Captions should not cover the SKU

Captions should not cover the SKU. The pack shot is the product: label, logo, shade name, a window in the pouch. Burned-in type on that label is a second headline that hides the first. Move the captions. Do not crop the carton to make room for a two-line block. The mute viewer needs both the words they can read and the box they can recognise.

The label is not a background

A pack shot is the still that has to stay readable. In a live-action commercial it is the hold at the end. In a generate it is usually the input still image-to-video is not allowed to melt. Hands, type on the carton, lid physics: those are the three failure modes. Captions are a fourth when finishing puts a transcript on the logo.

The SKU in the frame is proof. Lifestyle b-roll can be invented. The label cannot. Covering it with Classic, Glass, or a Paper card does the same damage as a model that crawled the type: the buyer cannot see what they are being sold. Mute autoplay makes this worse. There is no voice to say the name while the caption sits on it. There is a box with words on it, and different words on top of those words.

This is not the chrome problem. Safe area is username, caption field, CTA, Shorts AI overlay. Keep type out of that furniture too. This page is the product problem: even in a perfectly empty lower third, the pack may already live there. Two keep-outs. Chrome is one. The carton is the other.

Cropping the label to “clear the captions” is the other fail. You get a clean two-line block and a bottle with no name. The feed does not know the SKU. The PDP does, after the click you may not get. Move type. Leave the pack.

Generated I2V already wants to melt letters. If you then cover what survived, you spent a product row for a talking background. Lock the still until the label is sharp. Prompt motion around the pack. Then place captions in the hole that is not the pack and not the chrome.

Two keep-outs, one 9:16 hole

Vertical feeds share a ratio and do not share a UI. Bottom ~200px on a 1080×1920 export is a working keep-out for furniture, not a contract. Shop tabs, paid CTAs, and YouTube Shorts’ 27 May 2026 on-video AI label all grow the band. Recheck.

The pack often sits in that same lower third because that is how UGC hands hold a bottle. Hands enter from the bottom. The label faces camera in the lower half. A default center-bottom caption lands on the logo. That default is a collision, not a spec.

Where the words can go:

Mid-frame, above the pack. Two lines, not three. Three lines will still hit chrome or chin.
Top variant. Several Studio families offer top. VEED position can sit higher. Top chrome exists (status, search). Do not slam a title into the status bar. Top is a band, not the pixel at y=0.
Brief hold, then gone. If the pack is the last two seconds, let captions yield. The hold is the SKU. A caption that stays through the pack-shot is a caption that covers the SKU.

Word-highlight scales the live token. Scale plus a low baseline hits labels and buttons. Park karaoke higher than a static block, or shorten the cue. DYNAMIC 2× does not earn a seat on the carton.

Name supers, legal rows, and CTA cards are also type. Two-line speech plus a disclaimer plus a pack is three jobs in one band. Sequence them. The legal line can sit above the chrome for a hold. Speech can sit mid-frame. The pack keeps the lower picture. If you cannot fit all three, drop a graphic, not the label.

Do not shrink the font until the block fits on the logo. At phone size the paragraph is unreadable and the SKU is still covered. Split the cue. Move it.

Move type, do not crop the label

Cropping feels like a layout win in the editor. On a product ad it is a SKU miss.

A tighter crop that loses the wordmark, the shade, or the window in the pouch is a different product. Buyers match boxes. Marketplace ads that cannot show the listing image’s label get returns and complaints. Generated packs that already softened the type cannot afford a crop on top.

Fixes, in order:

Reposition the caption: up, or a top variant, or a shorter cue.
Reframe the picture so the pack sits where captions will not — often a little lower, with space mid-frame — without cutting the label.
Cut to a locked pack-shot still for the last beat with no speech captions, only the carton. Overlay a price if you must, not a transcript.
Recrop only if the label remains fully readable at 9:16 on a phone.

Burned-in captions will not restack when you notice the cover in comments. Placement is a pre-burn decision. Sidecar SRT cannot save a Reel. The feed shows pixels.

Contrast is a sibling issue, not a substitute. Outline or a backdrop so the caption holds on a busy plate. A high-contrast block on the logo is still on the logo. WCAG contrast for burned-in type keeps the letters visible. This page keeps them off the proof.

BASIC 1× two-line blocks are easier to park than DYNAMIC pops. Product ads usually want BASIC anyway. If the organic twin uses karaoke, that twin still cannot sit on the SKU.

QC the pack, then the chrome, then burn

Lock picture. If the label is already melted, regenerate or cut to the still you own. Do not caption a bad pack and hope type compensates.

Transcribe. Fix names and numbers to the PDP. Then place:

Draw a keep-out on the label (the actual letters on the carton, not “the bottle generally”).
Draw a keep-out on bottom chrome and the Shorts overlay zone when that label applies.
Put two lines of speech in the hole that remains.
If no hole remains, you have too many graphics. Remove one.

The AI caption generator and add captions to a video will put type where the preset says. 165 codes, 21 BASIC, 9 DYNAMIC, 45 Studio CSS presets. None of those lists knows where your carton is. You do.

Watch the export on mute, on a phone, in the app’s draft player. You are looking for a covered wordmark, not for vibes. If the brand name on the box is unreadable when the cue is up, the caption is in the wrong place. Move it. Burn again.

Captions should not cover the SKU. Move the type. Leave the label.

Place the next pack-shot burn in Versely.

FAQ

Can I crop tighter so captions fit at the bottom?

Only if the full label stays readable at 9:16 on a phone. If the crop eats the wordmark or the shade, you moved the wrong object. Move the caption.

Do top captions always solve a pack in the lower third?

Often. Check top chrome (status, search, username on some surfaces) and faces. Top is a band. It is not a dump. Preview in the app, muted.

Should the last pack-shot hold still have speech captions?

Usually no. Let the carton speak. If you need a price, author a small overlay that does not sit on the letters. A transcript on the logo is a cover.