Description
Prompting alone has one big annoyance: it doesn't listen. A character changes face between shots, a room looks different from every angle, and you re-roll a dozen times for one usable frame. ArtCraft builds the scene first — compose in a 2D canvas or a 3D scene, set character poses and the camera — and only then hands it to a model for images or video, so the results are controllable and repeatable.
It's an open-source desktop app for macOS, Windows and Linux with no subscription required. You're not locked to one model either: Nano Banana, GPT Image, FLUX, Seedream, Grok, Midjourney, Sora and more sit behind one interface, which suits indie directors, designers and illustrators planning shots.
3D scene staging: Layer backdrops, foregrounds and props in 3D, and block camera angles, object placement and depth with asset kits before generating.
Character posing and identity transfer: Pose a mannequin and set the camera, then have your character appear in that pose and position, so one character holds up across shots.
Image to location and 3D mesh: Turn one image into a reusable environment for multiple shots in the same room, or turn it into a 3D object you can rotate and frame.
2D canvas editing: Image layers, background removal and drawing tools, plus masks and inpainting to say exactly where a picture should change.
Quick prompting: When the idea is still rough, go straight from text to image, edit with a prompt, or animate an image into video, then refine on the canvas.
Pick your model: Choose image, video, music and sound, 3D mesh or world-generation models per task instead of hopping between sites.
It's an open-source desktop app for macOS, Windows and Linux with no subscription required. You're not locked to one model either: Nano Banana, GPT Image, FLUX, Seedream, Grok, Midjourney, Sora and more sit behind one interface, which suits indie directors, designers and illustrators planning shots.
Features
3D scene staging: Layer backdrops, foregrounds and props in 3D, and block camera angles, object placement and depth with asset kits before generating.
Character posing and identity transfer: Pose a mannequin and set the camera, then have your character appear in that pose and position, so one character holds up across shots.
Image to location and 3D mesh: Turn one image into a reusable environment for multiple shots in the same room, or turn it into a 3D object you can rotate and frame.
2D canvas editing: Image layers, background removal and drawing tools, plus masks and inpainting to say exactly where a picture should change.
Quick prompting: When the idea is still rough, go straight from text to image, edit with a prompt, or animate an image into video, then refine on the canvas.
Pick your model: Choose image, video, music and sound, 3D mesh or world-generation models per task instead of hopping between sites.
