance 2.0 Mini |
| Requested duration | 8 seconds |
| Published file | 720 × 1280, 9:16, 8.096 seconds |
| Video and audio | H.264 video + AAC audio at 44.1 kHz |
| Web file size | 2,831,822 bytes, approximately 2.83 MB |
| Reference input | One joint PNG image of two fictional adults |
| Submitted / downloaded | October 2, 2026, 13:25:24 / 13:32:25 UTC |
| Observed wait | About 7 minutes from submission to recorded download, including queue and polling |
Listen to the generated words
This draft transcript was produced automatically and can contain recognition errors. The render did not reproduce all intended wording exactly. Listen to the embedded video to judge clarity and the second line yourself.
The public studio accepts a topic of up to 240 characters, not a locked line-by-line script. If a name or exact sentence is essential, inspect the finished audio before sending the video as a gift.
| Turn | Draft transcript of the audible output |
|---|---|
| First performer | We bring the laughs, we bring the glow. |
| Second performer | Best friends forever, still in the show. |
Use this case as a quality checklist
- Compare both characters against the joint reference at several frames, not only the opening poster.
- Check lips, fingers and the microphone for distracting artifacts.
- Listen to the second line and decide whether the generated wording works for your intended occasion.
- Allow extra time for a gift. A single successful example does not establish a success rate or a deadline guarantee.
What you can learn from this one render
| Evidence | Useful conclusion | What remains open |
|---|---|---|
| Actual input and output | You can compare this pair’s appearance directly | Transfer from unrelated real-world selfies was not measured |
| Generated audio track | The downloadable clip includes original sound | A familiar name or exact sentence needs a listening check |
| Measured 8.096-second file | The published clip is a short complete performance | Other requested lengths produce different pacing |
| Recorded seven-minute interval | Leave time to watch, download and review | This single interval is not an average or a delivery guarantee |
Questions before you create
Can I use this reference to test the workflow?
You can view the published reference and video here. To create your own performance, upload your own authorized subjects. Using the same image or topic still creates a new result.
Was this a normal customer order?
No. It was an editorial demo rendered with a directed prompt. The model, reference and output are documented; the public studio's topic-based workflow does not expose that full prompt.
Does seven minutes mean my video takes seven minutes?
No. It is the observed submission-to-download interval for this one demo. Queue load, model choice and polling affect completion time.