# Models The engines that make 10xvid's videos and images: what each makes, in which shapes and lengths, and how it checks its own work. ## Flagship model ### Motion V1 A video engine built for one job: ads for your app. It reads your site, follows the Format DNA and draws every frame, then checks the film before you see it. - Draws from your real pictures, colours and type. It does not generate pixels from a prompt, so text stays sharp and your product looks like itself. - Makes vertical 9:16, square 1:1 and landscape 16:9 from one Format DNA, each laid out for its frame, not cropped - 5 to 30 seconds, as a 1080p MP4 at 30 frames a second, mixed to the loudness feeds expect - A hundred styles, or Pure, your site's own design language - A voiceover, or none ## Models overview | Model | Makes | Shapes | Length | |---|---|---|---| | Motion V1 | Video ads | 9:16, 1:1, 16:9 | 5 to 30 seconds | | Motion V1 Flash | Video ads | 9:16, 1:1, 16:9 | 5 to 30 seconds | | Image V1 | Image ads | 9:16, 1:1, 16:9 | A still PNG | You choose the engine each time you make something: in the chat or on the blueprint, through the API with `engine`. Images are made by Image V1. ## Motion V1 Motion V1 is 10xvid's flagship engine. A team of steps works on each film: a reader takes in your site, a strategist finds who the ad is for and the one idea that sells it, a director casts every shot from more than 40 designs while a critic argues back, a designer lays each shot out for every shape, an animator moves it by measure, a composer scores it, and a judge watches every frame before you do. ### What it is good for - Ads to get your app installed, from your own screens - A new feature or offer, shown in a few seconds - Testing many angles and styles at once, without filming anything ### Key features - **One link in.** Every film starts from a site it has read, never from a blank prompt. - **A plan you can change.** The Format DNA holds the hook, the story, the pace, the camera and the look. Change a part and the film follows. - **Three shapes in one run.** Each shape is composed for its frame. - **Two sites never get the same film.** A film shares at most half of its designs with another site's. ### Languages It writes in the language of your site. English is where it is tested most. ### Considerations - It uses the pictures and screens your site already shows. A physical product needs its own photos there. - There are no presenters, no avatars and no stock clips. - A video takes a while to make. Through the API, ask for it again or let a [webhook](/docs/api/webhooks) tell you when it is done. ## Motion V1 Flash A quicker video engine. One pass makes each video: it reads your site, writes the words from what your site says, picks a voice and makes music for the ad, builds the film from your screens, colours and type, looks at its frames and fixes what is broken, then renders it. ### What it is good for - Many quick versions of an ad to test - A first ad for a new app ### Key features - **About five minutes a video**, once its turn comes. Videos are made one after another. - **Your site's own words.** Nothing is invented: no reviews, ratings or results your site does not show. - **Every shape.** Vertical 9:16, square 1:1 and landscape 16:9, each made for its frame. - **A voiceover, or none.** With none, the words are on screen over music made for the ad. - **A site is read once.** Every later video of the same site starts from that reading. ### Considerations - It follows your site and your notes on the format, not the Format DNA, the hundred styles or the look setting. - It has no quality gate. It checks its own frames before it renders, and every video it makes is delivered. - It makes videos only. Images are made by Image V1. ## Image V1 The image engine. One session reads your site, sketches three ways in, chooses a kind of ad and a way of drawing it, lays the poster out as a page from your own pictures, colours and type, and checks every word on it before it is drawn. ### What it is good for - Feed and story image ads to run beside your videos - Many different static ideas to test at once ### Key features - **Judged before you see it.** A judge reads each image at 8 of 10. One under the bar is made again from what the judge said, and the better of the two is yours. - **Text that stays sharp.** Nothing is cut, covered, too small or too pale: each word is checked on the drawn picture. - **Every shape.** Vertical 9:16, square 1:1 and landscape 16:9, each laid out for its frame. - **Your blueprint when there is one.** With a blueprint, the image follows its plan. Without one, it reads your site. ### Considerations - It uses the pictures your site already shows, and the ones you add. It does not generate photos. - An image takes about eight minutes, its second pass included. ## The quality gate Every Motion V1 film is measured before delivery, then judged by a judge that does not know which film is ours. | What is checked | What it means | |---|---| | Text | Readable in the first second. Nothing clipped, overlapping or too small. | | Layout | No panel cut by the frame. Safe margins kept. | | Motion | No snaps, no jitter, nothing that lands twice. | | Sound | There, and never busier than the picture. | | Brand | Your logo where it belongs, and a clear call to action. | | Craft | Judged blind, beside films we rejected and films we admire. | A film that misses is made again from what failed, on the shot and at the second it failed. The best version is the one you get. More in [Videos and quality](/docs/concepts/videos). ## Limits | | Motion V1 | Motion V1 Flash | Image V1 | |---|---|---|---| | Made at once, for an account | 10 | 10 asked for at once, made one after another | 10 | | Styles asked for in one format | 10 | Not used | Not used | | Length | 5 to 30 seconds | 5 to 30 seconds | A still | | Shapes | 9:16, 1:1, 16:9 | 9:16, 1:1, 16:9 | 9:16, 1:1, 16:9 |