Repository navigation
What's Changed
- Make 3D models: ask for an object, a prop, a room, or a logo and the AI builds it out of parts (boxes, spheres, cylinders, cones, rings, lathed profiles, extruded outlines, and meshes) with
make_3d_model, saved as a.glbthat opens in Blender and every game engine.send_3d_modelshows an existing.glb,.stl, or.objby @Natuworkguy - Open 3D models beside the chat in the web UI: drag to turn, scroll to zoom, right-drag to pan, with buttons to reset the view, stop the spin, and see the wireframe. three.js ships with Flash, so it works offline by @Natuworkguy
- Build words in a blocky pixel font for logos and signs, so text comes out as letters instead of a box per letter by @Natuworkguy
- Show a vision model a render of each model it makes, with its size and a warning when it floats off the ground, so it can see and fix its own mistakes by @Natuworkguy
- Show models in the colours they were given, starting in front of them by @Natuworkguy
- Open Markdown and text documents the AI sends beside the chat in the web UI, where you can edit them and comment on passages by @Natuworkguy
- Show the AI's browser live beside the chat, step by step, and let it move, click, and drag the mouse at points on the page by @Natuworkguy
- Fix the AI stopping after saying what it would do next ("I'll move the text forward") without doing it: Flash now tells it to carry on by @Natuworkguy
- Put test scripts and throwaway files in the scratch folder, and finished files you asked for in your Downloads folder by @Natuworkguy
- Fix a second
flash --webon Windows quietly sharing a port already in use instead of saying so by @Natuworkguy
Full Changelog: v0.5.11...v0.5.12