Set up 3D stages and character poses before generating AI video and images
Imagine placing actors on a virtual stage, setting camera angles, and then rendering the final video with a chosen model. ArtCraft is a tool that allows you to visually control composition and character consistency, which is difficult to achieve with text input alone. It serves as an integrated development environment (IDE) for artists, designers, and filmmakers, focusing on combining 2D and 3D elements to create precise and repeatable results.
Pre-generation scene composition
The core of this tool is setting up the scene before generating images. You can layer 3D backgrounds and props to create depth, or remove backgrounds from images to use them as props in 2D and 3D scenes. You can also convert images into 3D meshes to precisely adjust position, rotation, and composition. Using the character posing feature, you can pre-set character poses and camera positions before generating the final shot, reducing the trial and error needed to achieve the desired composition.
Support for various models and providers
ArtCraft supports various AI model providers through a single interface. Image generation includes 16 model families such as Nano Banana, GPT Image, FLUX, and Seedream. For video generation, you can choose from 25 model families including Seedance, Kling, Veo, Sora, Vidu, MiniMax, and Flux. Music and sound are supported by Suno and Seed Audio, 3D mesh generation by Hunyuan 3D, Tripo3D, Meshy, and Rodin, and world generation by Marble and TripoSplat. Some models may currently be disabled or have limitations in the desktop app.
Rapid exploration and precise editing
When exploring ideas quickly, you can generate images with text prompts or instruct image editing with models like Nano Banana Pro and GPT Image. For more precise adjustments, you can use drawing, masking, and inpainting on the canvas to directly specify where the image should change. When transitioning from image to video, you can animate the image by selecting the desired video model and generation controls. This workflow allows you to handle everything from rapid idea exploration to precise scene composition within a single process.
Platform support and installation
Stable versions are provided for Windows and macOS, while Linux requires building from source. The latest builds can be found in GitHub releases. Documentation is also provided for development environment setup, desktop build scripts, roadmap, and license information. Additional provider integrations for Kling, Google, Runway, and Luma are planned, and subscription support for aggregators like OpenArt and Freepik is under review.
Things to check before use
This tool is suitable for users who are familiar with AI image and video generation but find it difficult to achieve desired composition and consistency through prompt engineering alone. It is particularly useful for film production or complex compositing work that requires 3D scene composition and character posing. However, since some models are disabled or limited in the desktop app, you should check the support list in advance if you must use a specific model. Linux users need to understand development environment setup as they must build from source.