How to turn a 360 panorama rendering into a video: flat 9:16 views, a slow pan, or six views for AI

You rendered 360 panoramas for a VR viewer or a client link, and now someone wants a vertical video for Instagram or the website. A panorama holds every direction at once, but uploaded as it is, it plays as a stretched strip; parts of it have to become ordinary flat views first. This guide explains what an equirectangular panorama contains, how to pick a field of view for a 9:16 frame, how to extract still views and a slow panning clip with one free command-line tool, how to pull six views for an AI film, and the pitfalls that make extracted views look soft or bent.

Key takeaways

• An equirectangular panorama is a 2:1 image covering 360° across and 180° top to bottom, so every degree of view gets only width ÷ 360 pixels.
• A 1080 × 1920 view with a 47.5° horizontal field of view covers about 76° vertically and needs a panorama roughly 7,700 pixels wide to stay sharp in the center.
• FFmpeg's v360 filter converts an equirectangular panorama into flat views and, driven by a time expression, into a slow vertical pan.
• Keep the pitch at 0° so vertical lines stay vertical, keep the field of view moderate so the edges don't stretch, and stay away from the floor and ceiling, where equirectangular images are most distorted.
• A panning clip from one panorama never moves through space; for a film that travels, extract views from two or three panoramas and send six of them to an image-to-video service.

How to turn a 360 panorama rendering into a video

To turn a 360 panorama rendering into a video, convert parts of the equirectangular image back into flat, rectilinear views. Extract still 1080 × 1920 views at a moderate field of view and edit them together, render a slow pan by rotating the view's yaw over time, or extract six views from different panoramas and give them to an image-to-video tool. All three start with the same step: choosing a field of view the panorama has enough pixels for.

What an equirectangular panorama actually contains

Many renderers save a 360 panorama as an equirectangular image: the full sphere unwrapped into a rectangle twice as wide as it is tall. Chief Architect's X18 Reference Manual (read as of October 2026) says its Export 360 Panorama tool keeps "the ratio between Width and Height" at "2 to 1," and recommends that you "position the camera as near as you can to the center of the scene before exporting." V-Ray for Rhino's Render Output page says its VR Spherical camera type "locks the Aspect Ratio to 2:1," while the VR Cubemap type locks it to 6:1, a different layout made of six square faces. Enscape's panorama export, per Chaos's export options overview (last updated February 16, 2026), comes as a Mono or a Stereo panorama at Low, Normal or High resolution. Use a mono panorama for video; a stereo one holds two views, one per eye.

The 2:1 shape sets the budget. A panorama 8,192 pixels wide spreads those pixels across 360°, about 22.8 pixels per degree. A phone screen showing a vertical view only shows a slice of that circle, so the slice has to carry enough pixels on its own.

Field of view: how much of the panorama a 9:16 view can use

A flat view's horizontal field of view decides how much it shows and how many panorama pixels feed each output pixel. For a 1080 × 1920 frame, the vertical field of view follows from the horizontal one, and so does the panorama width you need for roughly one panorama pixel per output pixel in the center of the frame:

These come from the geometry of a rectilinear view, not from any one renderer. They explain the most common complaint: a 4,096-pixel panorama viewed at 47.5° is enlarged almost twice in the center, so it looks soft no matter how sharp it looked in the VR viewer. A wider field of view needs fewer pixels, but straight-lens geometry stretches everything toward the edges, so a 75° view of a room makes furniture at the sides look pulled. For a typical architectural panorama, start at 47.5° to 55°, and render the panorama at the largest size your renderer offers if video is a likely use.

Extracting flat 9:16 views

You can do this in any tool that reframes 360 images. The method below uses FFmpeg, a free command-line program, and its v360 filter, which converts between projections; the commands were tested with FFmpeg 8.0.

One still view

ffmpeg -i pano.jpg -vf "v360=input=e:output=flat:yaw=30:pitch=0:h_fov=47.5:v_fov=76.1:w=1080:h=1920:interp=lanczos" -frames:v 1 view-01.png

input=e means equirectangular, output=flat means an ordinary rectilinear image, yaw turns left or right in degrees (from −180 to 180), pitch tilts up or down, h_fov and v_fov set the field of view from the table above, w and h set the output size, and interp=lanczos picks a sharper resampling than the default bilinear. Change yaw to face a different direction; 0 is the center of the panorama image. If your renderer saved a cube map instead, v360 also reads cube-map layouts such as c6x1 and c3x2, but the face order must match, so check one test frame.

