AI Videos: The Three Kinds You Can Make, and What Each Costs
Try it: write down where the video will be watched (a vertical feed, a product page, a channel) and what you already own (words, photos, a clip). Those two answers pick the kind below almost every time.
Kind 1: a story episode
You bring: a script or a novel (.txt, .docx, .md or pasted), or one line of idea that the AI Create tab writes into a script. You get: a cast with consistent faces, locations and props, a storyboard for each episode, and video with spoken dialogue, ambience and a score. The read splits the script into episodes of a minute or two each, so a longer script means more episodes, not longer ones.
Shape and length. Landscape 16:9 or portrait 9:16, chosen once in Step 1. One generation is a segment of up to 15 seconds, or up to 30 seconds on the long-take tiers; the segments are joined into the episode, and an episode can hold up to 60 of them. A segment may be much shorter than its ceiling: two short shots can run only a few seconds. Quality is the tier you pick per segment, from 480P up to 2K. Dialogue can be in any of 15 languages, one voice per character.
Cost shape. Reading the script is paid once, at 4 credits per 3,333 English characters. Each episode’s storyboard is 1 credit per 100 counted characters. Video is 10 credits a second on the default tier and the join is 1 credit a second. Worked examples at the default tier: a 60-second episode is about 660 credits ($4.11); a 90-second episode about 990; a series of ten 90-second episodes from a 30,000-character novel about 10,026 credits, $62. All before retakes. Where credits go in one episode · AI movies: how long they run.
Kind 2: an ad of 15, 30 or 60 seconds
You bring: one to four photos of the same product, dish, shop or place, and optionally one line saying what it is and who it is for. You get: an ad with a script, a cast, sets, sound, music written for it and an end card carrying your brand.
Four pickers decide it. Ad vibe (Epic, Funny, Warm, Minimal, Twist, Mouth-watering close-ups, Show how it works, A day in the life) changes the story and the end-card colours but not the price. Length is 15, 30 or 60 seconds, which is one, two or four 15-second segments, each shot on its own and joined with the end card last. Format is landscape or portrait. Quality is 480P, 720P or 1080P, and the menu prints the all-in price beside each one. Length and quality move the price; vibe and format do not.
Cost shape. The one price on the start button is a cap, not a bill. Reading the photo is free, cleaning it is one sheet, the script is a writing charge, the breakdown is a read, sheets are per image, shots are per second at the tier, and the final cut is per second. Each stage charges its own actual price, so the total lands at or below the number you saw. An ad is never the free first project, and it cannot draw on the free setup images. The ad’s one price is a cap · Ad vibe, length, format and quality.
Kind 3: a remake of a clip you shot
You bring: one MP4 or MOV, up to 5 minutes at the time of writing and within the megabyte limit printed on the upload box, and optionally one line naming who is who. You get: the clip read shot by shot into a cast, sets, props and a finished storyboard, then filmed again with your own characters in the look and language you choose.
What carries over and what does not. The blocking, the order of shots and the dialogue structure are read from your clip. The faces are not: the source is never handed to the video model, and the people in the result are your cast’s. Reading a clip does not give you a licence to its story, so use material you have the rights to.
Cost shape. The read is priced by the clip’s length alone, in minutes rounded up; the number of cuts never enters the price. Everything after the read (sheets, storyboard, video, join) is priced as in a story project. How to remake a video with AI · Source video limits, read time and price.
The three side by side
| Story episode | Ad | Remake | |
|---|---|---|---|
| You bring | Script, novel or one-line idea | One to four photos | One MP4 or MOV |
| Length | As long as the script; segments of 15 or 30 seconds | 15, 30 or 60 seconds | Follows the clip; you edit the storyboard |
| Shape | 16:9 or 9:16 | Landscape or portrait | 16:9 or 9:16 |
| Quality | 480P to 2K per segment | 480P, 720P or 1080P | 480P to 2K per segment |
| Price shape | Read once, storyboard per episode, video per second, join per second | One capped price; stages charge their own | Read by minutes, then as a story |
| Free first project | Yes: up to 15 setup images and the first storyboard | No | No |
All three end in the same editor: cast, sets, props, episodes and segments. You can fix one shot by editing its text and regenerating only that segment. Four ways to start a project.
Choose by where it will be watched
| Where it will be watched | Start with | Why |
|---|---|---|
| A vertical feed (TikTok, Shorts, Reels) | A 9:16 story episode of about a minute | Open on a hook and end on a turn; the shape cannot change after Step 1. Posting guide |
| A product or shop page | A 15 or 30 second ad | One idea, your brand on the end card, a price you can see first. Ad from one photo |
| A channel with a returning cast | A multi-episode story in 16:9 | One cast carries across every episode. Making a series |
| A scene you already shot | A remake | Keeps your blocking, replaces the people. Recast a remake |
What none of the three is
These are built for stories, ads and remakes with people in them. They are not screen recordings, slide narration, or a single still brought to life on its own, and they do not keep a real person’s face from a photo or use famous characters and brand mascots. A generation takes minutes, not seconds, and two identical requests can give two different clips, which is why you read the storyboard before you pay for video.
| Kind | Starts from | Typical first spend |
|---|---|---|
| Story episode | Words | The read, then setup images (15 free), then one segment |
| Ad | Photos | The capped price on the start button; stages charge less if they can |
| Remake | A clip | The read, priced by the clip’s minutes |
FAQ
What kinds of AI video can I make here?
Three: story episodes from a script, idea or novel; 15, 30 or 60 second ads from up to four photos; and remakes of a clip you shot, filmed again with your own cast.
How long can an AI video be?
A single generation is up to 15 seconds, or 30 seconds on the long-take tiers. An episode joins segments, and a series is as long as the script. Ads are 15, 30 or 60 seconds.
How much does an AI video cost?
At the default tier a 60-second story episode is about 660 credits, $4.11 at the lowest credit price, before retakes. An ad shows one capped price before you start.
Which kind is cheapest to try?
A story project, because the first drama gets free setup images and a free first storyboard, and each account has one free preview of up to 5 seconds. Ads and remakes do not.
Can I make a vertical video?
Yes. Choose 9:16 Portrait in Step 1 of a story or remake, or Portrait format in an ad. The shape cannot change after you continue.
Does the video have sound?
Yes. Dialogue is spoken inside each clip, one voice per character, with ambience, and each episode gets a score you can switch off.
Can I turn one photo into a moving picture?
Not on its own. A photo can become an ad, or the face or the room that a story follows. A single still brought to life is not what this makes.
Can I use my own footage?
You can remake it: the clip is read and filmed again with new characters. The source clip is not given to the video model and the faces are not kept.
Pick the kind, then read before you pay
Every kind shows its price before video starts. Read the storyboard first; it costs nothing to edit.
Updated November 2, 2026