Convert 16:9 Gameplay to 9:16 Without Cropping the HUD

The geometry is brutal: a 1920x1080 frame cropped to fill a 1080x1920 canvas keeps a 608-pixel-wide column — roughly 32% of your original frame. The other 68% gets deleted. In a talking-head video that is fine, because the face is in the middle. In an FPS clip, the middle is a crosshair and a wall. The information viewers actually read — the kill feed, the minimap, ammo and health — lives in the corners you just threw away.

That is why every generic converter chokes on game footage: to convert 16:9 gameplay to 9:16 without cropping away the HUD, a tool has to treat the corners as the payload — and these tools were built for podcasts. Their default advice (crop to the action, let AI track the subject) deletes exactly what makes a game clip legible. Follow it and your viewers see someone shooting at nothing, with no kill feed confirming the hit.

This guide covers the fix that actually works — the blurred pillarbox — three ways: in CapCut, in Premiere or Resolve, and as a single copy-paste ffmpeg command, plus a batch loop for converting a whole session. It is one piece of the larger workflow in our guide to making gaming clips vertical for TikTok and Shorts.

What crop-to-fill actually deletes from a game clip

Run the numbers on a standard 1080p clip. To fill 9:16, the frame is scaled until its height hits 1920, which makes it 3413 pixels wide — then everything outside a centered 1080-pixel window is cut. Here is what falls outside that window in most shooters:

• Kill feed (top-right): gone. In Warzone, Apex, Battlefield, and Delta Force this is the only on-screen proof of who you hit and how. A montage without a kill feed reads as unverified.
• Minimap (top-left or top-right): gone. Viewers use it to understand rotations and where the next fight comes from.
• Ammo, health, shields, killstreaks (bottom corners): gone. The low-ammo, one-shot clutch moments lose all their tension when nobody can see the counter.
• Round economy and scoreboard (top edge in Valorant and CS2): partially or fully cut, so a 1v4 retake looks like a random kill.

Hit markers and damage numbers survive because they sit near the crosshair — which is the trap. The clip still looks like gameplay, so people ship it, and then wonder why the retention graph falls off after the first kill. The viewer cannot follow the play.

AI subject-tracking reframe fails on gameplay for a more fundamental reason: there is no subject to track. The tools are trained to follow faces and moving objects. Gameplay has no face, the crosshair is already dead-center and never moves relative to the frame, and the information is static UI in the corners. An AI reframe of an FPS clip confidently produces the same center crop you would have done by hand — it just charges a subscription for it.

The three ways to make a 16:9 clip vertical

1. Crop to fill

Scale to fill the 9:16 canvas, cut the sides. Full-screen video, zero effort, and the HUD is gone. Acceptable only for cinematic b-roll where no UI matters.

2. AI subject-tracking reframe

Tools like Opus Clip follow a detected subject and slide the crop window around. Built for talking heads; on gameplay it degrades into method 1 with extra latency and a monthly fee.

3. Blurred pillarbox (full-frame stack)

Two layers: a copy of your clip scaled up to fill the whole 1080x1920 canvas and heavily blurred (the background), with the untouched 16:9 frame centered on top (the foreground). 100% of the original frame survives, the blur reads as intentional rather than lazy, and the vertical file is what TikTok, Shorts, and Reels expect. This is the method the TikTok clipping community standardized on, and the only one of the three that respects the HUD.

If you want a deeper head-to-head with retention reasoning, we wrote a full comparison in blurred pillarbox vs crop-to-fill for FPS Shorts.

How to do the blurred-pillarbox conversion manually

In CapCut

1. Create a new project and import your clip.
2. Set the aspect ratio to 9:16. Your clip letterboxes to the center of the canvas.
3. Select the clip, open Canvas (Background on mobile), and choose Blur.
4. Pick the second or third blur strength. Too weak and the background competes with the gameplay; too strong and it turns into gray mush.
5. Export at 1080p, 30 or 60 fps to match your source.

CapCut is the fastest free GUI route, but it re-encodes on export and the free tier has been shifting features behind Pro. We covered the details and workarounds in our CapCut blurred background guide.

In Premiere Pro or DaVinci Resolve

1. Create a 1080x1920 sequence/timeline at your clip's frame rate.
2. Place the clip on two stacked tracks.
3. Bottom track: set Scale to 178% so it fills the full height, then add Gaussian Blur at 75-100 and enable "Repeat Edge Pixels" (Premiere) so the edges do not darken.
4. Top track: set Scale to 56.25%, which fits the full 1920-wide frame into the 1080-wide canvas. Leave position centered.
5. Export H.264, 1080x1920. Use your GPU encoder if available.

