AI Renderer

The primary objective of this research is to establish a pipeline that can create a 360 image from a prompt and a 3D interior model, which can then be viewed on WebVR.

Project Details

Year: 2023

Status: Conceptual

Sector: Visualization

Typology: Computation

Scope/Role: Research

Workflow

The proposed workflow involves using Grasshopper and Rhino to generate a base image and a depth image. To generate the prompt, the English text is converted into a stable diffusion prompt, using ChatGPT. The inputs are streamed from the desktop to Colab, where the stable diffusion generates the image.

Base geometry for Al rendering conceptual living room interior without texture

Generating depth pass from base geometry

Image without structured prompt

"minimal living room with marble floor and pop coloured sofa, morning time sunlight from window and dark staircase light coloured wall"

Image with structured prompt after chatgpt

"minimal living room, marble floor, pop coloured sofa, morning time sunlight, window, black staircase, light coloured wall, super detailed, hyper realistic, octane render, vray, cinema 4D --ar 2:1 --uplight"

Shylesh Kumar

(he/him)

Computational Designer in Catalonia

Interested in generative design, computational geometry, machine learning, new media, and game design. With a broad experience with Architectural Sustainability and the integration of parametric 3D modeling and simulation tools into the process, I have worked on projects at various scales - from buildings to planning, and have been involved in research and education programs.

Smortr > Shylesh kumar >AI Renderer

Page updated

Google Sites

Report abuse