agenticroom.setyongr.chatgpt.site

render_scene_snapshot

Render the current 3D room to a JPEG image (data URL) so the arrangement can be judged visually. view presets: “live” uses the editor camera exactly as last left; “orbit”/“top”/“front”/“side” frame the standard room overviews without moving the user’s camera. Returns the image as a data: URL with pixel dimensions; output is downscaled to maxWidth. Only the 3D canvas is captured — UI panels, overlays, and text are never included. Deterministic; never mutates state.

Input schema

{
  "type": "object",
  "properties": {
    "view": {
      "type": "string",
      "enum": [
        "live",
        "orbit",
        "top",
        "front",
        "side"
      ],
      "description": "Which camera view to render (“live” = current editor camera)."
    },
    "maxWidth": {
      "type": "integer",
      "description": "Output width cap in pixels (256-2048, default 1024); height keeps the canvas aspect."
    }
  },
  "additionalProperties": false
}