AI Video Aspect Ratios Explained: 9:16 vs 16:9 vs 1:1 (and When to Use Each)
Pick the right aspect ratio before you generate — platform-by-platform guidance for TikTok, Reels, YouTube and feeds, plus how ratios work across AI video models.
Aspect ratio is the one setting you can't fix after generating. Crop a 16:9 clip to vertical and you lose half the frame — usually the half with the subject in it. Thirty seconds of planning here saves entire re-generations, so here's the complete decision guide.
The three ratios that matter
9:16 — vertical (TikTok, Reels, Shorts, 视频号)
Full-screen on phones, which is where short-form video lives. If your clip is destined for any feed that scrolls vertically, generate at 9:16 — don't generate wide and crop. Composition tip for prompts: vertical frames favor single subjects and depth ("walking toward the camera down a neon street") over side-by-side action.
16:9 — widescreen (YouTube, web embeds, presentations)
The cinematic default. Landscapes, establishing shots, product hero videos on a website — anything viewed on a laptop or TV. Camera-movement prompts ("drone shot sweeping across…", "slow pan over…") have room to breathe here that vertical simply can't offer.
1:1 — square (feed posts, thumbnails, grids)
The compromise format: acceptable everywhere, optimal nowhere — which is exactly why it's useful. One square generation can serve an Instagram feed post, a product grid tile, and a blog inline video without recropping. Macro subjects (jewelry, food, abstract loops) sit beautifully in a square.
How ratios work across AI models
Not every model exposes the same control — this trips up almost everyone:
- Kling 2.6 and Kling 3.0 Turbo accept an explicit ratio: 9:16, 16:9, or 1:1. You choose before generating.
- Hailuo 02 and Wan 2.6 decide automatically from your prompt and input. Writing "vertical smartphone video of…" nudges them, but it's a suggestion, not a setting.
- Image-to-video always follows your source image. A vertical photo produces a vertical video. If you need a 9:16 result from a wide photo, crop the image first — that's the only reliable lever.
One idea, three ratios
Because generations are credit-priced individually, the "shoot once, crop later" habit from real cameras inverts: generate per-platform instead. A Kling 2.6 5-second clip costs 28 credits, so producing the same concept in 9:16 + 16:9 + 1:1 runs 84 credits (~$2) — and each version is composed for its frame, not amputated into it. That's the difference audiences actually notice.
Quick reference
| Destination | Ratio | | -------------------------------- | ----------------------------------------------------------------- | | TikTok / Reels / Shorts / 视频号 | 9:16 | | YouTube main, website hero | 16:9 | | Instagram feed, grids, blogs | 1:1 | | Undecided / multi-platform | generate 9:16 first (largest audience), add others if it performs |