BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//programme.europython.eu//europython-2026//speaker//GYZDC
 U
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-europython-2026-7NWR9R@programme.europython.eu
DTSTART;TZID=CET:20260717T135500
DTEND;TZID=CET:20260717T142500
DESCRIPTION:What if AI could help design 3D worlds without taking creative 
 control away from humans?\n\nIn this talk\, I will present a hosted Blende
 r agent that creates and edits 3D scenes from natural language. The agent 
 runs Blender headlessly inside Docker\, communicates with it through a TCP
  socket protocol\, and exposes Python-powered tools for scene creation\, o
 bject manipulation\, materials\, screenshots\, rendering\, and asset impor
 t. It is built with the Microsoft Agent Framework and Azure AI Foundry\, u
 sing Python as the bridge between the LLM\, Blender’s scripting API\, cl
 oud storage\, and the hosted runtime.\n\nMore than a technical demo\, this
  talk explores a potential future pipeline for AI-assisted 3D creation: a 
 workflow where artists\, game designers\, architects\, and developers can 
 iteratively generate\, inspect\, modify\, and refine 3D environments while
  keeping human control at every stage. Instead of replacing creative tools
 \, the agent becomes a programmable collaborator that can accelerate repet
 itive work\, propose variations\, and turn high-level intent into editable
  Blender scenes.\n\nThrough live examples\, we will discuss what this coul
 d mean for the future of video game world-building\, virtual production\, 
 and 3D architecture: faster prototyping\, more accessible scene creation\,
  and new ways to move from idea to interactive environment. Attendees will
  leave with a practical architecture for building Python agents that opera
 te real creative software\, and a grounded view of both the opportunities 
 and current limitations of human-in-the-loop AI for 3D design.
DTSTAMP:20260524T121644Z
LOCATION:Chamber Hall B (S3B)
SUMMARY:When Python Agents Meet 3D: Automating Blender from Natural Languag
 e - David Rousset
URL:https://programme.europython.eu/europython-2026/talk/7NWR9R/
END:VEVENT
END:VCALENDAR