Rules that keep lines straight

• Pitch 0. A level view keeps vertical lines vertical. Tilt it up or down and walls and columns converge, the same keystone effect as tilting a camera.
• Moderate field of view. Stay at or below about 55° unless the room is small and you need the width.
• Avoid the poles. The top and bottom rows of an equirectangular image are stretched across the full width, so views of the ceiling or floor come out soft.
• Mind the seam. The left and right edges of the panorama meet behind the camera; v360 wraps across it, but check that your renderer didn't leave a visible join there.

A panning video from one panorama

A pan is the same extraction with the yaw changing over time. Save this line as pan.txt:

0-8 [expr] v360 yaw 'lerp(-20,20,TI)';

Then run:

ffmpeg -loop 1 -framerate 30 -t 8 -i pano.jpg -vf "sendcmd=f=pan.txt,v360=input=e:output=flat:h_fov=47.5:v_fov=76.1:w=1080:h=1920:interp=lanczos,scale=out_range=tv,format=yuv420p" -c:v libx264 -crf 18 pan.mp4

That makes an 8-second, 1080 × 1920 H.264 clip at 30 fps that turns smoothly from 20° left to 20° right of the panorama's center. Shift both numbers to aim the pan elsewhere, and keep it slow: 40° over 8 seconds reads as calm; 120° over 4 seconds reads as a phone being swung around.

Know the limit before you build a film this way. Every frame comes from one camera position, so nothing moves relative to anything else: no parallax, no walking through a door, no approach. A pan works as one shot in an edit, or as a short loop on a website, but a 30-second film of pans from one point feels static.

Six views for an AI film

If you rendered panoramas in several rooms, you already have the raw material for a film that travels. Extract one or two flat views from each panorama, then choose six that work as a set: an exterior or approach, the main living space, a second room, a detail such as a stair or kitchen island, and a closing view. Six views from a single panorama share one viewpoint and make a weak set. The six roles a project video needs explains how to choose.

Arch2Video takes exactly six images from one project (JPG, PNG, HEIC, TIFF or WebP, up to 30 MB each; long side at least 512 px, 1,500 px or more for the sharpest scenes), so a 1080 × 1920 extraction qualifies as long as it's sharp in the first place. It returns a 1080 × 1920 vertical H.264 MP4 of about 30 seconds: six connected five-second AI-generated scenes, an original soundtrack and an optional studio logo. The AI director plans the scene order. Generation usually takes an hour or more, and you're emailed when it's ready. It costs USD 15 per video, a one-time card payment with no subscription; uploading is free and the price appears after the upload. There is no free trial and no free video.

The motion is an AI interpretation of each view, not a path through your model, so content past the frame edge is invented. Every film includes a small cast of original, fictional adults in at least one scene, and people in your renders are not animated. Review the result with an AI video quality checklist before posting, and present unbuilt work as concept work.

Common pitfalls and fixes

If you still have the 3D model, rendering flat 9:16 stills directly will always beat extracting them from a panorama; the Rhino and Chief Architect guides cover those exports, and rendering export settings for Instagram covers sizes and color.

Frequently asked questions

Can I post a 360 panorama rendering directly as an Instagram Reel?

Not usefully. A Reel is a flat 9:16 video, and an equirectangular panorama uploaded as a still or a video frame shows as a stretched 2:1 strip. Convert it first: extract flat 1080 × 1920 views at a moderate field of view, or render a slow pan from it, and post that.

What field of view should I use for a vertical view from a panorama?

Start with a horizontal field of view of 47.5°, which gives about 76° vertically in a 1080 × 1920 frame and looks like an ordinary lens. Go to 55° for a little more room, and avoid going much past 65°, where the edges visibly stretch. Narrower views need a larger panorama to stay sharp: about 7,700 pixels wide at 47.5°.

Why do my views from a 360 panorama look blurry?

Because the panorama spreads its pixels around a full circle, and your view uses only a slice of it. A 4,096-pixel-wide panorama has about 11 pixels per degree, so a 47.5° view at 1080 pixels wide is enlarged almost twice in the center. Re-render the panorama at a higher resolution, use a sharper resampling such as Lanczos, or accept a slightly wider field of view.

Can AI turn one 360 panorama into a walkthrough video?

Not a true walkthrough. A panorama is one camera position, so a pan from it only rotates; an image-to-video tool given views from it can invent motion, but not a measured route through the plan. For a film that moves from room to room, use views from several panoramas, or render a camera path from the model if accuracy matters.