# Reframe

> Turn a recording of people talking into a vertical clip that shows whoever is speaking. One entry adds Transcribe, Speaker Frames and Speaker View, wired.

Source: https://nodaro.ai/docs/nodes/video/reframe

**Reframe** turns a recording of people talking into a vertical clip that shows whoever is speaking. It is one entry in the node picker that adds three nodes, wired together and grouped: [Transcribe](https://nodaro.ai/docs/nodes/audio/transcribe) with speaker detection, [Speaker Frames](https://nodaro.ai/docs/nodes/video/speaker-frames) and [Speaker View](https://nodaro.ai/docs/nodes/video/speaker-view). There is no Reframe node: a saved, exported or imported workflow holds those three nodes.

Reframe has no price of its own. Each of its three nodes charges its own price when it runs. See [Credits](#credits).

Speaker Frames and Speaker View run on Nodaro Cloud. On a self-hosted install, [connect the install to Nodaro Cloud](https://nodaro.ai/docs/self-hosting/cloud-connect), and they work as their own pages describe.

## When to use it
- You have a podcast or interview video and want a 9:16 clip for Shorts, Reels or TikTok that shows whoever is talking.
- You want a speaker-following crop without building an edit with Edit Plan and Camera Switch.
- You want the three nodes wired correctly in one step, and then adjust each one in its own settings panel.

## Quick start
### Select the video

Click the node that outputs your video, for example an **Upload Video** node, so that it is the only node selected.

### Add Reframe

Press Tab and choose **Video › Reframe › Reframe**. You can also drag a connection from a video output onto the canvas and choose **Reframe**. The video is wired into all three nodes.

### Run and review

Click **Run**. Speaker View starts at **Preview** quality, so the first run is the low-cost one. When the framing is right, use **Render final**.

Workflow: Reframe wires one video into Transcribe, Speaker Frames and Speaker View, and Speaker View renders a vertical clip that follows whoever is talking.

- Upload Video → Transcribe (audio)
- Upload Video → Speaker Frames (video)
- Upload Video → Speaker View (video)
- Transcribe → Speaker Frames (transcript)
- Transcribe → Speaker View (transcript)
- Speaker Frames → Speaker View (face tracks)

## Inputs
The group has no input of its own. One video feeds three inputs: the **Audio** input of Transcribe, and the **Video** inputs of Speaker Frames and Speaker View.

- **From a connection.** Drag a connection from a video output onto the canvas and choose **Reframe**. That video is wired into all three nodes.
- **From a selected node.** Add Reframe from the node picker or the sidebar while exactly one node with a video output is selected. That video is wired into all three nodes. With no node selected, or several, nothing is wired.
- **Later.** Otherwise the group shows a **Wire a video** badge. Connect a video to any one of the three inputs, and it is wired to all three at once. This happens only once, while no input has a video yet. An audio source wired into Transcribe is not a video, so the badge stays.

## Outputs
The outputs are those of the three nodes. The clip is the **Video** output of [Speaker View](https://nodaro.ai/docs/nodes/video/speaker-view#outputs). Speaker View also outputs the edit as drawn and the transcript remapped through it, ready for [Add Captions](https://nodaro.ai/docs/nodes/video/add-captions).

## Settings
Reframe has no settings of its own. Each node keeps its own settings panel, so you change them as on the node's own page. Reframe adds the nodes with these settings:

| Node | Connected from | Set to |
| --- | --- | --- |
| [Transcribe](https://nodaro.ai/docs/nodes/audio/transcribe) | The video, into **Audio** | Speaker detection on |
| [Speaker Frames](https://nodaro.ai/docs/nodes/video/speaker-frames) | The video, into **Video**. The transcript of Transcribe, into **Transcript** | The node's defaults |
| [Speaker View](https://nodaro.ai/docs/nodes/video/speaker-view) | The video, into **Video**. The transcript of Transcribe, into **Transcript**. The **Face tracks** of Speaker Frames, into **Face tracks** | **9:16**, **Cut**, **Glide** and **Preview** quality |

The three nodes are added in one step, so one undo removes the whole group. The group is a frame, not a wrapper: you can move, rename, recolour or ungroup it like any group. Deleting one node leaves the others as ordinary nodes.

