Launching today
Anthroposcaper
Tag a 2D plan, get a 3D urban environment model
59 followers
Tag a 2D plan, get a 3D urban environment model
59 followers
Draw a 2D plan in the browser or import a DXF, then tag its faces and lines. Turn your own GLB models into templates, and Anthroposcaper lays them along every matching edge, corners and junctions included, into a textured 3D scene. Change one tag or drawing and the scene rebuilds. Export to GLB for your 3D editor or game engine. Connect your AI agent over MCP: it helps set up templates, configure tags and materials, and keep the project organized. You stay in control of the result.
Free Options
Launch Team
Context.devAPI to scrape anything for AI, 20,000+ developers trust us.
Promoted
@pommaniulo Can the exported GLB files keep separate materials and object groups so they’re easy to edit later in Blender or a game engine?
@jedidiah_behar5 Yes on materials: each one is exported as a separate PBR material with its textures embedded, and meshes are split by material, so in Blender or an engine you can tweak or swap them one by one.
Objects come as separate nodes, not one merged mesh: each region of the plan, each stretch of a template along an edge, and each placed model (as an instance sharing the same mesh data). What's missing today is meaningful names and grouping: nodes are named generically, like generated_12.
I don't come from gamedev, so the export structure is based on general reasoning. If you tell me how you'd like it organized for your engine (naming, grouping, merging, pivots), I'm happy to add export settings for that. Grouping by tag and template (road, curb, fence...) is a quick addition either way.
DXF import makes this more practical for existing projects. Does it work with plans that already contain lots of layers and annotations?
@harrywilson Yes. Layers come in as they are, with their names, and you can hide, lock or exclude any of them from region building. That's the key part for busy plans: lines on annotation layers (grids, frames, leaders) would otherwise cut regions, so you just switch those layers off for the build.
Text, dimensions, hatches and blocks are skipped on import, so regions are taken from the lines themselves; if real geometry lives inside blocks, explode them first. On medium-sized plans it runs briskly; with a really heavy plan it's worth excluding what you don't need.
Are there any formatting limitations to watch for?
@garyshenk Mostly on the import side:
• DXF: lines, polylines, circles, arcs and images come in, along with layers and Z heights. Blocks, splines, ellipses and text are skipped, so explode blocks and convert splines to polylines before exporting.
• GLB/glTF: textures must be embedded (a GLB, or a glTF with embedded images); externally referenced image files are skipped. Animations aren't used: templates are static geometry.
• The plan itself: you don't need to close contours by hand, regions are found from line intersections, but lines do need to actually meet. A visible gap leaves a region open.
If you need anything that isn't supported yet, just let me know and I'll add it quickly.
Can the agent also detect missing tags or inconsistent template settings in a project?
@jeankang Yes, that's one of the more useful things to ask it. The agent can read the whole setup: every region with its tags and the parameters they resolve to, every template as the build engine parses it (layer, distribution mode, models), and a report of the last build: for each template, how many edges it matched, what it generated, and if it produced nothing, at which step it stalled (no matching edge, the edges didn't link up, the stretch is shorter than one copy). So "find untagged regions" or "why didn't the fence build here?" are things it can answer and then fix, with every change on the undo stack.
It's not a built-in validator that runs on its own, though: you ask, and the agent does the checking.
@jeankang A bit more on what the agent can do, since it's easy to underestimate. It isn't limited to reading settings: it has access to the geometry and, if you allow it, can take screenshots of the viewport. So it can look at an imported model and set up a template from it, create the tags your project needs, draw a plan from a picture you give it, and even tag the regions on that plan from the picture.
The image part depends on how well your AI model reads images. In my tests that was the weakest link: working with geometry and settings went well, recognizing what's in a picture less so.
Could this work well for creating environments for games where roads buildings and street elements need to stay aligned when the layout changes?
@alheri_murya Yes, with one honest caveat. All generated geometry is bound to the lines of the 2D plan and built from them. Move or reshape a line, press Build 3D, and everything attached to it (curbs, sidewalks, fences, repeated street elements) is rebuilt along its new position, corners included.
What's not there yet is constraints between the lines themselves. If you move one edge of a road, the parallel sidewalk line won't follow on its own; you move it too. A constraint solver for that is on the roadmap.
Could this work well for creating environments for games where roads buildings and street elements need to stay aligned when the layout changes?
@mikkellarsen Same question came in just above, so the short version: generated geometry is rebuilt from the plan lines, so moving a line moves everything attached to it. Constraints between the lines themselves are on the roadmap.
https://www.producthunt.com/products/anthroposcaper?comment=5914216
Is the export worflow mainly built around professional 3D editors or game engines too?
@haolin Export in glb format — it is widely supported. You can use it in both professional 3D editors and game engines.