# How to Make a Talking-Photo Video in n8n: A Free Workflow You Can Import

> Source: <https://www.stork.ai/blog/n8n-talking-photo-video-workflow>
> Published: 2026-11-16 22:31:48+00:00

[Make one free](https://www.stork.ai/tools/talking-photo?utm_source=house_ad&utm_medium=inline&utm_campaign=talk-photo)

[Developers: there is an API](https://www.stork.ai/developers/talking-head-api?utm_source=house_ad&utm_medium=inline&utm_campaign=talk-photo-second)

## What you need[#](#what-you-need)

1. A **self-hosted n8n** . Unverified community nodes can't be installed on n8n Cloud ([n8n's installation docs](https://docs.n8n.io/integrations/community-nodes/installation/) ). We submitted the node for n8n's verification on November 16, 2026; until that is approved, Cloud users can't add it.docs.n8n.io
2. A **Stork account and API key** . Create the key at[stork.ai/dashboard/api](https://stork.ai/dashboard/api) . It starts with sk_stork_ and is shown once.stork.ai
3. A **photo with one clear face** , reachable at a public HTTPS address. The workflow ships with a sample photo so you can run it before you have your own.

You do not need credits to follow this tutorial. Talking Photo has a free tier of 2 videos a day; free videos carry a small Stork logo.

## Step 1: install the node[#](#step-1-install-the-node)

In n8n, open **Settings**, then **Community nodes**, click **Install a community node** and enter **n8n-nodes-storkai**. Tick the box that says you understand the risk of unverified code, then click **Install**. n8n's own guide to this screen is [here](https://docs.n8n.io/integrations/community-nodes/installation/gui-install/). The package is on [npm](https://www.npmjs.com/package/n8n-nodes-storkai) and its source is on [GitHub](https://github.com/jupiterspins/n8n-nodes-storkai); it has no runtime dependencies and was published by GitHub Actions with a provenance statement, which is what n8n [requires for verification](https://docs.n8n.io/integrations/creating-nodes/build/reference/verification-guidelines/).

## Step 2: add your API key[#](#step-2-add-your-api-key)

Add the **Stork AI** node to a workflow and click **Connect to Stork AI**. Paste the key and save. Open the credential again and n8n shows **Connection tested successfully**: the node checks the key against the Stork API, so a typo is caught here and not in the middle of a run.

## Step 3: import the workflow[#](#step-3-import-the-workflow)

Download [script-to-talking-photo.json](https://github.com/jupiterspins/n8n-nodes-storkai/blob/main/examples/script-to-talking-photo.json) (or copy its [raw contents](https://raw.githubusercontent.com/jupiterspins/n8n-nodes-storkai/main/examples/script-to-talking-photo.json)). In n8n, create a workflow, open the three-dot menu and choose **Import from File**. Then open the last node and pick your Stork AI credential. The workflow has three nodes:

| The three nodes of the example workflow |  | 
|---|---|
| Node | What it does | 
|---|---|
| Manual Trigger | Starts the run when you click Execute workflow. | 
| Edit Fields | Holds three values: script (the words), photo_url (the face) and voice_id (the voice, en-leo by default). | 
| Stork AI | Resource Talking Photo, operation Create, tier Free. It sends the three values and waits for the finished video. | 

Click **Execute workflow**. The Stork AI node stays busy while the video renders, then outputs one item. The field you want is **video_url**: an MP4 you can download, post or pass to the next node. In our run the item arrived after 167 seconds, for a video 5.6 seconds long.

## Step 4: make it yours[#](#step-4-make-it-yours)

- **Change the words and the face** in the Edit Fields node. The built-in voices are listed at[api.stork.ai/api/v1/voices](https://api.stork.ai/api/v1/voices) ; each has an id like en-ava or en-leo.
- **Feed it from real data.** Replace the Manual Trigger and Edit Fields with a Google Sheets, Airtable or RSS node and map its columns to the three fields. Each incoming row becomes one video.
- **Use your own voice recording.** Set**Voice From** to**Audio File** and give the node an audio URL. The photo then lip-syncs to your recording.
- **Remove the logo.** Set**Tier** to**Paid** . Paid videos are charged per minute from prepaid credits; current prices are on the[developer pages](https://stork.ai/developers) . Run the**Estimate** operation first if you want the price before you spend anything.

