One cut, nine native edits: how we re-shape a single story
The fastest way to look like a brand that does not understand a platform is to post the same crop everywhere. We start from one finished story, then build nine separate edits, each cut to fit the place it lands. Not nine reposts. Nine different files, made from one source, in about the time it used to take to format two.
This post is about the second half of that sentence: the method, the tools, and where it breaks.
What you'll learn
- Why "post it everywhere" quietly loses to "fit it everywhere"
- The one-source, nine-edit pipeline we run on our own content
- The exact tools and aspect ratios behind each native cut
- The trap of treating captions as an afterthought
- What this makes possible that a human formatter cannot match
The old way: one export, copied around
For years the standard move was to render a single hero video, usually 16:9 because that is what the editor opened by default, and push it to every account. TikTok got a letterboxed widescreen with two black bars eating half the screen. LinkedIn got a caption written for Twitter. The thumbnail was whatever frame the platform grabbed.
It worked in the sense that something went out. It failed in the sense that each platform's own ranking system reads native formatting as a signal, and a letterboxed video on a vertical feed reads as imported. The reach you lose is invisible, so nobody fights for it.
The deeper problem is that "platform-native" is not one rule. It is nine sets of rules that change a few times a year. A team that formats by hand either ignores most of them or burns a person on reformatting full-time. Both are bad.
The new way: one source of truth, nine derived edits
We treat the finished story as a single source asset with a manifest. The manifest is a small file that says, in plain terms, what the story is: the master video, the key frames, the spoken transcript with timestamps, the core claim, and the call to action. Everything downstream reads from that one file, so when the story changes, every edit changes with it.
From that manifest we generate nine edits. Here is the method, channel by channel.
Vertical, fast, sound-on
TikTok, Instagram Reels, and YouTube Shorts share a shape, 9:16, but not a temperament. We cut all three vertical, then change the open.
- We use **FFmpeg** to reframe 16:9 to 9:16 with a center-weighted crop, then a `scale` and `pad` pass so nothing important leaves the safe zone.
- The hook differs per channel. Shorts rewards a question in the first second. Reels rewards motion. TikTok rewards a face or a hand entering frame. Same story, three different first two seconds, pulled from different timestamps in the same master.
- Captions are burned in, because most plays are muted. We pull the transcript, align it with **Whisper** word timestamps, and render karaoke-style captions in the brand font so the words land on the beat.
Square and feed-first
Instagram's main feed and Facebook still favor a 1:1 or 4:5 frame. We render a 4:5 edit, slightly tighter than the vertical, with the caption doing more work because these feeds are read as much as watched. The first line of the caption is the hook here, not the video.
Wide and patient
YouTube long-form and LinkedIn keep the 16:9 master, but the framing of the post differs hard. YouTube gets a title built for search and a custom thumbnail rendered from a key frame, not an auto-grab. LinkedIn gets a plain-text open that states the lesson before the video, because that feed rewards the comment, not the click.
The two that are not video
X (Twitter) often performs better as a short thread with a clipped 30-second cut than as the full piece. Pinterest wants a tall still with text baked in, linking back. Both are generated from the same manifest: the thread from the transcript, the Pin from a key frame plus the core claim.
That is nine: TikTok, Reels, Shorts, Instagram feed, Facebook, YouTube, LinkedIn, X, Pinterest. One story underneath all of them.
A worked example, start to finish
Take a single piece we ran: a 90-second explainer on how we tag consent at signup. One master video, 16:9, one narration track. Here is what the nine edits became, concretely.
- The master stayed wide for YouTube, with a search-shaped title and a thumbnail rendered from the frame where the diagram appears.
- The three vertical cuts each opened on a different beat. Shorts opened on the question "who owns your email list." Reels opened on the diagram animating in. TikTok opened on a hand pointing at the screen, pulled from a later timestamp and moved to the front.
- The 4:5 feed edit led with a written first line, because that surface is read before it is watched, so the caption carried the hook the video would have carried elsewhere.
- LinkedIn got the flat lesson up top, "your signup form is a data asset or a liability, here is the difference," with the video below as proof.
- X became a four-tweet thread built from the transcript, with a 30-second clip in the first tweet.
- Pinterest got a tall still of the diagram with the core claim baked in as text, linking back.
Same 90 seconds of source. Nine genuinely different arrivals. The reframing and caption burn ran unattended; the per-channel hooks took one person about twelve minutes to review and adjust two of them.
The aspect-ratio map, written down
The reason this scales is that the formatting rules live in a file, not in a person's head. We keep a small table that says, per channel, the aspect ratio, the safe-zone margins, whether captions burn in, and where the hook comes from. When a platform changes a spec, we change one row, and every future edit inherits it.
- 9:16 with a 14 percent top and bottom safe margin: TikTok, Reels, Shorts.
- 4:5 with the caption as the hook: Instagram feed, Facebook.
- 16:9 with a custom thumbnail: YouTube. 16:9 with a text-first open: LinkedIn.
- Text-derived: X thread. Still-derived: Pinterest.
This table is the actual product. The videos are downstream of it. Most teams keep these rules as folklore, which is why their formatting drifts every time a different person posts.
Where the time actually goes
People assume the editing is the cost. It is not. The reframing, the caption burn, the thumbnail render are scripted, so they run while we do something else. The cost is the decision layer: which hook per channel, which 30 seconds become the X clip, which frame becomes the Pin.
So we put the machine on the mechanical part and keep the judgment for ourselves. The pipeline produces nine draft edits. A person looks at the three hooks and the thumbnail and changes maybe two of them. That review is ten minutes. The rendering is unattended.
We know this because we run it on Temerarii Media's own posts, every day. When we changed from one shared export to nine derived edits, the work did not get bigger. The repetitive part got automated and the choosing part got faster, because we were choosing from drafts instead of from a blank timeline.
The caption is half the post
A blunt thing we learned on ourselves: on muted feeds, the caption is not metadata, it is the video. If the first three burned-in words do not say something true and specific, the play ends before the story starts.
So we write captions per channel, not per story. The vertical caption is a spoken-rhythm hook. The feed caption is a written lede. The LinkedIn caption is a flat statement of the lesson. Same idea, three registers, because the reader on each surface arrived in a different mood.
What this makes possible
Once one story reliably becomes nine native edits, two things change.
First, you can publish more without making more. The bottleneck moves from production to having something worth saying, which is the right place for the bottleneck to live.
Second, you can run honest tests. Because all nine edits trace back to one manifest, you can compare them on equal footing. Same story, nine frames, nine hooks. The platform tells you which framing it wanted. You stop guessing about reach and start reading it.
The point was never to flood nine feeds. It was to respect that each feed is a different room, and to walk into each one dressed for it, without hiring nine people to pick the outfits.
If you want to watch this run on real content, the experiments at [temerarii.xyz](https://temerarii.xyz) show the machine working in public, and the rest of this week's posts in the [blog](/blog/) take the other pieces apart the same way.
Written by Dominique Davis, Founder, Temerarii Media.