X video editor: green screen, captions, and the trap

#Quick answer

The X video editor lives in the iPhone app. Open the composer, add a clip, and you get a trim bar, green screen recording that uses an X post or a camera roll photo as your background, text and emoji overlays, and auto-generated captions you can now edit by hand or hand to Grok to correct or translate. The 4 October 2026 rebuild was almost entirely about those captions. None of it has been announced for Android, and the only way to attach a real closed-caption file is still a .srt upload on the web.

Last updated: October 2026

#TL;DR

X built a video editor in three moves over four months, and the coverage all froze at move one. Search for it and you get a wall of July launch posts listing green screen and multilingual captions, which was accurate on 7 July and has been incomplete since 16 August. The thing worth knowing is not in any of them: the October rebuild made the captions editable, the caption tooling is split across two platforms that do not share it, and the overlay tool sits inside a rule that can cost you a payout if you point it at someone else's footage.

#What you actually get, and when each piece landed

Three dated releases, all on the iPhone app.

Underneath all of that sits the control that has been there for years and still does most of the work: trim. X's video help page documents it for both the iPhone and Android apps. Drag either side of the bar at the bottom of the preview, tap Trim, then preview with the play button before you post. Nothing fancy, available everywhere, and it is the only editing step most posts need.

Why X bothered is on the record. Nikita Bier, then X's head of product, framed the July launch as giving creators "the tools to create original content & reward those creators." Read that next to what X was doing to its payout program the same month and the strategy is obvious enough. Hold that thought, because it comes back below as a rule you can trip over.

#The October rebuild is about captions, not effects

If you went looking for new filters on 4 October, you found a layout change and left disappointed. The substance is in the caption flow.

Before, the captions the editor generated were what you got. Now they arrive as editable text in the composer, and you have three ways to deal with them: fix them yourself, ask Grok to correct them, or ask Grok to translate them into another language. Allegra Jacchia, the X engineer who announced both this and the August overlays, described the result as cleaner, more intuitive and much easier to use.

Why that matters more than it sounds is in X's own documentation. Its video help page says the plainest possible thing about where auto captions come from: "Speech-to-text is generated using Microsoft Azure Cognitive Services." It is a general-purpose transcription service. Hand it a product name, a handle, a bit of jargon or an accent it has not seen much of and it will produce something confident and wrong.

X has said as much elsewhere for years. Its caption file guide warns that automatic captions miss some audio outright and mistranslate other parts of it, that "sometimes the mistakes are neutral or funny, but sometimes they're offensive," and that the best way to get quality captioning is to write the file yourself. Until October, the app gave you no way to act on that advice. Now it does, at least for overlays.

#Green screen, and where the background comes from

Green screen replaces what is behind you while you record. The interesting part is the source: X lets you pick either an existing post on X or a photo from your camera roll.

That second option is the generic one. The first is the one built for this platform. You record yourself reacting in front of the actual post, which means the thing you are talking about is visible in frame instead of quoted above it. If you have been doing that with a screenshot and a quote post, this is a shorter path to the same result, and it keeps the reply in the conversation rather than spawning a second one. Our guide to quoting a post on X covers why that distinction changes who sees it.

#X has three caption systems and they do not share anything

This is where people get stuck, and it is not their fault. There are three different things on X called captions.

Auto captions. Generated by X from your audio, viewer-side. X's help page says they are shown automatically on iOS and Android when a video plays in the timeline, and that to see them in full screen you have to enable captions at the system level: Closed Captions on iOS, Captions on Android. Not every video gets them.

An uploaded caption file. The real accessibility route. Upload your video, select Upload caption file (.srt) below it, choose the file, select Done, and the language name replaces the upload prompt. X's page ends with a one-line note that decides the whole thing: "Currently this feature is available on web only." Worth a footnote: that page says SRT stands for Secure Reliable Transport, which is a video transport protocol. The subtitle format is SubRip. Your file still works.

Burned-in overlays. What the iOS editor makes. These are rendered into the video as part of the picture, so they show for everyone regardless of settings, in any language you translate them into, styled how you like. They are also not machine-readable, so they do nothing for a screen reader.

So the device that can style and translate captions cannot attach a caption file, and the surface that can attach a caption file cannot style or translate anything. Pick based on who you are posting for. If accessibility is the point, use the web and an .srt file, and give the video a description the same way you would add alt text on X. If reach on a muted autoplaying timeline is the point, burn the overlay in on your phone. We go deeper on the viewer side in X video captions.

#The overlay trap: a text overlay is not an edit

Here is the part no launch post mentions, and it is the one with money attached.

X retired Creator Revenue Sharing on 7 September 2026 and replaced it with Original Content Rewards, whose terms took effect on 7 August 2026. That program only pays on original content, and X's help page for it spells out what fails. Content copied from other creators or re-uploaded from another platform fails. Aggregated compilations fail. And so does minimally modified content, which X defines as a reproduction carrying only slight changes, naming a changed word or two, a filter, a speed adjustment, or a text overlay that does not meaningfully transform the original.

Read that against what shipped on 16 August. The tool X gave you is a text overlay. Used on your own footage it is an edit. Used on a clip you pulled from somewhere else, it is the exact example X's own policy lists as not qualifying.

What does qualify, by the same page, is adding genuine perspective: reacting to something, analysing it, breaking down a news story, adding context other people do not have. Which is a reasonable description of what green screen plus your own voice produces, and a poor description of a reposted clip with a caption slapped on it. The editor can get you to either outcome. The policy pays for one. If you are planning around those payouts, start with X monetization requirements and the application flow in X Creator Studio.

#There is still no music

Worth saying plainly, because it is the single most common question about every short-video editor and the answer here is unusual.

No dated announcement for the editor, the overlays or the composer rebuild mentions an audio track, a sound library or licensed music, and X's video help pages do not document one. There is no X equivalent of picking a trending sound. If you want music under a clip, mix it in before you upload.

Two consequences. Your audio is final once it is uploaded, so get it right in your own editor first. And the music is your problem: anything you use is subject to the same takedown process as any other media, which is worth understanding before you post rather than after, via how an X copyright claim works.

#Where this fits in a posting routine

This removes a real step. Recording, trimming, captioning and posting now happen in one app on iOS, and the October caption editing means you no longer have to choose between fast and accurate. For a reaction clip it is genuinely the shortest path available.

What it does not touch is everything around the video. The caption text still has the X character limit applied to it, the file still has to clear the limits in our X video length guide, and a good clip posted into a dead window still underperforms a mediocre one posted into a live one. The one number worth planning around is timing: engagement on X runs a twin-peak daily curve, and video is no exception.

The other half is the conversation after you post. Video gets you attention; replies are what turn that attention into followers, and that is the part that does not compress into an app feature. If you would rather not spend your evenings on it, X-Autopilot runs that reply loop on your own Mac, on your own connection, with the session cookie never leaving your machine. That is a smaller detection surface than a shared cloud IP pool and a delegated login, not an absence of risk, and we set out the whole trade-off in is X-Autopilot safe.

#Bottom line

X now has a real video editor, and it is an iPhone product: green screen from a post or a photo, text and emoji overlays, and since 4 October 2026, captions you can correct yourself or hand to Grok to fix and translate. The trim control is the only piece Android has documented. Closed-caption files are still web only. There is no music. And the overlay tool, pointed at footage that is not yours, lands you in the one category X's payout rules explicitly refuse to pay for.