> For the complete documentation index, see [llms.txt](https://help.intangible.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.intangible.ai/get-started/your-first-project.md).

# Your first project

Describe a project, confirm the brief, let Intangible agents build the scene, then render keyframes, video and an asset pack. One prompt and two photos to a finished set of shots.

This walkthrough takes one project from a sentence to a downloadable pack. The example is an airbag rover: a lunar landing in five shots, made from one prompt and two photos of the landing site. Use your own subject if you have one; the steps are the same.

You type, decide once, and check the results. Intangible does the building. Most of the time is spent waiting on the agents and the renders.

The video runs the same steps on a different project: a commercial for a smart grocery trolley, from a prompt and three photos to a downloaded asset pack.

{% embed url="<https://www.youtube.com/watch?v=pdM5x3PQ97c>" %}

## Before you start

Sign in. If you have not set up your account yet, see [Set up your account](/get-started/sign-up-and-set-up.md). Projects use credits: the brief, the references, the renders and your messages to the agents all draw on your balance. See [Credits and tokens](/learn-more/concepts/credits-and-tokens.md).

{% stepper %}
{% step %}

## Describe it and bring your files

On Home, type what you want to make in the prompt field. Say what it is, what happens and how many shots you need:

> A pod wrapped in airbags is cut loose 20 meters above a lunar crater. It bounces down the slope, opens, and a rover drives off. Five shots.

Click **+** at the bottom left of the field and add what you already have. The example uses two photos of the site: an orbital view of the crater and a close-up of the landing area. Photos, PDFs and 3D files all work.

![Home with the airbag rover prompt typed in the field, two crater photos attached as tiles under it, and Guide me, Auto and Create project along the bottom](https://1179478100-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FSWrjIqlgVAblx6vHGJiM%2Fuploads%2Fgit-blob-565adfcd4f15e8d440041dc0d733185807d2e794%2Ffirst-project-home-01.png?alt=media)

See [Start a project](/get-started/start-a-project.md) and [Home](/canvas/home.md).
{% endstep %}

{% step %}

## Choose Guide me and create the project

Set the menu next to **+** to **Guide me**. The agents will ask a few questions and write a brief for you to confirm before they build. **Skip** skips the questions and starts building straight from your prompt; use it once you know how your prompts land.

Click **Create project**. Your files upload and the project opens on the canvas.
{% endstep %}

{% step %}

## Answer the questions and confirm the brief

The agents ask their questions in the chat panel, bottom right, one at a time as choice cards. For the rover they asked whether there were CAD models of the pod and rover, whether there were brand guidelines, what the piece needs to do, how it should feel and which image people should remember. Click an answer. The status reads "waiting for your answer" whenever a card needs you.

Then the brief appears as a card at the top of the canvas's first column, and your photos move into a group for the subject they show. For the rover the brief lists five shots: an establishing wide as the pod is cut loose, the first impact, the big bounce, the pod opening, and a resolving wide as the rover drives off. Anything the agents guessed is marked *(assumed)*; read those first. Click a section to fix its text.

When it reads right, pick the confirm option in the chat panel, for example **Confirm, start building**. That is the one decision the build needs from you.

![The Brief card in the canvas's first column with Domain, Direction, Subjects, Tone & Style, Scenes & Shots, Audience and Other decisions, the Groups column with Airbag pod, Rover and Lunar crater, and the chat panel asking Confirm the brief, or edit it first? with Confirm, start building and I'll edit it first](https://1179478100-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FSWrjIqlgVAblx6vHGJiM%2Fuploads%2Fgit-blob-9acf8d4560dd6c8a7c3be1931ff3b2c67a319ad7%2Ffirst-project-brief-01.png?alt=media)

See [The brief](/canvas/the-brief.md) and [Direct Intangible agents](/get-started/direct-intangible-agents.md).
{% endstep %}

{% step %}

## Watch the agents build

The status changes to "building your scene" and the step list shows each piece as it lands on the canvas:

* **Reference sheets,** one per subject group. The rover example gets three: the airbag pod, the rover and the lunar crater.
* **The scene,** built at true scale with the crater, the pod, the lander and the rover inside it. The models live in the scene; they get no cards of their own.
* **Motion.** The drop, the bounces, the opening and the drive-off are animated at lunar gravity.
* **Shots,** one per shot in the brief, framed for your approval. In **Storyboard**, the scene card is followed by five shot cards, 01 Cut loose to 05 Drive-off.

When the agents finish, the status reads "All done! How does it look?" and they report what is in the scene, each shot's status, and any problems they could not solve. Nothing is rendered yet. A choice card asks **What next?**, with options such as fixing two shots first or rendering all five. Ask for a fix first if a shot needs one, for example "Fix shots 2 and 3, but don't render anything." Check the canvas for the cards rather than trusting the status alone.

