When Instagram began testing "Features" — long-form horizontal video up to about 90 minutes — on October 26, 2026, almost every take treated it as a new place to post. The change that lands earlier in the workflow, and that almost nobody is talking about, is on the production side: after a decade of filming vertical, phone-first, and fast, creators now have to shoot for a wide, lean-back screen that someone watches for twenty or forty minutes, and shoot it so the same session can also be cut into the vertical clips the format links back to it. That is a different craft — wider framing composed with the vertical crop in mind, audio that survives long runtimes, a steadier set, and clip-aware filming that leaves clean, self-contained moments in the recording. This guide is about the input side of the format shift: how to actually produce a horizontal long-form piece that works both as a Feature and as a clip source, which vertical-era habits to unlearn, where the extra production cost is and isn't worth it, and how to hit the cadence the format rewards without staging a 90-minute shoot every week.
When Instagram began testing "Features" on October 26, 2026 — long-form, horizontal video that can run up to about 90 minutes, aimed at podcasts, interviews, and comedy — nearly every take framed it the same way: Instagram now has a place to post long video. True, and it is the half of the story that changes the least about your day. The change that lands earlier, and that almost nobody is talking about, is on the production side. For ten years Instagram trained creators to film one way — vertical, close, phone-first, hook in the first second — and the format it is now building rewards close to the opposite. The thing to solve is not where the file goes; it is how you shoot it.
This guide is deliberately about that input side. It is not the strategic question of whether to go long at all — when a Feature beats a Reel, how the reach and monetization math works — which is worked through in Instagram long-form video strategy. Nor is it the output question of what to do with the finished recording, which the companion guide on repurposing long-form horizontal Instagram video covers in full. This page sits between them, on the part most coverage skips: how you actually produce a horizontal long-form piece that works both as a lean-back watch and as the source of the clips that promote it.
The facts first, because much of the format is still a limited test and it is easy to overcommit. On October 26, 2026, Instagram's Adam Mosseri and product VP Tessa Lyons introduced "Features," a long-form, horizontal video format reported at lengths up to roughly 90 minutes, aimed at the podcasts, interviews, and comedy that have lived on YouTube. Mosseri — who noted he had long been against long-form on Instagram — framed it as "symbiotic" with short-form rather than a replacement for it. The test began with creators in Brazil and is expanding to a small group in the US, so most accounts cannot post a true Feature yet, and the 90-minute figure is described as one of several limits being trialed, so treat it as provisional. The format is built for a lean-back screen, which is why the Instagram TV app and Apple TV support are part of the same push. (Instagram has tried long-form before — IGTV launched in 2018 and was shut down in 2022 — so the commitment is worth watching, not assuming.)
One detail does more to reshape production than the length number: Features is horizontal and clip-linked at the same time. Under the test you can link short vertical Reels — reportedly up to eight — to a single long video. That means the master you film has two jobs at once. It has to hold up as a widescreen piece someone watches for minutes on a TV-class screen, and it has to survive being cut into vertical clips that reach new people and point back to it. You are no longer filming a Reel. You are filming a horizontal session that must also be a good clip source — and those two goals pull your framing, audio, and pacing in directions the vertical era never asked about.
A decade of Reels, Stories, and TikTok built a specific set of reflexes, and most of them actively hurt a horizontal long-form piece. The instinct to frame tight and vertical, filling a 9:16 phone screen with a single face, leaves you with nothing to work with in a wide 16:9 composition. The willingness to live with phone-mic audio, fine for a clip that is over in seconds, becomes a liability across forty minutes. The jump-cut pacing that keeps a short Reel moving reads as frantic in a format people chose precisely because they wanted to sit with something. Even the hook reflex changes: a Feature still needs a reason to stay, but a lean-back viewer who opted into a 30-minute conversation is not the thumb-stopping stranger a Reel is built for.
The reframe is to produce for the way the content is consumed. A Feature is watched leaning back, often on a bigger screen, by someone who already decided to give you their time. That rewards composition, sound, and structure over speed and compression. None of this means abandoning short-form craft — you still need it for the clips you cut from the master. It means holding two production modes in your head at once, which is the actual new skill the format demands.
