Getting Started with ArtCraft: A Beginner's Tutorial

ArtCraft calls itself "the IDE for artists." Like an IDE for code, it bundles several tools in one place: a prompt box, a 2D canvas, a 3D stage, and a video tool, all connected to dozens of AI models. That breadth can feel like a lot at first. This tutorial gives you a simple path through it, one tool at a time.

What you need before you start

• A computer running macOS or Windows, or any modern browser for the web app. Official desktop builds are a universal macOS installer (Apple Silicon and Intel) and 64-bit Windows installers. See the download guide for details.
• An ArtCraft account, so your projects and generations are saved.
• A way to pay for generations. AI models cost money to run. You can buy an ArtCraft credit plan, or connect accounts you already pay for. The pricing guide explains both options.

Step 1: Install ArtCraft and sign in

1. Pick desktop or web

   Download the installer from getartcraft.com or the GitHub releases page, or open app.getartcraft.com in your browser. The desktop and web versions share the same studio.

2. Sign in

   Log in with your ArtCraft account. If you plan to use ArtCraft credits, check your balance before generating video, which uses credits much faster than images.

3. Connect providers (optional)

   If you already pay for a supported service, connect it so those generations use your existing account. The official README lists Grok (Grok Imagine and Grok Video), Midjourney, Sora (Sora 1, Sora 2, and GPT Image 1), and World Labs (Marble).

Step 2: Make your first image from a prompt

Start the way you would in any AI image tool: write a prompt and pick a model. A good prompt names the subject, action, setting, shot, lighting, and style:

a lighthouse keeper in a yellow raincoat climbing a spiral staircase,
inside a stone lighthouse at night, low-angle wide shot,
warm lantern light against cold blue moonlight, cinematic film still

Which model? For a first test, choose an inexpensive, fast model and save the premium ones for final renders. According to the official credit estimates, 1,000 credits buy about 250 Nano Banana images but only about 67 Nano Banana Pro images. The image generator guide compares the image models, and the prompt library has ready-made prompts to copy.

Step 3: Edit instead of re-rolling

When an image is close but not right, editing is cheaper and more predictable than generating from scratch. ArtCraft offers two kinds of editing:

Phrase prompted edits as instructions and say what to keep: "Change the raincoat to red. Keep the face, pose, and lighting unchanged." More examples are in the editing prompts.

Step 4: Build the scene instead of describing it

This is where ArtCraft differs from a plain prompt box. Its official tagline is "text prompting is neat, but artists crave control." Instead of hoping the model understands "the cup is on the left, slightly behind the laptop," you place things yourself:

• 2D compositing: stack image layers, remove backgrounds, and sketch on a canvas.
• 3D compositing: place backdrops, foreground elements, and props at real depths. See the 3D compositing tutorial.
• Character posing: pose a mannequin and set the camera before rendering. See the character posing tutorial.
• Image to Location: keep one environment consistent across several shots. See the Image to Location tutorial.

You don't need all of these on day one. Try one when prompting alone stops giving you the composition you want.

Step 5: Turn an image into video

ArtCraft's image-to-video tool animates a still frame with the video model of your choice: Seedance, Kling, Veo, Vidu, MiniMax, and others. Animating a good first frame gives far more control than text-to-video, because the composition is already fixed.

1. Prepare the first frame

   Use a finished image from the steps above, ideally staged so the subject and camera are exactly where you want them.

2. Choose a video model

   Test motion with a cheaper, faster model, then switch to a premium model such as Seedance 2.0 for the final render. See the Seedance 2.0 guide.

3. Describe only the motion

   The image already shows the scene. Write what moves and how the camera moves: "she turns toward the window, slow push-in, curtains drift in the breeze."

A simple beginner workflow

Putting it together, here is a workflow that wastes few credits:

1. Explore ideas with text-to-image on a fast, inexpensive model.
2. Pick the best result and fix details with prompted or canvas editing.
3. If the layout or character placement is wrong, rebuild it with compositing or posing rather than rewriting the prompt.
4. Render the final still on a premium image model.
5. Animate it with a fast video model, then render the keeper on a premium one.

Where to go next

• Browse every model and its current status in the model catalog.
• Copy prompts from the prompt library.
• If something goes wrong, see how to fix failed generations.