Groups do not nest. When you add Reframe inside an existing group, the new group is placed beside it. Reframe cannot be added while a run is in progress, or on a workflow you can only view.

## Credits
Reframe has no price of its own. Each of its three nodes charges its own price when it runs:

- **Transcribe** is charged at its own price. See [Transcribe](https://nodaro.ai/docs/nodes/audio/transcribe).
- **Speaker Frames** is charged per started block of 1,000 sampled frames. It samples the whole clip at 2 frames per second, so a 14-second clip is 28 frames, one block. See [Speaker Frames](https://nodaro.ai/docs/nodes/video/speaker-frames#credits).
- **Speaker View** is charged per minute of the clip, at the rate of its **Quality** and tier. See [Speaker View](https://nodaro.ai/docs/nodes/video/speaker-view#credits).

The editor shows one summed estimate for the group: on a chip on the group itself, and as a **Reframe** row in every run confirmation. The figure is the sum of the three nodes' own estimates, so it cannot differ from what the run is checked against. Before the run, Speaker Frames and Speaker View read the clip at its recorded length. When the length of the clip is not known before the run, they quote their ceilings. The run then charges what each node actually used.

## Tips
- **Start with the preview.** Speaker View starts at **Preview** quality. Check the framing, then use **Render final**.
- **Correct a face in Edit framing….** When a face is matched to the wrong speaker, open **Edit framing…** on Speaker View and choose who the face is. See [Face tracking](https://nodaro.ai/docs/nodes/video/speaker-view#face-tracking).
- **Use a wide video.** Speaker View needs extra width to follow a speaker. A video that is already 9:16 or narrower is refused before anything runs.
- **A failed node keeps what was charged.** Each node is its own run. When Speaker Frames fails, Transcribe is already charged, and the later nodes are not.

## Troubleshooting
| Message or problem | Cause and fix |
| --- | --- |
| The group shows **Wire a video**. | No video was connected. Connect a video to any one of the three inputs. |
| Reframe is not in the node picker. | One of the three nodes is not available on this install, or a price is missing. Reframe is offered only where Transcribe, Speaker Frames and Speaker View all are. |
| The video is refused as already as narrow as 9:16. | The video has no extra width to follow a speaker across. Use a wider video, or choose a wider aspect ratio on Speaker View. |
| A speaker is not followed. | Speaker Frames could not tell which face is that speaker. Click **Which face is …?** on Speaker View or Speaker Frames, and choose the face. |

## From the API
Reframe is not a node type, so the API and MCP have no Reframe call. To build the same thing, create the three nodes in a workflow and wire them as in [Settings](#settings). You can also call the routes in a row: transcribe the video with speaker labels, then `POST /v1/speaker-frames` with `videoUrl` and `transcript`, then `POST /v1/speaker-view` with `videoUrl`, `transcript` and `tracks`. See [Speaker Frames](https://nodaro.ai/docs/nodes/video/speaker-frames#api) and [Speaker View](https://nodaro.ai/docs/nodes/video/speaker-view#api).

## Frequently asked questions

### What does Reframe do?

Reframe turns a video of people talking into a vertical clip that shows whoever is speaking. It adds three nodes to the canvas in one group, wired together. Transcribe finds who speaks when, Speaker Frames finds each face, and Speaker View renders the 9:16 clip.

### Is Reframe a node?

No. Reframe is an entry in the node picker that adds three ordinary nodes, Transcribe, Speaker Frames and Speaker View, inside a group. A saved workflow holds those three nodes, never a node called Reframe.

### How much does Reframe cost?

Reframe has no price of its own. Transcribe, Speaker Frames and Speaker View each charge their own price when they run. The group shows one summed estimate before the run.

### Does Reframe need an edit or Camera Switch?

No. Speaker View is fed the plain video and cuts it at each speaker's turns, so Reframe needs no Edit Plan and no Camera Switch.

### What if a face is matched to the wrong speaker?

Open Edit framing… on the Speaker View node and choose who each face is. Save framing stores the corrections on the Speaker Frames node, and Update preview renders the clip again without running Speaker Frames again.