Burn Captions for a Silent Feed
Burn the captions into the picture. The same file can go to other apps and still read with the sound off.
Burn the captions into the file so the same reel can play in a silent feed on other apps. Jun speaks one line for Hearth Bakery, and that line is already in the picture, upper-middle, before you upload. TikTok and Shorts are often watched without sound. Burned-in captions still show the words when the file moves. One person speaks. A realistic face or voice is labeled as AI. Leave the platform label in place. The bottom and the right stay clear so the app buttons do not cover the burned-in line. If the words can fall off the file, they were never in the picture. Leave the upper-middle empty in the still, burn the line that they close at four into that band, export one vertical file, and label every upload. The first three seconds show those burned-in words and Jun's face while he says they close at four. A caption that exists only in the upload form, or words on the bottom and the right, is rejected before posting. One person speaks for fifteen to forty-five seconds, and the post says the face and voice are AI.
Put the words in the pixels
A caption that exists only in the upload form disappears when you download the file or send it to another app. Burn the line into the upper-middle of Jun's frame. The words match what he says: they close at four. Pick a caption look you can read on a phone. The safe zone is upper-middle, away from the bottom bar and the right-hand buttons. Burn the closing line into the upper-middle, match it to the spoken hour, and read it on a phone before upload. The first three seconds show the hour in the pixels while Jun says they close at four. A caption that disappears when the file moves to another app is the version you reject.
Same file, more than one app
Export once, vertical, and that file is the one you post. Burned-in captions help on TikTok and Shorts because the sound may be off and the app will not reuse another app's subtitle track. You still label a realistic face or voice as AI on each upload. The disclosure is not a substitute for the burned-in offer, and the offer is not a substitute for the disclosure. Export once, then post that same vertical file where silent autoplay is normal, and label each upload after the words are in the pixels. The opening three seconds show the burned-in offer while he says they close at four. A download that lost the words, or an upload that skips the AI line, is rejected before posting.
One speaker, one line
Only Jun is in the frame, so the caption has one voice to follow. A second line of chatter makes the burned-in text late. The first three seconds carry the point, and the burned-in words are visible in those seconds. Approve the still with empty upper-middle space before you add the type, so the face is not fighting the letters. One voice, so the burned-in text has one line to follow. Approve empty upper-middle space before the type goes in. The first three seconds show the words clear of the face while Jun says they close at four. A second line of chatter, or letters fighting the eyes, is the frame you refuse.
A worked example
Jun at Hearth Bakery says they close at four. That sentence is burned into the upper-middle of the vertical file. The same file is what he can post where a silent autoplay is normal, including TikTok and Shorts. The bottom and right stay clear. The realistic face and voice are labeled as AI. The platform label stays on each upload. Empty upper-middle, burned-in hour, one vertical file, then the label. The first three seconds show the sentence in the picture while Jun says they close at four. A caption that lived only in the upload box is rejected before posting. The post says the face and voice are AI.
Before you post
- Captions are burned into the picture.
- The words sit upper-middle.
- The realistic host is labeled as AI.
Where Scro fits
Scro keeps one person speaking and offers 13 caption styles that can be burned into the frame. Export at 9:16. Label a realistic face or voice as AI, and leave the platform disclosure on the post. Burn one of 13 caption styles into the upper-middle, keep one speaker, and export after the words survive a download. The first three seconds show the burned-in line while Jun says they close at four. Words under the reel buttons are rejected before posting. The caption says the face and voice are AI.
Burn the captions in. Finished samples are on the examples page.
Related: 15 to 45 Seconds for the Reel.