A soft subtitle is usually a separate timed-text track that a compatible player or editor can manage without changing the video image. Hardcoded or burned-in subtitles have been rendered into the picture, so the best first option is a clean source export; if that is unavailable, visual cleanup is a separate task. Do not confuse this with open versus closed captions: those terms describe whether viewers can turn text on or off, not by themselves where the text is stored.
Published · Updated
People often use captions, subtitles, open, closed, hardcoded, burned-in, and soft as though they all describe the same thing. They do not. One group describes what the text is for, another describes whether viewers can control its display, and a third describes how it is delivered or rendered. Sorting those meanings out is the fastest way to avoid trying to repair pixels when a track can simply be switched off.
In accessibility guidance, captions normally include spoken dialogue plus important sound information, while subtitles are often used for language understanding or translation. In everyday publishing, the words can overlap and regional usage varies. For an editor deciding what to do next, the more important question is whether the text remains a separate timed track or has already been rendered into the visible picture.
| Term pair | What it describes | What it does not prove |
|---|---|---|
| Captions vs. subtitles | The role of the text for the viewer—often same-language accessibility information versus language translation or comprehension | Whether the text is a track, a sidecar file, or pixels in the exported image. |
| Open vs. closed | Viewer control: open text is always shown; closed text can be shown or hidden in a compatible player | Whether always-visible text has definitely been burned into the video pixels. |
| Soft / track vs. hardcoded / burned-in | How the text is delivered: a separately managed timed track versus text rendered into the image | Whether every player will expose the same controls or support the same subtitle format. |
| Sidecar vs. embedded track | Where a separate track lives: in a companion file such as SRT or VTT, or as a distinct stream inside a media container | That a missing sidecar file means the words are necessarily hardcoded. |
Start with the player, then verify in the source workflow if you can. On YouTube, for example, a visible CC control lets a viewer turn captions on or off; if turning it off removes the words, that player was displaying a separate caption track. The same principle applies in any editor or player that lets you select a subtitle or caption track independently of the video layer.
Do not draw the opposite conclusion too quickly. No CC button can mean there is no accessible track, but it can also mean that the player, container, account setting, or platform is not exposing it. A screenshot only shows what was rendered at that moment; it cannot by itself tell you whether the text originated from a track or was baked into the file. When the stakes are higher than a quick viewing decision, inspect the original timeline or ask the source owner for the clean master and any caption files.
A separate timed-text track is the least destructive case. It can be a sidecar resource supplied next to the video, such as an SRT or VTT file, or an independent track packaged with the media. Web video commonly associates text tracks with a video through a track element, and a single video can offer several caption or subtitle tracks in different languages. The picture itself does not need to change when the viewer switches or hides one.
Use the tool that owns the track. On a video platform, turn off, replace, or manage the subtitles there. In an editor, remove the relevant caption or subtitle track from the export while keeping a separate accessible version for the next delivery. On YouTube Studio, subtitle and caption files contain timed text and can be added or managed separately from the video, which illustrates why this is a metadata and publishing task rather than an image-cleanup task.
| Where you find it | Recommended action | Result to verify |
|---|---|---|
| Player caption menu | Turn the selected track off or select another language | The words disappear while the video image remains unchanged. |
| Editing timeline caption track | Edit, hide, or exclude that track in a new export | The video layer remains intact and an accessible track is retained where needed. |
| Publishing platform subtitle manager | Replace or remove the timed text resource for the authorized version | Timing, language, and viewer controls behave as intended on the target platform. |
| Sidecar SRT or VTT file | Keep, revise, or omit the track according to the new delivery | The video still plays normally and the correct language is available when the track is attached. |
Hardcoded or burned-in subtitles are part of the rendered image. This is common when a social clip, screen recording, or final export has already combined the captions with the footage. The first choice is still to recover the project that made the video or request a clean master. A new export without the text layer preserves every underlying pixel that was originally available.
When no editable source exists and you own the footage or have permission to edit it, the task becomes visual cleanup. A lower subtitle band over a plain, stable background is usually easier to assess than text crossing a speaker's face, a product, hands, hair, fast motion, reflections, or a scene cut. Cropping can hide the band but also changes composition; covering it can break as the background moves. Review the entire affected range, not a single frame.
Separate tracks are easier to revise for language, terminology, timing, and viewer preference. They also let a publisher offer more than one language without permanently adding every version to the image. Burned-in text can be intentional—for example, a social-video style that must be visible with sound off—but it makes later corrections and localization more destructive.
That does not make one format universally better. A track depends on the target player supporting and exposing it. Open text is useful when the message must be seen regardless of the player setting. The important production decision is to preserve an editable caption source whenever you can, even if you also deliver a version with visible on-screen text.
A YouTube viewer clicks CC and the words disappear: the platform was showing a separate caption track, so no image repair is needed. A team has an MP4 plus an SRT or VTT resource: the timed text can be revised or omitted from a new delivery, while the video remains intact. A creator only has a vertical export with styled captions already visible over every scene: the source may be a burned-in version, so the team should first look for the original project or a clean master.
These examples are deliberately about the next action, not labels alone. A file can be delivered through several routes, and platforms differ in what controls they expose. The safe workflow is to confirm the actual source and output you have before you remove anything.
If you have confirmed that the visible words are inside the picture, no clean source exists, and you are authorized to edit the clip, use a workflow designed for hardcoded subtitles. RemoveText.video can help prepare a clean master by selecting the subtitle area and reconstructing the affected frames while preserving the original framing. It does not remove a subtitle track, add new subtitles, or translate a video.
Do not use visual cleanup to remove ownership, attribution, provenance, safety, or legally required disclosure marks. And do not promise an invisible result before review: text that hides moving people or detailed image content may leave a repair that is unsuitable for publication.
Before
After
Not necessarily. Open describes text that is always shown to the viewer; hardcoded or burned-in describes text rendered into the picture. They often appear together, but a label alone is not enough to choose a repair method.
No. A screenshot only proves that the words were visible at that moment. Check a player control, the original timeline, or the available caption resources before deciding whether the text is a separate track or part of the picture.
It may mean no track is available, but it can also reflect the player, file format, platform, or current playback context. Treat it as a reason to inspect the source, not proof that the subtitles are burned in.
You can prepare a cleaner master and then add a new subtitle track or a newly designed caption layer in your editing workflow. The old text should be removed first so the two versions do not overlap.
No. A separate subtitle or caption track should be turned off, replaced, or excluded in the player, publishing system, or editor that manages it. RemoveText.video is for eligible text that is already visible inside authorized video frames.