## The other two things the node makes[#](#the-other-two-things-the-node-makes)

**Speech** turns text into an MP3 voiceover. In our test a 61-character line came back as a 4.5-second file 5 seconds after the request, and cost 5 credits (5 cents) on the paid tier. **Singing Video** takes a photo and a song and returns a performance video cut between several camera angles. It is the slow one: about 8 minutes for 15 seconds of video and half an hour or more for a whole song.

For a job that long, do not leave the node waiting. Turn off **Wait for Completion**, put the address of an n8n [Webhook node](https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.webhook/) in **Webhook URL**, and continue the workflow from that webhook. Stork calls it when the video is ready, and signs the call so you can check that it is genuine. A job that fails is not charged.

Enjoying this? Get one like it in your inbox each morning.

one email a day · unsubscribe in two clicks · no third-party tracking

## Use it as a tool for an AI Agent[#](#use-it-as-a-tool-for-an-ai-agent)

The node is marked as usable by agents, so it appears as **Stork AI Tool** under an AI Agent's Tool input. We tested the smallest setup we could: n8n's AI Agent, a local [qwen2.5 model with 3 billion parameters](https://ollama.com/library/qwen2.5) running in Ollama, and the Stork AI tool set to Singing Video, Estimate, with the length left for the model to fill in. Asked "How much does a 30 second singing video cost?", the agent called the tool with seconds set to 30 and answered: "A 30 second singing video costs 600 credits ($6)."

One caution from that test. Before we told the model exactly which fields to read, its first answer mixed up the price with the account balance. Small models need a system message that names the fields to use. And if you let an agent call Create, remember that it spends real credits: have it call Estimate first, or keep it on the free tier.

## What it does not do[#](#what-it-does-not-do)

- **It does not run on n8n Cloud yet.** That waits for n8n's verification.
- **It has no trigger node.** Webhooks are set per job with the Webhook URL field and received by n8n's own Webhook node.
- **It does not write the script.** Put an LLM node in front of it if you want the words generated.
- **It animates one face in one photo.** It works on people and on human-proportioned art such as cartoons and paintings. Do not expect it to animate pets or objects.

## Frequently asked questions[#](#frequently-asked-questions)

### Is there an n8n node for making talking-photo videos?[#](#is-there-an-n8n-node-for-making-talking-photo-videos)

Yes. The community node n8n-nodes-storkai adds a Stork AI node with three resources: Talking Photo, Speech and Singing Video. Each has Create, Estimate, Get and Get Many. It installs from npm in any self-hosted n8n.

### Can I use the Stork node on n8n Cloud?[#](#can-i-use-the-stork-node-on-n8n-cloud)

Not yet. n8n Cloud only installs community nodes that n8n has verified. The node was submitted for verification on November 16, 2026. Self-hosted n8n can install it today.

### How much does it cost to make a video from n8n?[#](#how-much-does-it-cost-to-make-a-video-from-n8n)

Nothing on the free tier: Talking Photo allows 2 free videos a day with a small Stork logo. Paid videos are charged per minute from prepaid credits, where 1 credit is 1 US cent. The node's Estimate operation returns the price of a job without making it.

### How long does a talking-photo video take in n8n?[#](#how-long-does-a-talking-photo-video-take-in-n8n)

In our run on November 16, 2026, a 5.6-second free-tier video came back 167 seconds after the workflow started. The node waits for the result by default, so the next node receives the finished video address.

### Can an n8n AI Agent make videos with it?[#](#can-an-n8n-ai-agent-make-videos-with-it)

Yes. The node can be attached to an AI Agent as a tool. In our test a small local model called its Estimate operation correctly. An agent that calls Create spends real credits, so have it estimate first or keep it on the free tier.