With one ffmpeg command

This is the whole technique in a single filter chain — paste it, change the filenames, done:

ffmpeg -i clip.mp4 -filter_complex "[0:v]scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920,boxblur=20:5[bg];[0:v]scale=1080:-2[fg];[bg][fg]overlay=(W-w)/2:(H-h)/2" -c:v h264_nvenc -cq 21 -c:a copy short.mp4

Line by line: the first branch scales the source up until it covers the full 1080x1920 canvas, center-crops it, and blurs it (boxblur=20:5 is a radius-20 blur applied 5 times — strong enough that the background never distracts). The second branch scales the untouched frame to 1080 wide (608 tall for 16:9 sources; -2 keeps the height encoder-friendly). The overlay centers the sharp frame on the blurred one. -c:a copy passes audio through untouched — no reason to re-encode it. On systems without an NVIDIA GPU, swap -c:v h264_nvenc -cq 21 for -c:v libx264 -crf 19. The full filter reference lives in the ffmpeg filters documentation.

Mind the platform safe zones

Every short-form platform draws UI on top of your video: captions, the engagement rail, the sound attribution. Post a clip where the ammo counter sits under TikTok's caption block and you have cropped it just as surely as a center-crop would. Approximate reserved areas on a 1080x1920 canvas:

Here is the quiet advantage of the blurred pillarbox: a centered 16:9 frame spans roughly pixels 656 to 1264 vertically on a 1920-tall canvas. TikTok's caption zone starts around pixel 1536 — your entire game frame, HUD included, clears it by more than 270 pixels. Crop-to-fill, by contrast, puts gameplay under every overlay on the screen. The one overlap to watch is TikTok's right-side icon rail, which can graze the right edge of the centered frame near its bottom corner; upload one test clip and check before you batch a whole session. YouTube publishes its own format guidance in the aspect ratio help docs if you want the source numbers.

The automatic way: native 9:16 export in GameClipper Advanced

Full disclosure: GameClipper Advanced is our app, and we built its vertical export because we were tired of running every clip through the two-layer stack by hand. It is a Windows 10/11 desktop tool that works on footage from any recorder — ShadowPlay, OBS, a capture card, Xbox Game Bar — and does the conversion as part of the clipping pass:

1. Import your full VOD — no cutting required first.
2. Calibrate once per game: draw a box around the kill feed and capture your in-game name. It saves as a profile.
3. The computer-vision scan finds every kill in the VOD, with OCR name confirmation so teammate kills do not sneak in.
4. Review the auto-built timeline: preview each clip, nudge starts and ends, drop the weak ones.
5. Export vertical 9:16. The full 16:9 frame is centered with a blurred pillarbox generated automatically — the HUD is never cropped, on any clip, with zero per-clip setup. Exports run through GPU NVENC — an NVIDIA GPU is recommended, not required.

Everything runs locally; your footage never leaves your machine. The free version includes the full detection, editing, and vertical export pipeline forever, with a small watermark on exports — a one-time $19 upgrade removes it. No subscription, which matters when the AI-reframe tools charging monthly cannot even handle a kill feed.

Batch-converting a whole session

Volume is the actual job: if you post daily, you are converting Friday night's whole session, not one clip. Three realistic options:

• ffmpeg loop: drop all your cut clips in a folder and run this in PowerShell:

Get-ChildItem *.mp4 | ForEach-Object {
  ffmpeg -i $_.FullName -filter_complex "[0:v]scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920,boxblur=20:5[bg];[0:v]scale=1080:-2[fg];[bg][fg]overlay=(W-w)/2:(H-h)/2" -c:v h264_nvenc -cq 21 -c:a copy ("vertical_" + $_.Name)
}

• Editor templates: save the two-layer stack as a preset in Premiere or Resolve and drag clips through it. Works, but it is still one clip at a time and you still had to find and cut every kill first.
• Detection-to-export in one pass: the reason we bother with kill-feed detection at all is that cutting is the slow half. GameClipper scans the whole VOD, finds the kills, and exports them all vertical in one run — the conversion stops being a separate chore. If your source is a stream archive, the same idea powers our Twitch VOD to TikTok workflow.

Common mistakes that still ruin vertical exports

• Cropping first, asking later. Once you export a crop, the corners are gone for good — always convert from the 16:9 master.
• Exporting below 1080x1920. Platforms upscale small files badly and text-heavy HUDs smear first. 1080x1920 minimum, always.
• Blur too weak. A barely-blurred background full of readable motion competes with the actual clip. If you can read the duplicate kill feed in the background, double the blur.
• Black bars instead of blur. They play fine technically, but on a feed full of full-canvas videos they read as unedited and cost you the swipe-past.
• Re-encoding audio for no reason. Copy the audio stream (-c:a copy) — every needless transcode costs quality.
• Ignoring safe zones. A perfect pillarbox export can still lose its bottom edge to a caption — check the table above before adding text.

Your 60-second pre-upload checklist

• Full 16:9 frame present and sharp, centered on 1080x1920?
• Kill feed, minimap, and ammo counter all readable at phone size?
• Background blurred hard enough that nothing in it is legible?
• Hook text in the top blurred band, nothing in the bottom 25%?
• Exported 1080x1920, source frame rate, audio stream copied?

That is the whole discipline: the frame stays whole, the blur does the formatting, and the platforms' UI never touches your HUD. Editor stack, ffmpeg one-liner, or GameClipper Advanced applying the pillarbox to every detected kill automatically — the rule never changes: keep the 16:9 master, convert from it, and never ship a clip whose kill feed you cannot read at phone size.