# How to Start a Video Podcast That Clips Well

Start a video podcast that clips well: a camera for each voice, faces framed big enough for a 9:16 crop, 4K files, and the specs YouTube and Spotify publish.

- Page: https://www.snipri.com/blog/how-to-start-a-video-podcast
- Author: [Cihan Düzgören](https://www.snipri.com/about), who builds and runs Snipri, and wrote this guide
- Published: November 2, 2026

![Plan the crop before you record: one camera and one microphone for each voice; 19% of the frame's height, the face size that gets a full-height crop; and 4K, a 1215 × 2160 crop, more than the clip's 1080 × 1920. Beside it, two vertical clips Snipri's reframing cut from the film Tears of Steel: a close-up of a grey-haired man, and a two-shot split into two panes, a young woman above him.](https://www.snipri.com/blog/how-to-start-a-video-podcast/cover.webp)

**Short answer.** To start a video podcast that clips well, plan the vertical crop before you record. Give each voice its own camera and microphone, frame each person from the chest up so the face is at least a fifth of the frame's height, record in 4K if you can, and publish the full episode in 16:9 on YouTube or Spotify. Then cut 9:16 clips from it.

- Give each voice a camera and a microphone, and add a wide camera that holds everyone for the exchanges.
- Frame each face at a fifth of the frame's height or more: in Snipri, a face under 12% keeps the whole shot, and one of 19% or more gets a crop as tall as the frame.
- Two people can share a camera when they sit far apart: Snipri gives each a pane once their faces are more than 22% of the frame's width apart.
- Record in 4K if you can: a 9:16 crop as tall as a 4K frame is 1215 × 2160, more than the 1080 × 1920 a vertical clip needs.
- Publish the full episode in 16:9, and keep clips out of the YouTube playlist you mark as a podcast.

A video podcast is published as one long horizontal episode, and its clips are cut vertical from the same recording. This setup plans for both. The platform specs come from YouTube's, Spotify's and Apple's own pages, read on November 2, 2026, and the framing numbers from Snipri's reframing code.

## What do you need to start a video podcast?

You need a camera and a microphone for each voice, a wide camera that holds everyone, light on every face, and a way to record each camera and microphone on its own track.

| Part | What to start with | What it changes for the clips |
| --- | --- | --- |
| Cameras | One for each person, plus one wide shot of everyone | A close camera gives a vertical clip one face to crop to |
| Microphones | One for each person, just below the chin | A clean voice makes accurate captions, and a mouth in view lets speaker detection see who talks |
| Resolution | 4K, or 1080p at least | How much a 9:16 crop has to be enlarged |
| Frame rate | One rate from recording to export, such as 24, 25 or 30 frames a second | YouTube asks for the frame rate you recorded |
| Light | A soft light in front of each face | An evenly lit face is easier to find and follow |
| Recording | Every camera and microphone on its own track | The edit can cut to whoever is talking |

## Where should the cameras go?

Put a close camera beside each person, aimed across the table at the other, and a wide camera on the open side that holds everyone. Each close camera sees its person almost face on while they talk to the other, and the wide one keeps the exchanges, the reactions and anyone the close cameras miss.

The edit cuts between the three, and a clipping tool reframes whichever shot is on screen. So the framing of each camera decides what a vertical clip can show.

![A plan of two people at a table seen from above, the host on the left and the guest on the right, each with a microphone in front. Camera A faces them from the open side, camera B stands beside the guest aimed at the host, and camera C beside the host aimed at the guest. Below, the shot each gives: A holds both people, each inside a pane; B and C each hold one person from the chest up inside one window as tall as the frame.](https://www.snipri.com/blog/how-to-start-a-video-podcast/camera-plan.webp)

*A drawing, not to scale: three cameras for two voices, and the 9:16 windows Snipri's reframing puts in each shot by the rules this guide describes.*

## How big should a face be in the frame?

Frame each person from the chest up, so the face, from forehead to chin, is at least a fifth of the frame's height. At that size, a vertical crop of one person can use the frame's full height.

Snipri's [auto reframe](https://www.snipri.com/features/auto-reframe) measures each face against the frame's height, and two thresholds in its code decide the crop:

- **Under 12%,** it keeps the whole shot and fills the rest of the vertical frame with a darkened, blurred copy of it.
- **From 12%,** it crops to the face, with a tighter window that it enlarges more, up to 2.5 times.
- **From 19%,** the window is as tall as the frame, the most a 9:16 crop of it can hold.