The single most useful production habit for this format is composing the wide shot so it can still be cropped to vertical later. A straight center-crop of a 16:9 recording is how a repurposed clip ends up with a guest sliced off the edge or a speaker reduced to a dot in a sea of empty floor. You avoid that at the shoot, not in the edit: keep whoever is speaking inside the center vertical "safe" zone of the frame, stage two-shots so each person is individually croppable rather than spread to the edges, and leave headroom and margin so a 9:16 reframe has somewhere to sit. Film wide for the Feature, but never forget that a vertical clip has to come out of the same pixels.
If you change one thing about how you produce, change the sound. A 40-second Reel forgives bad audio because it is gone before it grates; a 40-minute horizontal piece does not — poor sound is the fastest way to end a lean-back watch, and it is far more noticeable on the TV-class screens this format is built for. A dedicated or lavalier mic per speaker and a reasonably treated room matter more here than a better camera. This is the least glamorous part of the shift and the one with the highest return: viewers tolerate imperfect video far longer than they tolerate audio they have to strain through.
Beyond framing and sound, a long horizontal piece rewards a steadier, better-lit, more structured production than a Reel. Shots should hold rather than jump; light should be good enough to survive a big screen; and the piece needs a shape — an open, a through-line, segments — so a viewer can orient and stay. Slower pacing is allowed, but slow is not the same as shapeless, and the most common way a long piece fails is not that it is too long but that it has no structure to carry the length. Produce the session as a thing with a beginning, a middle, and an end, not as unbounded footage you will fix later.
The link-eight-Reels mechanic is Instagram telling you, in product terms, that the intended unit is one long piece plus the clips cut from it. The highest-leverage thing you can do at the shoot is make that extraction easy by filming so that self-contained moments already exist in the recording. That means asking one clean question at a time rather than stacking three, letting a subject land a complete thought before you cut in, leaving a half-second of silence around a strong line so a clip has clean in and out points, and jotting the timestamps of the moments that land while you still remember them. A clip-friendly master is produced at capture; it cannot be fully rescued in the edit.
The alternative is the failure mode most creators back into: finishing a 90-minute recording, then scrubbing the timeline hoping clean, standalone clips are buried in there somewhere. Sometimes they are; often the best moments got stepped on, or trail off, or depend on ten minutes of context that will not fit in a vertical clip. The craft of actually pulling and reframing those moments — selection, the 16:9-to-9:16 recompose, captions — is its own discipline, covered in AI clips from long-form content and in the repurposing guide. But that work is dramatically easier when the master was filmed to be clipped in the first place. See the glossary on long-form video and horizontal video for the format definitions underneath all of this.
Honesty keeps this useful, because a horizontal long-form shoot is a real production event, not a quick clip, and three limits are worth stating. First, do not make a long piece just because the format exists: if your idea lands in five minutes, filming ninety to fill the container trades a strong short piece for a weak long one, and padding is the surest way to tank retention. The length has to be earned by content that genuinely has depth to carry — a real conversation, a teardown, a story that needs room. Second, the monetization case is still unsettled: reporting around the launch indicated Instagram was not broadly sharing ad revenue the way YouTube does, beyond a few bespoke early deals, so produce for reach and audience depth rather than a platform payout. Third, most accounts cannot post a true Feature yet — it is an invite-limited test — so you can build the production habit now, but the surface itself is not universally available.
Even a perfectly produced master is one week's input, and the format does not reward one-off pieces — it rewards a steady drop that a returning audience can rely on. That is the production trap. A 90-minute horizontal shoot, done with the framing, audio, and clip-awareness above, is an event you cannot realistically stage every week, and the weeks you skip are the weeks the cadence breaks. The craft of shooting well solves the quality problem; it does not solve the supply problem, and supply is what quietly kills most long-form ambitions.
Kompozy is a content generation and multi-platform publishing engine, and for this format the angle that matters is the production input itself, not just the post-shoot cleanup. For the weeks you cannot film, its Persona HeyGen format stands up longer-form, avatar-led video with a consistent voice and face — horizontal-destined long content produced without a camera day — and Persona Shorts fill the vertical cadence between shoots, so your long-and-short rhythm is not hostage to whether you could get everyone in a room this week. Everything is governed by one Persona Brief, which is what keeps a generated piece and your genuinely filmed horizontal master reading as the same creator rather than two different voices.
