SceneMixer on keeping a cast consistent
SceneMixer holds the single five for cast consistency. A five needs a named, reusable store of character references that the tool applies without being asked each time, which is a documentation standard rather than a measurement of how stable the faces are. As of 2026-09-12.
| Point | As the rubric has it, or as the table records it |
|---|---|
| The question this column asks | Can the same character be kept across separate generations? |
| What this tool rates | 5 of 5 |
| Tools holding the same value | One of the sixteen rated here |
| What a five looks like | A named, reusable store of character references applied automatically |
| What sits at the bottom of the scale | No mechanism described anywhere on the vendor pages |
| Tools left unrated on this column | None of the sixteen |
Inclusion rule. The rubric's own wording for this criterion, with this tool's value beside it. Nothing here is taken from a demo, a benchmark or a review. Order. The question first, then the value, then the rule it was given under.
1The whole column, with this row marked
| Tool | Cast consistency |
|---|---|
| CapCut | 1 of 5 |
| Dreamina | 3 of 5 |
| Hailuo (MiniMax) | 4 of 5 |
| invideo AI | 2 of 5 |
| Kling AI | 4 of 5 |
| LTX Studio | 4 of 5 |
| Luma Dream Machine | 3 of 5 |
| Morph Studio | 3 of 5 |
| Pika | 2 of 5 |
| PixVerse | 3 of 5 |
| Runway | 3 of 5 |
| SceneMixer | 5 of 5 |
| Sora (OpenAI) | 2 of 5 |
| Veo / Flow (Google) | 3 of 5 |
| Vidu | 4 of 5 |
| Wan (Tongyi Wanxiang) | 3 of 5 |
Inclusion rule. Tools rated on this rubric: a public English product page addressing at least four of the seven criteria. An unrated cell means nothing usable is published on that point. Order. Alphabetical by tool name.
2Why this column decides whether a series is feasible
A drifting face is the failure that ends serial production. Every other criterion affects how good an episode looks; this one affects whether sixty episodes can be the same story. Four tools reach four here and one reaches five, so the field is thin at the top of the column that matters most.
The four-versus-five distinction is whether the references are applied automatically or attached by hand each time. Over a season that is the difference between a pipeline and a discipline, and disciplines fail on the episodes nobody had time to check.
3What the rating deliberately does not claim
A five here says the mechanism is described and reusable. It does not say faces never drift, because nothing published by any vendor would support that claim and this register does not run its own trials.
A reader should expect drift anyway and read the cell as saying where the controls are. A documented store of references is something a production can audit when a face goes wrong; an undocumented one leaves nothing to check.
4The crowded middle of this column is the real story
Seven of the sixteen rated tools sit at three, which is the most crowded value anywhere on the rubric. Three means consistency is asserted and the mechanism is not described, which is the least actionable answer a vendor can give.
That crowding is why a single cell at the top is worth a page. The gap between three and five is not a difference of degree; it is the difference between a claim and something a colleague can be handed.
Ratings describe what each vendor publishes on its own pages, read 2026-09-12, and are given under the rule on how scoring works. This column across every tool is on cast consistency; this tool across all seven is on SceneMixer. Nearby cells: CapCut, cast consistency, SceneMixer, shot control. All of them: single ratings.