Agent Details • NoodleFlow

Born elsewhere

Aug 13, 2026, 11:40 AMRead-onlyPublic

Public page

Configure

Inputs

What the person running this agent provides

Inputs are collected before the run starts. Reference them from your instructions as ${variable_name}.

Input photo

Variable name

Type

Required

Origin country

Variable name

Type

Required

Your name

Variable name

Type

Required

Countries to visualize

Variable name

Type

Required

Goal & instructions

What should it achieve, and how should it behave?

Your goal is to create a video showcasing how the person would look if they were born elsewhere.

For that, the user will upload their photo, provide their origin country. User input:
Origin country - ${origin_country}

---

The first thing you should do is to prep the original photo to showcase the country where the person was actually born. Can you use Nano Banana Pro Edit to render an image using the following template:

`The attached person was born in ${origin_country}. Create a new image by changing the background to match the country he was born. As if he is representing the country. Don't include the flag. Use other elements to represent the country`

Let's call this edited image `origin_photo`.

---

Now create a few other images with Nano Banana Pro Edit to create a different image showcasing how the person would look if they were born in another target country. Use origin_photo as the input to each of this.

Target countries - ${visualization_countries}

Use the following prompt:

`The attached person was born in ${origin_country}. Create a new image of what he would look like if he was born in target_country. Change his facial features, skin type, hair type (and whatever else make sense) to match the phenotype of a person from that country. Also change the background to match this new country. But still make him resemble the original person. As if he is representing the country. Don't include the flag. Use other elements to represent the country. And change the clothing to match the vibe/tradition of the country.`

---

Then generate transition videos. Basically the first video segment should use start frame as the origin_photo and end frame as target_photo for target country #1. Then, next video segment should use first frame = target photo country #1 and end frame = target photo country #2, and so forth. Use a prompt like: the person turns right and smiles to his other version while standing still (not moving) and camera pans to the other version and the other person is looking left smiling as well and then looks at the camera. 
Generate all these transition videos and output all of them.
Don’t generate audio on these videos.
Each video should be as short as possible

If the user provided their name, can you also render text overlay on each video showing what their name would look like in that country. Again, try to match the feel of the name but adapt it to the culture of that country with their conventions, stereotypical endings, etc.

User’s name: ${user_name}

And also render that country’s flag followed by country name underneath

Make to so that this text overlay renders in the last 1 sec of previous video and then 3 sec of next video. So if we are doing transition from country x to y to z, then country y’s text should appear for the last 1 sec of first transition video and the first 3 sec of second transition video. This is assuming videos are 4s long. If they are different length, adjust somewhat proportionally. The idea is for the text to appear at the time when a new country is just introduced and stay there until the next country is introduced

And make sure the text type for names is of type title and for country name and flag is caption

For the text that should extend all the way until the end of the video, just set the end param as empty and it will extend all the way automatically. That way it’s more robust

—-

At the end stitch them all together into one video.

Type $ to insert an input variable. References are replaced with the runner's values at runtime.

Planner model

The LLM that plans and coordinates each run

Claude Sonnet 5

Allowed tools

The models & abilities it may reach for

Guardrails

Set limits on what your agent can do

Budget cap

per run · pauses to ask if exceeded

Max steps

tool calls per run · then it stops and returns what it has

Banned actions

Output

The shape the agent should return

Declare the fields every run must return. You define the shape; the agent fills in the values.

Field

Type

What goes here

Configure

Inputs

What the person running this agent provides

Inputs are collected before the run starts. Reference them from your instructions as ${variable_name}.

Input photo

Variable name

Type

Required

Origin country

Variable name

Type

Required

Your name

Variable name

Type

Required

Countries to visualize

Variable name

Type

Required

Goal & instructions

What should it achieve, and how should it behave?

Your goal is to create a video showcasing how the person would look if they were born elsewhere.

For that, the user will upload their photo, provide their origin country. User input:
Origin country - ${origin_country}

---

Let's call this edited image `origin_photo`.

---

Target countries - ${visualization_countries}

Use the following prompt:

---

User’s name: ${user_name}

And also render that country’s flag followed by country name underneath

And make sure the text type for names is of type title and for country name and flag is caption

For the text that should extend all the way until the end of the video, just set the end param as empty and it will extend all the way automatically. That way it’s more robust

—-

At the end stitch them all together into one video.

Type $ to insert an input variable. References are replaced with the runner's values at runtime.

Planner model

The LLM that plans and coordinates each run

Claude Sonnet 5

Allowed tools

The models & abilities it may reach for

Guardrails

Set limits on what your agent can do

Budget cap

per run · pauses to ask if exceeded

Max steps

tool calls per run · then it stops and returns what it has

Banned actions

Output

The shape the agent should return

Declare the fields every run must return. You define the shape; the agent fills in the values.

Field

Type

What goes here