Then, when you have filmed a real master, the same engine turns that single session into the full deliverable set so one shoot is not one post: Clipped Shorts cut the strongest moments into captioned vertical clips — the exact clip set the Features funnel asks you to link back to the full video — and the same source fans into a carousel of the key points, quote graphics, a blog recap, and a newsletter, every output on the same brief. Autopilot schedules and publishes the whole set across the eight social platforms plus blog and email from one queue, behind a per-post review gate so a human still signs off. Be precise about the boundary: filming the horizontal master, enrolling in the Features test, linking clips inside Instagram's app, and any TV placement all happen in Instagram's product — Kompozy removes the production ceiling, not the platform steps. For the step-by-step upload task see posting long-form video on Instagram, and for turning this into a weekly rhythm, scaling Instagram Reels production.
Instagram going horizontal is a shift in what and how you film, not just where you post. Shoot a wide 16:9 frame composed with the vertical crop already in mind, treat audio as the upgrade that actually matters, give the piece a structure that can carry its length, and film clip-aware so one session is both a Feature and the clip funnel that feeds it. Only go long when the content earns it, and go in for the reach and audience depth rather than a payout that may not come. The craft is learnable in a shoot or two. The genuinely hard part is sustaining the cadence a depth surface rewards — and that is an input problem, solved by generating some of the video and industrializing the output of the rest.
You film in a wide 16:9 frame, but compose it with the later vertical crop in mind — keep whoever is speaking inside the center vertical zone so a 9:16 clip does not decapitate them. Fix the audio first, because a 40-minute piece forgives bad sound far less than a 40-second Reel does. And film clip-aware: ask clean questions, let people finish thoughts, and leave beats so self-contained moments exist in the recording that can later be cut into vertical Reels.
The "Features" format tested from October 26, 2026 is specifically horizontal — a widescreen format built for a lean-back screen, which is why the Instagram TV app (reaching Apple TV alongside Fire TV, Google TV, and Samsung sets) is part of the same push. That is the deliberate break from the vertical Reel. Longer vertical Reels still exist separately, but the long-form surface Instagram is building, and the one that plays on the living-room screen, is landscape.
Filming the long session so that clean, standalone moments already exist in the recording, instead of hoping to find them later. In practice: ask one clear question at a time, let the subject land a complete thought before cutting in, leave a half-second of silence around strong lines so a clip has clean in and out points, and note the timestamps of good moments as you shoot. The clips are cut from the master, so a clip-friendly master is made at capture, not rescued in the edit.
You need better audio, mainly. A phone Reel can survive mediocre sound because it is over in seconds; a 30- or 60-minute horizontal piece watched on a TV-class screen cannot — bad audio ends a lean-back watch fast, so a dedicated or lavalier mic and a treated room matter more than the camera. Beyond that, steadier framing and decent light help, because a bigger screen shows flaws a phone feed hides. You do not need a studio; you need the session to hold up at length.
A Reel is vertical, close, phone-first, and built around a hook in the first second and tight pacing. Horizontal long-form inverts most of that: wide framing for a lean-back screen, room for a slower build, audio and structure that have to hold attention for minutes rather than seconds, and a shot composed so it can later be cropped to vertical. The Reel is a single deliverable; the horizontal master is a source that has to serve both the full watch and the clips cut from it.
The hard part is cadence: a depth surface rewards a steady drop, but a 90-minute horizontal shoot is a production event you cannot stage every week. Kompozy is a full AI content generation and multi-platform publishing engine — its Persona HeyGen format stands up longer-form avatar-led video with a consistent voice for the weeks you cannot film, and when you do film, Clipped Shorts cut the master into captioned vertical clips while the same source fans into carousels, a blog, and a newsletter on one Persona Brief. Autopilot then publishes the set across the eight social platforms plus blog and email.
Instagram's October 2026 "Features" test gives horizontal video up to about 90 minutes a home, and it reshapes production, not just posting. After a decade of vertical, phone-first filming, creators now shoot 16:9 for a lean-back screen — wider framing composed with the eventual vertical crop in mind, audio that holds up over long runtimes, and a steadier set. The key discipline is filming clip-aware, so one horizontal session is both the full Feature and the source of the vertical Reels linked back to it.
Get started → · ← All guides · Compare Kompozy vs other tools