Synthetic Visions: AI Text-to-Matter through Neural Radiance Fields

Instructor: Dr. Yota Adilenidou, Director of Arch-hives Ltd. 

​

Workshop lI
Location: Online (Session link to be sent upon subscription)
Dates: September 6-8, 2025​

Time: 13.00 - 17:30 BST (UTC+1)​

Level: Beginner: Architecture students of all levels, Researchers and Professionals

Maximum number of students: 25

Workshop fee: £350

​

Arch-hives introduces the second workshop on Generative AI - Text to Matter: From Image to 3D reconstruction. ​​

​​​​

This workshop explores a cutting-edge workflow that harnesses AI applications to transform 2D images into immersive 3D environments using point clouds and mesh reconstruction. Participants will work with Neural Radiance Fields (NeRFs), a technology that enables AI-generated videos to be decoded into frames captured from multiple angles and camera positions. These are then synthesised into high-resolution 3D reconstructions that capture the spatial and material character of the subject.

The focus of the workshop is the concept of Hybrid/Hacked Materiality through a Text-to-Matter approach. ​ AI generated images are decoded and exploded into point clouds. 

The process moves from image to geometry, generating outcomes that carry both visual and structural depth. AI-generated concepts are developed around the theme of materiality—where virtual forms begin to suggest physical presence. These 3D models are not just representational; they are prepared for potential fabrication through methods such as 3D printing or CNC milling.

​

Participants will gain insight into AI-assisted spatial capture and modelling, while exploring how computation, vision, and material thinking can converge into a hybrid creative practice.

​

- Generate and control images through AI platforms

- Convert images to videos through AI platforms

- Work with NeRFs to convert multi-angle video frames into spatial data
- Generate point clouds and high-resolution mesh reconstructions
- Engage with workflows that bridge AI-generated concepts and physical materialisation
- Develop digital 3D models prepared for fabrication (e.g., 3D printing or CNC)

         ATTENTION: fabrication itself is NOT part of the workshop
- Reflect on hybrid practices where computation, vision, and materiality converge

- Hands on complicated Source installations of NerfStudio and NVidia Instant NGP.

​

System Requirements: 
•    Windows 10/11
•    Graphic card: NVIDIA RTX 3060 and above

​

Software/Applications: 
•    Midjourney -> requires paid subscription (https://www.midjourney.com/)
•    Stable Diffusion -> requires installation (instructions to be given prior to course)
•    Luma AI / Runway ML require paid subscription (https://lumalabs.ai/, https://runwayml.com/)
•    Nvidia Instant NGP -> requires installation (instructions to be given prior to course and during the course)
•    NerfStudio -> requires installation (instructions to be given prior to course and during the course)
•    MeshLab -> requires installation (free) (https://www.meshlab.net/#download)
•    Touchdesigner -> requires installation (free) (https://derivative.ca/download)
•    Meshy AI and Tripo 3D ->requires free registration (https://www.meshy.ai) (https://www.tripo3d.ai/)

 

​​

​

Workshop Schedule

​

Day 1
13.00-14.00 Intro Lecture
14.00-15.00 Midjourney and text to Image platforms SD, control and modes

15.00-15.15 Break
15.15-16.15 Text/Image to Video
16.15-17.30 Luma AI - Runway ML

​

Day 2
13.00-14.00 Intro to Neural Radiance Fields - Luma Captures
14.00-15.00 Installation Nerfstudio

15.00-15.15 Break
15.15-17.30 Working with Nerfstudio

​

Day 3
13.00-14.00 installation NVidia Instant NGP
14.00-15.00 Working with NVidia NGP

15.00-15.15 Break
15.15-16.15 Meshlab / Text to 3D
16.15-17.30 Touchdesigner

​​

 

Limited availability! please email at yota@arch-hives.net if you don't manage to book. If there is request upcoming workshops might be organized.

​

Instructor: Yota Adilenidou, holds a Diploma of Architecture from A.U.Th. an MSc. AAD from GSAPP, Columbia University and a PhD by Architectural Design by the Bartlett School of Architecture. She is the Director of Arch-hives Ltd., a practice that focuses on the research of computational methodologies and digital fabrication for the evolution and activation of matter and form. She has taught many years in Europe and UK and is currently a Lecturer (Teaching) in Design at the Bartlett School of Architecture at UCL and a Visiting Lecturer at University of Westminster.

​

​