Convert Landscape Photos to 9:16 for Reels & Shorts with AI
Turn horizontal photos into vertical Reels and Shorts in 60 seconds with AI outpainting — no cropping, no awkward dead space, no losing the subject. Exports and hook overlays included.
Product Marketing
Every creator who has ever tried to repost a horizontal photo to Reels or Shorts has hit the same wall: the platform expects 9:16 vertical, the photo is 16:9 or 3:2 horizontal. The platform-default fix is to either letterbox the photo with black bars top and bottom or center-crop it into a vertical slice that loses most of the composition. Neither option works. Letterboxed posts look amateur and the algorithm treats them as low-quality. Center crops cut off products, faces, and the entire left and right context that made the original photo worth posting.
AI outpainting solves this in a way that no manual cropping ever could. Instead of removing pixels to fit the vertical aspect ratio, the AI invents new pixels above and below the original frame, extending the sky, the wall, the table, the floor — whatever context surrounds the subject. The original photo stays intact at full width; the new 9:16 frame wraps the old one with believable extensions. The result reads as a native vertical photo, not a converted one.
This walkthrough covers the exact workflow for converting any landscape photo into a Reels or Shorts post: identifying the subject's safe zone, outpainting top and bottom, choosing the right 9:16 crop, running one boost pass to even out the new pixels. Exporting both the 1920-tall master and the 1350-tall feed derivative that Instagram needs separately. Two minutes per photo once your template is set.
- AI outpainting beats center-cropping for landscape→vertical conversion — it adds pixels instead of removing them.
- Subject must stay in the upper or center 75% of the frame; Reels and Shorts overlay UI on the bottom 25%.
- Export two versions: 1080×1920 for fullscreen playback, 1080×1350 for Instagram's feed preview crop.
- One pass of AI enhancement after outpainting hides the seam between original and AI-generated regions.
- 3-5 word hook overlay in the upper third gives viewers a reason to keep watching past the first half-second.
Why landscape photos fail on Reels and Shorts (and why cropping doesn't fix it)
Reels and Shorts both use 9:16 as the canonical aspect ratio. The video player fills the screen vertically; anything that isn't already 9:16 gets one of two default treatments. Either the platform letterboxes — adds black bars top and bottom to fit the horizontal photo in the middle of the screen — or the user crops the photo to a tall vertical slice manually before uploading. Letterboxing produces a post that looks broken; the algorithm has been observed deprioritizing such posts because they read as low-effort. Manual cropping loses everything outside the narrow center slice. Usually means losing the parts of the photo that made it worth posting.
The alternative most creators reach for is taking new vertical photos for vertical platforms. Is correct but expensive and slow. Reshoots double the production time of every campaign and require the subject to still be available. For evergreen content, archival product photography, and any photo that was perfect in 16:9 but never reshot in 9:16, the only real path is converting the existing file. The math of conversion is straightforward: a 16:9 photo is 1.78 wide for every 1 tall. A 9:16 photo is 0.56 wide for every 1 tall. To convert, you need to add about 3.16× the original height in new vertical real estate above and below the existing frame. You cannot crop your way there.
AI outpainting is the only tool that produces this kind of pixel growth without re-shooting. The model extends context-matched sky, walls, ground, or backdrop above and below the original photo. The extensions look natural because they're generated from the actual pixels of the source. It's not a creative liberty — it's a reconstruction of the frame the photographer would have captured if they'd stood further back with a vertical sensor.
- Reels and Shorts both use 9:16 — letterboxing or center-cropping a 16:9 photo destroys it.
- Manual vertical reshoots are the right answer but cost double; not feasible for archival photos.
- Outpainting adds 3.16× the original height in believable pixels — the only viable conversion path.
- AI extensions are reconstructions, not creative liberties — they extrapolate the actual photo context.
Identifying the subject's safe zone
Before opening any editing tool, look at the original landscape photo and decide which rectangle of the frame must survive the conversion intact. For most photos this is obvious — the subject's face, the product, the focal element. For complex compositions with multiple subjects or asymmetric framing, the safe zone is the smallest rectangle that contains every element you want preserved.
Photos where the subject sits in the dead center convert easily; outpainting just adds context above and below. Photos where the subject is on the left or right edge are harder. You may need to outpaint asymmetrically (more on one side) and accept that the final crop will not be a simple center crop. Photos where the photographer's framing was already vertical-friendly (subject upper-third, lots of empty space around) are the easiest of all and may only need outpainting on one side.
If the subject is too wide to fit in any 9:16 crop without center-cropping, the photo cannot be cleanly converted. Group shots taken with the camera held very wide, panoramic landscapes. Architectural shots that depend on the full width are examples. For those cases the better strategy is to repurpose the photo as a static post (carousel slide or 1:1 grid post) rather than fighting the aspect ratio.
- Safe zone = the smallest rectangle containing every element you must preserve.
- Centered subjects convert easily; edge subjects require asymmetric outpainting.
- Already-vertical-friendly compositions only need outpainting on one side.
- Wide groups, panoramics, and width-dependent architecture are not cleanly convertible — repurpose as static posts instead.