![The canvas after the build: the Airbag pod, Rover and Lunar crater groups, each with its reference sheet beside any images you uploaded for it, and the Scene 1 card with its five shot cards in the Storyboard](https://1179478100-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FSWrjIqlgVAblx6vHGJiM%2Fuploads%2Fgit-blob-15c8f4a5f911f73d5b0db975780bbc4d0085817a%2Ffirst-project-built-01.png?alt=media)

See [What Intangible agents build](/canvas/what-intangible-agents-build.md) and [Status and steps](/chat-agents-intangible/status-and-steps.md).
{% endstep %}

{% step %}

## Step in by hand where you need to

What the agents build in the scene is ordinary scene content, so you can change it by hand in Build or Compose. To change it yourself, select a card and use its toolbar:

* Scene card, **Edit in Build**: move, add or swap objects in the set.
* Shot card, **Edit in Compose**: reframe the camera or change the lens.

Press **0** or click the canvas icon at the top to go back. Your changes are already on the cards.

For a change you can describe, ask in the chat panel instead and name the card: "add a third camera low on the slope in scene 01".

See [Take control](/get-started/take-control.md) and [Chat with Intangible](/chat-agents-intangible/chat-with-intangible.md).
{% endstep %}

{% step %}

## Render a keyframe for every shot

Select the scene card and click **Batch visualize** in its toolbar. The panel lists every shot with its own prompt, filled in from what is in the shot. If a prompt comes back empty, write one: **Generate** stays off until every shot has a prompt. Pick quality and model in the footer, then click **Generate**.

One keyframe per shot lands in a new **Visualizations** group. Every keyframe comes from the same 3D scene, so the example gets the same pod, the same rover and the same crater in all five frames.

![The Batch visualize panel open from the scene card, with Scene, Style and Lighting tabs, a prompt for each shot (the last one runs below the edge of the panel), and 1K, the model picker and Generate in the footer](https://1179478100-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FSWrjIqlgVAblx6vHGJiM%2Fuploads%2Fgit-blob-fb1d2954a91c62d0713dd20c0a5b097398a30b51%2Ffirst-project-batch-01.png?alt=media)

See [Batch visualize](/canvas/batch-visualize.md).
{% endstep %}

{% step %}

## Turn keyframes into video

Ask the agents in the chat panel, naming the shots:

> Turn the render of shot 03, the big bounce, into a short video. Just that one shot.

Before they start anything, the agents show a choice card with each option's model, length and cost. Animating the render makes the model invent the bounce; one option follows the scene's real bounce instead. Your pick starts the video, and it lands in the shot's Visualize media list and in **Visualizations**.

To fix one frame before you animate it, open **Edit image** on its card, draw around the part to change and say what you want. To give one shot more control, open its render card with **Edit in Visualize**: in the example, shot 04 gets the rover reference there, so the model knows what is inside the pod.

See [Keyframes to video](/canvas/keyframes-to-video.md), [Edit media](/canvas/edit-media.md) and [Visualize](/visualize/visualize.md).
{% endstep %}

{% step %}

## Make an asset pack

Click **Add an asset pack** in the bottom toolbar. Drag the five render cards and the video card onto the pack. It lists them by scene and shot and reads **Draft · Not generated yet**. Click **Generate**, then **Download** on the toast or the pack card. The ZIP holds `images/`, `videos/` and a `manifest.json`, and only what you dragged in; drag the reference sheets in too if the next team needs them.

![Asset Pack 1 beside the Visualizations group, its renders and video in sections by shot, Download in its header, and the toast Your asset pack Asset Pack 1 is ready with Download](https://1179478100-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FSWrjIqlgVAblx6vHGJiM%2Fuploads%2Fgit-blob-0cfa870258d01de0a5d9d342edea9aa893d6193a%2Ffirst-project-pack-01.png?alt=media)

See [Asset packs](/canvas/asset-packs.md) and [Deliver](/get-started/deliver.md).
{% endstep %}
{% endstepper %}

**Try it with your own idea.**

<a href="https://studio.intangible.ai/?utm_source=help.intangible.ai&#x26;utm_medium=docs&#x26;utm_campaign=page-cta&#x26;utm_content=your-first-project" class="button primary" data-icon="arrow-up-right">Start a project in Intangible</a>

## If something comes out wrong

Fix it where it lives, not in the prompt:

* **Wrong framing:** select the shot card, click **Edit in Compose** and move the camera.
* **Wrong subject:** give the subject a better reference, or fix its 3D model in Build.
* **Wrong look:** change the model, style or lighting in Batch visualize or in Visualize.
* **One wrong detail:** draw on the frame with Edit image.

## What's next

* [How Intangible works](/learn-more/concepts/how-intangible-works.md): why one 3D scene sits under every shot.
* [Work well with Intangible agents](/chat-agents-intangible/work-well-with-intangible-agents.md): prompts and habits that get better first builds.
* [References](/learn-more/concepts/image-reference.md): the input that keeps a subject looking the same in every shot.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://help.intangible.ai/get-started/your-first-project.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.