# AI video makers for social: loops, captions and motion [Canonical HTML page](https://gradiently.design/guide/ai-video-maker) "AI video" covers three very different tools: ones that generate footage, ones that put a presenter on screen and ones that animate a design. For most social posts, the third kind does the most work. ## The short version - AI video makers fall into three groups: tools that generate footage from a prompt, tools that create presenters or avatars, and design tools that animate text and backgrounds. - Most short social videos are vertical, at 1080×1920 for Stories, Reels, TikTok and YouTube Shorts. - Captions should sit in the middle of a vertical frame, clear of the platform's buttons along the bottom and right edge. - A seamless loop, where the last frame flows back into the first, suits short social video because it replays without a visible jump. - Generated footage can contain errors such as warped text or hands, so check every frame before posting. An **AI video maker** for social media needs to do three things well: produce vertical video at the right size, keep captions readable on a phone and loop cleanly when the platform replays it. Different tools get there in different ways. Some generate footage from a prompt, some put a talking presenter on screen, and some animate a design you already have. For announcements, quotes, offers and tips, an animated design is usually the fastest route to something that looks like your brand. ## Three kinds of AI video maker | Type | What it does | Good for | Watch out for | | --- | --- | --- | --- | | Footage generators | Create video clips from a text or image prompt | Moods, b-roll, abstract scenes | Warped text, odd hands, inconsistent style between clips | | Presenter and avatar tools | Turn a script into a talking person on screen | Explainers, training, translations | Can feel uncanny; disclose synthetic people | | Motion design tools | Animate text, shapes and backgrounds in a layout | Announcements, quotes, tips, offers, loops | Needs a good design to start from | Pick the tool for the post. A Friday opening announcement does not need generated footage; a mood piece might. Many social videos are really moving graphics: words on a background, a logo, a short message. A motion design tool gives you control over exactly those parts, and the result stays on brand because it starts from your own layout. [Animated social posts](https://gradiently.design/guide/animated-social-posts) has ideas for this kind of video. ## Sizes for social video Vertical video fills a phone screen and is the default on Stories, Reels, TikTok and Shorts. Feed posts still accept portrait and wide video, so make the vertical version first and adapt it. The full list is in [social video sizes](https://gradiently.design/guide/social-video-sizes). - Reel, Story, TikTok, Shorts: 1080 × 1920 - Instagram and LinkedIn portrait: 1080 × 1350 - Square post: 1080 × 1080 - X and Bluesky post: 1600 × 900 Four frames to scale. One vertical master covers four platforms; the other three are for feeds. ## Captions people can actually read Many people watch with the sound off, so on-screen words carry the message. AI can transcribe speech into captions quickly, but placement and size are still your job. In a vertical frame, the platform's own interface sits across the bottom and down the right side. Captions placed there get covered. ### Avoid - Captions in the bottom fifth, under the platform's buttons - Small type that only reads on a laptop - Whole paragraphs on screen at once - White text straight over a busy background - Effects on every word ### Do - Captions in the middle band of the frame - Large type, checked on a real phone - Two short lines at a time - A calm, contrasting area behind the words - One simple entrance, used consistently [Video captions design](https://gradiently.design/guide/video-captions-design) goes into type, timing and safe zones. If you are using a background that moves, readability can change from second to second, which is where a background that responds to the words helps. ## How to make a seamless loop A seamless loop is a short video whose last frame flows straight back into its first. Platforms replay short videos automatically, so a clean loop looks intentional while a hard cut back to the start looks like a mistake. Loops also suit still words: the background drifts, the message stays put. 1. **Keep the words still** Let text arrive once, then hold. Moving words are harder to read and harder to loop. 2. **Move the background slowly** Gentle drift reads as calm and premium. Fast motion competes with the message. 3. **Match the first and last frames** Motion that returns to where it started joins invisibly. Check the join by watching it replay three times. 4. **Choose a sensible length** Long enough to read the words twice, short enough that it replays before attention drifts. [Video loop length](https://gradiently.design/guide/video-loop-length) has guidance per platform. 5. **Export at the platform size** MP4 at 1080×1920 for vertical. Upload the file as made, so the app has nothing to rescale. A vertical event story reading Doors open at 8 with a gradient background designed to loop A vertical event loop: still words in the middle band, the background drifting behind. A portrait running club tip reading Stretch before you run on a moving gradient background A portrait tip for the feed, made from the same approach with a different Mark. There is more on the technique, including how to test the join, in [seamless loop video](https://gradiently.design/guide/seamless-loop-video). ## Making social video in Gradiently Gradiently is a design tool built around Marks, living gradient and material backgrounds with their own light, grain and motion. Text can animate in with a rise, pop, wipe or typewriter, and the Mark keeps its calmest region behind the words so captions stay readable while it moves. On Pro you get a video timeline with captions, music, camera, screen and voice recording, and export as MP4, WebM or GIF where the browser supports recording, up to 4K with no watermark. The Designer, Gradiently's AI, can lay out a post on your Mark from a sentence and make the other sizes, so the vertical loop and the feed version come from one design. For recording yourself, see [record screen in browser](https://gradiently.design/guide/recording-camera-and-screen). > **Check generated footage frame by frame** If you do use a footage generator, scrub through every second before posting. Garbled signs, extra fingers and flickering edges are easy to miss at full speed and obvious once someone pauses. ## FAQ ### What is the best AI video maker for social media? It depends on the post. Footage generators suit moods and b-roll, avatar tools suit scripted explainers, and motion design tools suit announcements, quotes and offers that need to look like your brand. ### What size should a social media video be? Make vertical videos at 1080×1920 for Reels, Stories, TikTok and YouTube Shorts. For feeds, 1080×1350 portrait, 1080×1080 square or 1600×900 for X and Bluesky. ### Where should captions go on a vertical video? In the middle band of the frame, clear of the platform's buttons and captions along the bottom and the action buttons down the right edge. ### How do I make a video loop without a jump? Keep the words still, move the background slowly and make sure the motion ends where it started, so the last frame flows into the first.