A wide shot of a room puts faces far under the first threshold. In the scene below, the largest face is 6 to 7% of the frame's height. A close-up from the shoulders up puts it near half.

![A meter of face height from 0 to 60% of the frame's height. Under 12% is hatched: the whole shot. 12 to 19%: a tighter window, enlarged up to 2.5 times. 19% and over is lime: one window as tall as the frame. A wide shot of a room, its largest face at 6%, is marked under 12%, and a close-up of a grey-haired man, his face about half the frame's height, is marked in the lime part. Each frame is shown with its face box beside the vertical clip Snipri cut from it: the whole shot over a blurred copy for the wide one, a full-height crop of his face for the close-up.](https://www.snipri.com/blog/how-to-start-a-video-podcast/face-size.webp)

*Two scenes from the film Tears of Steel as Snipri's reframing cut them. The thresholds are the ones in its code, and the face shares are the worker's own measures on these frames.*

## Can two people share one camera?

Yes, if they sit far enough apart. Snipri gives two people a pane each, one above the other, when their faces are more than 22% of the frame's width apart and neither pane would show the other's face. Seated closer, they stay together in one window.

In a two-person shot, Snipri can also cut to whoever talks alone for a while, judged from the movement of each mouth and the sound. A quick exchange keeps both on screen. Keep each microphone below the chin, so it does not hide the mouth.

![A frame of a young woman and a grey-haired man standing apart in a room of books and red lamps, with a lime window round each face, beside the vertical clip Snipri cut from it: the woman in the top pane, the man in the bottom one. Measured over the shot: 49 to 54% of the frame's width between their faces against the 22% rule, 6.2 to 8.4 face widths against 1.8, and the smaller face's area 44 to 80% of the larger's against 10%.](https://www.snipri.com/blog/how-to-start-a-video-podcast/two-people.webp)

*A two-shot from the film Tears of Steel, split by Snipri's reframing into two panes. The rule's numbers are from its code; the measures are the worker's on this shot.*

The video puts the three shots side by side, from the farthest camera to the closest, each with the number that decided its crop.

[Video: A wide shot, a two-shot and a close-up, each as Snipri's reframing cut it](https://www.snipri.com/blog/how-to-start-a-video-podcast/three-framings.mp4)

*Three shots from the film Tears of Steel, from the farthest camera to the closest, each as Snipri's reframing cut it, beside the frame it came from. The wide shot's largest face is 6 to 7% of the frame's height, so the whole shot is kept; the two faces of the two-shot are 49 to 54% of the width apart, so each gets a pane; the close-up's face is about half the frame's height, so one window as tall as the frame follows it. The windows, face boxes and gauges are drawn from the worker's model, and the clips have no captions.*

Film: “Tears of Steel” by Blender Foundation, from [mango.blender.org](https://mango.blender.org/), under [CC BY 3.0](https://creativecommons.org/licenses/by/3.0/), cut and reframed, without sound.

## What resolution and frame rate should you record?

Record in 4K if your cameras can, and in 1080p at least. A vertical clip is 1080 × 1920. A 9:16 window as tall as a 4K frame is 1215 × 2160, so it is scaled down to fit. From 1080p the window is 608 × 1080 and has to be enlarged 1.78 times, and from 720p 2.67 times, which leaves the clip softer.

The gap grows when a face is small. For a face at 12% of the frame's height, the smallest Snipri crops to, a 1080p clip is enlarged 2.5 times and a 4K clip 1.41 times.

![Four rectangles drawn to scale: the 9:16 window as tall as a 720p frame, 405 × 720, enlarged 2.67 times; as tall as a 1080p frame, 608 × 1080, enlarged 1.78 times; as tall as a 4K frame, 1215 × 2160, scaled down to 0.89; and the 1080 × 1920 clip. A table adds that for a face at 12% of the frame's height, the smallest Snipri crops to, the clip is enlarged 2.67 times from 720p, 2.5 times from 1080p and 1.41 times from 4K.](https://www.snipri.com/blog/how-to-start-a-video-podcast/record-in-4k.webp)

*A drawing to scale, with the factors worked out from the clip size Snipri renders and the window rules in its reframing code.*

Keep one frame rate from the cameras to the export. YouTube asks for video encoded at the frame rate it was recorded and names 24, 25, 30, 48, 50 and 60 as common rates; Spotify takes 24, 25, 30, 50 or 60.

