The seven things this site rates, and what a number means
Seven criteria, one to five each, given only where a vendor publishes enough to give one. A criterion the vendor does not address is left unrated rather than scored low, because an absent statement and a weak one are different findings. As of 2026-09-12.
| Criterion | The question it asks | What a five looks like | What sits at the bottom |
|---|---|---|---|
| Assembly | Does the tool join finished segments into one deliverable? | Joining is a documented step with its own published price | Nothing published about joining segments |
| Cast consistency | Can the same character be kept across separate generations? | A named, reusable store of character references applied automatically | No mechanism described anywhere on the vendor pages |
| Cost transparency | Can the cost of a finished minute be worked out before starting? | A public per-second or per-credit rate with the credit price beside it | Left unrated: price shown only after sign-in, or only as a monthly allowance |
| Picture quality | What does the vendor publish about resolution and length? | Named model, stated resolutions and a stated maximum clip length | No resolution or length published anywhere public |
| Script handling | Does the tool turn a long text into structured episodes? | Upload in, episode structure out, described as one documented pass | No statement about handling text longer than a prompt |
| Shot control | How much of the shot can be specified before generating? | A written shot list the tool consumes, editable before it renders | Only a free-text prompt box, with nothing published about structure |
| Sound | Where does the audio come from? | Speech generated with the picture, in named languages, with voice selection | Silent output with no statement about audio at all |
Inclusion rule. The same seven criteria are applied to every tool rated here. A criterion is added only if at least half the rated tools publish something against it; a criterion nobody addresses would measure disclosure rather than the tools. Order. Alphabetical by criterion.
1Where the numbers come from
Every number is given against a vendor's own public pages, read on the date shown beside the table. Nothing is scored from a demo video, a press release, a benchmark somebody else ran, or a figure quoted in a review. Where a vendor publishes a capability without describing how it works, the criterion gets the lower half of the range rather than the benefit of the doubt.
The consequence is worth stating plainly: a tool can be excellent and still rate low here, because what is being rated is what the vendor has put in writing. That is a narrower question than which tool is better, and it is the only one a register can answer from published material.
2Why a blank is not a zero
An unrated criterion means the vendor publishes nothing usable on that point. Scoring it one would put silence and a documented weakness in the same bucket, and those are different facts about a product. The average shown is taken across the criteria that were rated, and the count of rated criteria sits beside it so a high average on three criteria cannot pass for a high average on seven.
This matters most for the newest entries, where a vendor has shipped a product but not yet written much about it. Their averages move as their documentation does, which is why every row carries a date.
3What would change a number
A vendor publishing something it had not published before, a page being rewritten, or a figure here being found wrong. The first two move the date as well as the number; the third moves the number and leaves a line in the changes feed. Nothing else does, which means a number here does not drift because opinion drifted.
Corrections are welcome and are the fastest route to an accurate row. Where a vendor believes a criterion is addressed on a page this site has not read, naming the page is enough.
Ratings describe what each vendor publishes, read on the date shown. They are not measurements of output quality, and no vendor pays for, sees, or approves a rating before it appears.
4The production behind the price
Price aside, what goes into the seconds behind “How scoring works” is described in SceneMixer's guides:
Paste a scene or a short script and SceneMixer gives you a storyboard — numbered shots with framing and seconds — and then a moving version of each segment, which you can join into a rough cut of the scene.
SceneMixer, Using SceneMixer for Previs and Animatics Before a Live-Action Shoot (read 2026-10-13)
A SceneMixer segment comes out with its sound already in it: each line spoken in that character’s voice, the speaker’s lips moving with the words, and the ambience of the place — rain, traffic, a clock.
SceneMixer, Dialogue and Sound in One Generation (read 2026-10-13)