The outpainting step
Upload the landscape photo to Magic Eraser AI Fill. Select the outpaint option and drag the canvas boundary upward (to extend the top) and downward (to extend the bottom). For most 16:9 → 9:16 conversions, you'll add roughly 100% of the original height in new pixels above and another 100% below, giving you a final canvas that is about 16:27 — well taller than 9:16. The AI extends sky, walls, ground, or whatever is at the top and bottom edges of the original.
Run the outpainting in two passes if the first result looks awkward. Some AI fill models produce strange details near the new-pixel boundary. Running outpainting again with the result as input usually smooths the transition. Two passes is plenty — more produces diminishing returns and starts to introduce repetition patterns.
Watch out for outpainting failure modes. If there are objects, edges, or distinctive textures near the top or bottom of the original frame, the AI may extrapolate them strangely. A tree trunk doubled, a horizon line jaggy, a wall with two sets of moldings. Look at the result at full size and re-do the pass if something looks wrong. The fix is usually rerunning the outpainting on just the problem region. Some tools let you mask just the broken area for a targeted regenerate.
- Outpaint roughly 100% of the original height on both top and bottom edges.
- Two passes usually smooths the boundary; more than two introduces repetition artifacts.
- Common failure modes: doubled tree trunks, jaggy horizon lines, repeated wall moldings.
- Targeted regenerate on just the broken region fixes most failures without redoing the whole frame.
Choosing the final 9:16 crop
After outpainting, the image is taller than 9:16. Crop to exactly 1080×1920 with the subject positioned mindful of how Reels and Shorts overlay their UI. Both platforms place caption chips, profile avatars, play buttons, and 'like' affordances along the bottom 20-25% of the screen. Anything in that band gets visually overlapped by the platform chrome, so the subject should never sit there. The upper third or the center is safe; the lower third risks being clipped or overlapped.
If you have flexibility on the horizontal positioning of the crop, place the subject's eye-line (if there's a face) or the center of mass (if there isn't) on the upper third. The classic rule-of-thirds placement at 33% from the top. This produces the most engaging composition for vertical mobile viewing. The bottom 67% becomes negative space, which is fine on mobile because the screen is short and the eye reads top-to-bottom.
For posts that use the photo as a backdrop with text overlay (which is the dominant Reels and Shorts photo format), placing the subject on the upper third and the text overlay in the center vertical band is the proven layout. The text sits over the subject's body or torso, never over the face, and never in the platform's chrome band.
- Crop to exact 1080×1920; keep subject above the bottom 20-25% chrome band.
- Subject eye-line on the upper third (33% from top) is the rule-of-thirds anchor.
- Negative space below the subject is fine; mobile screens read top-to-bottom.
- Text overlay sits over the body or torso, never over the face.
Enhancement and seam-hiding
Outpainted regions are AI-generated, and current AI fill models produce regions with slightly different texture traits than the original camera-captured pixels. The new regions are smoother and less detailed at fine scales. Under bright direct lighting the seam between original and outpainted regions can be visible as a faint band of softness.
One pass of AI boost applied to the full final image evens out the texture differences. The boost model treats the whole frame uniformly, sharpening fine detail in the outpainted regions to match the original and slightly softening any over-sharp edges in the original to match the outpainted regions. The net result is a uniform-looking frame where the seam is no longer detectable.
Do not run multiple boost passes. Each pass shifts texture traits toward the boost model's preferred output. After two passes the photo starts to look filtered. One pass is the right amount — enough to hide the seam, not enough to over-process.
- Outpainted regions are softer than camera-captured pixels; one enhancement pass evens them out.
- Enhancement treats the full frame uniformly — both the original and the AI-generated regions.
- Two or more passes over-processes the image; the photo starts to read as filtered.
- If the seam is still visible after one pass, the outpainting itself probably needs another pass first.
Exporting for Reels, Shorts, and the Instagram feed preview
Reels and Shorts both display 9:16 (1080×1920) when played in their dedicated viewing surfaces. Export that as the master. But Instagram's main feed shows a 4:5 (1080×1350) crop of the same Reel for users who scroll the home feed without entering Reels. Instagram auto-center-crops the 9:16 into 4:5, which usually produces an awkward result with the subject crammed against the top of the preview.
Export a second 1080×1350 derivative manually, recomposing the subject to fit the shorter feed crop without losing the hook. This sounds like extra work for every Reel but takes about 30 seconds. You're cropping the existing 1080×1920 down to a 4:5 with smarter framing than Instagram's auto-crop. For weekly creators, the time investment is small and the feed-preview quality difference is large.
YouTube Shorts uses 9:16 across all surfaces and does not need the 4:5 derivative. TikTok similarly uses 9:16 in the feed but does need a separate 1:1 (1080×1080) for the profile grid (covered in the companion TikTok thumbnail piece). Export targets vary by platform. One workflow that produces all three sizes from a single source is what saves you time.
- 1080×1920 master for fullscreen playback on both Reels and Shorts.
- 1080×1350 derivative for Instagram's feed preview — auto-cropping the 9:16 is reliably worse.
- YouTube Shorts uses 9:16 across all surfaces; no derivative needed.
- TikTok needs an additional 1080×1080 for the profile grid (separate workflow).
Sources
- Instagram Reels Specifications — Instagram Help Center
- YouTube Shorts: Create a Short — YouTube Help