A 4K file is large. At the 35 to 45 Mbps YouTube recommends for 4K at 24 to 30 frames a second, an hour is about 16 to 20 GB, so check the upload limit of the tool you clip with, Snipri's [plans](https://www.snipri.com/pricing) included, before you export.

## How should you record the sound?

Give each person their own microphone on its own track, just below the chin and angled up at the mouth. Separate tracks let you fix one voice without touching the other, and a close, clean voice makes the transcript, and with it the captions on every clip, more accurate. [Captioning podcast clips](https://www.snipri.com/blog/add-captions-to-podcast-clips) covers what to check before a caption goes out.

For the upload, YouTube recommends AAC-LC or Opus audio at 48 kHz, and Spotify AAC-LC, PCM or FLAC at 128 kbps or higher, in stereo.

## Where should you publish the episode?

Publish the full episode in 16:9 on YouTube and Spotify, and on Apple Podcasts through a hosting provider that supports video. On YouTube, create a podcast in YouTube Studio, from a new playlist or one you already have, and put only full-length episodes in it, in order. YouTube's help asks you not to add podcast features to a playlist of clips.

| Platform | What it asks for | Where it comes from |
| --- | --- | --- |
| YouTube | Full-length episodes in a podcast playlist, in order; MP4 with H.264 at the frame rate recorded; 16:9; 35 to 45 Mbps for 4K and 8 Mbps for 1080p at 24 to 30 frames a second | YouTube Help: Create a podcast in YouTube Studio, and recommended upload encoding settings |
| Spotify | MP4 recommended, MOV compatible; H.264 or H.265; 1080p or higher; 16:9; under 10 GB and 4 hours recommended, up to 60 GB and 12 hours | Spotify: Video specs |
| Apple Podcasts | Video through a hosting provider that supports it, which publishes for you; file types vary by provider; standard RSS video feeds remain available | Apple Podcasts for Creators: How to publish video on Apple Podcasts |

## How do you turn the episode into clips?

Upload the edited episode, the one that cuts between the cameras, and let the clipping tool find the moments and reframe each shot. Snipri's [podcast clip maker](https://www.snipri.com/tools/podcast-clip-maker) reads the transcript for moments that stand alone, cuts each to 9:16 by the rules above and burns the captions in.

Then review each clip as [How to make podcast clips](https://www.snipri.com/blog/how-to-make-podcast-clips) describes: a moment that makes sense without the rest of the episode, cut from its question to the end of its answer. For YouTube, [How long can YouTube Shorts be](https://www.snipri.com/blog/how-long-can-youtube-shorts-be) has the length limits, and [How to make YouTube Shorts from long videos](https://www.snipri.com/blog/how-to-make-youtube-shorts-from-long-videos) the rest of the workflow.

## Common questions

### How many cameras do I need for a video podcast?

One for each person and one wide shot is the setup in this guide. A single wide camera can work when everyone sits far enough apart and each face is at least 12% of the frame's height, so a vertical crop has a face to follow.

### Should I record a video podcast vertically?

No. Record the episode in 16:9, the standard shape on YouTube and the one Spotify recommends, and cut the vertical clips from it. Spotify shows a video of any other shape letterboxed or pillarboxed in some of its apps.

### Can I start a video podcast with a phone?

Yes. Record at least 1080p with the phone held horizontally, give each voice its own microphone, keep one frame rate for the whole recording, and frame each person from the chest up as you would with any camera.

### How big is a 4K video podcast file?

At the 35 to 45 Mbps YouTube recommends for 4K at 24 to 30 frames a second, an hour of video is about 16 to 20 GB. Spotify recommends files under 10 GB and takes up to 60 GB.

## Sources

1. [YouTube Help: Create a podcast in YouTube Studio](https://support.google.com/youtube/answer/12751636), checked November 2, 2026.
2. [YouTube Help: YouTube recommended upload encoding settings](https://support.google.com/youtube/answer/1722171), checked November 2, 2026.
3. [Spotify: Video specs](https://support.spotify.com/us/podcasters/article/video-specs/), checked November 2, 2026.
4. [Apple Podcasts for Creators: How to publish video on Apple Podcasts](https://podcasters.apple.com/support/5593-how-to-publish-video), checked November 2, 2026.

Read next: [How to Make Podcast Clips That Stand Alone](https://www.snipri.com/blog/how-to-make-podcast-clips)