A specific promise with no mechanism, against a mechanism with no promise
One page promises that leads stay the same people across every episode, down to face, wardrobe and chemistry. The other documents one to seven reference images per call and states that nothing persists between calls. The vaguer page promises far more. As of 2026-09-16.
| Point at issue | OpenArt | Vidu |
|---|---|---|
| What is promised | Face, wardrobe and chemistry, every episode | A consistent subject, per call |
| What is documented | No object, no step | The request, the image count, the input limits |
| Who is the reader | A producer choosing a product | A developer writing an integration |
Inclusion rule. The contrast is between the kinds of evidence, not between capabilities. A promise and a documented interface answer different halves of the same question. Order. The promise first, then the documented interface.
1The inverse relationship here is worth naming
The page that publishes the most detail promises the least, and the page that publishes no mechanism makes the largest claim. That is not a coincidence: a developer reference has to be true for an integration to work, and a product page has to be attractive.
A reader comparing the two on a capability grid would see one strong row and one weak row and conclude the opposite of what either page supports.
2Wardrobe and chemistry are the right things to promise
Clothing is what goes first in a generated series, and chemistry is a property of two characters at once. Naming them means the page was written by somebody who has watched it fail, which is genuinely informative even without a mechanism.
It also gives a buyer a test plan: three named attributes are three things to check in an afternoon. A trial can settle what the page cannot.
3Seven images and no store is a complete answer to a narrower question
The interface answers what happens on one request, exhaustively, including the minimum pixel size. It declines the season question entirely, which is honest and leaves the whole continuity strategy with whoever is calling it.
Neither page is deficient. They are documents for different buyers, and the register's job is to stop one being read as the other.
4The wording both sides published
- Cross-episode castLeads are described as staying the same people across every episode, with the same face, wardrobe and chemistryan outcome without a mechanism
- Cross-episode castNo reusable character or element stored between calls is documentedwhat the documentation does not describe
5The same column, read for a third tool
On the cross-episode cast, SceneMixer's published page puts it in one line: A full-body sheet per character, made while the script is analysed
. Its reading here, A reference sheet made once and cropped into every segment, sets out what that line settles for a series and what it leaves to a production run.
6Sources
Read from the drama generator page at openart.ai and the reference-to-video API reference at platform.vidu.com, checked 2026-09-16 and 2026-09-12 respectively. The full column is on cross-episode cast; the single readings behind this contrast are OpenArt and Vidu. Other contrasts filed here: Google Flow and Vidu, DramaSo and Dramily, TrendShort and Showrunner Studio.