plusX.Space

from Portfolio Wojciech Karnowka [2023]

by Wojciech Karnowka

OCULAR ENVELOPES

London, United Kingdom

Organisation:

Pilbrow&Partners / Research May 2021

Wojciech Karnowka

Architects and developers are actively pursuing the automation of diverse aspects within the design process. Despite its growing popularity, particularly in the media, it remains accurate to say that generative tools have yet to achieve widespread adoption. have been working on an algorithm aimed at partially automating office space planning. This approach is especially pertinent to refurbishment projects, where rapid feasibility assessment is crucial. The algorithm generates and assesses layouts based on multiple criteria, encompassing environmental considerations.

Originally conceived as a feasibility study at Pilbrow&Partners for PlusX, a UK-based developer, the project has evolved. While the initial studies were halted, I have continued to develop the generative model to facilitate automated fitting for potential projects. My aspiration is to refine the algorithm further, with the ultimate goal of introducing it as a opensource project.

This endeavor and its algorithms draw inspiration from a range of well-established and documented algorithms. The concept is influenced by the approach outlined in “Augmented Space Planning Using Procedural Generation to Automate Desk Layouts” (Anderson et al., 2018).

Borgo Miranda

Monte Bre, Lugano, Switzerland

Client:

Project Team:

Gruner&Friends (remote)

March 2022

Wojciech Karnowka

Calvin Sin

Gruner & Friends designed Borgo Miranda, the 5-star hotel resort in the ticinese village of Monte Bre. Monte Bre is situated on a 1.4 hectare mountainous site above the towns of Locarno and Ascona. It enjoys views towards the south over the Lago Maggiore.

The objective was to create a village that integrates into existing tissue of Montre Bre. The data collected from the site and its envrionment was used to evaluate and optimise the design - with particular focus on views, surfaces and cost. The position of each space is algorithmically decided based on topography, and scoring system developed to evaluate quaity of each configuration.

Topography

The colour gradient displays slope of the topography, with red depicing step areas and green flat areas. Algorithim then places building strategically to facilitate access, to reduce circulation space and minimise excavation.

Clusters

Buildings are clustered to minimise facade surfaces whilst mainitainig local typologies. Consequently, complexity is reduced in construction and design.

View Study

Each individual house in cluster has freedom to rotate and re-locate within X,Y direction to avoid overlapping views to acheive an overfall better view.

Cham

Cham, Switzerland

Gruner&Friends (remote)

March 2022

Wojciech Karnowka Calvin Sin

Gruner&Friends actively participated in an extensive urban study centered around the potential of the “Cham North” site. Spanning over 26,000 sqm, the site boasts a gentle slope that leads toward the Lorze stream.

Positioned along Knonauerstrasse, “Cham North” is poised to serve as the starting point for the forthcoming Cham - Hunberg bypass. This strategic location embeds the area into the transit network and positions it as a notable entrance to Cham.

Throughout the project, in collaboration with Calvin Sin, our design team took the lead in formulating scripts that could generate and assess various massing models based on different typologies. We further optimised specific parameters within the designated design domain, all while considering the site context and ensuring compatibility with the views from neighboring buildings.

Bottom Right: Selection of Building Typologies Explored Through Evolutionary Search and Analysis

Bottom: Evolutionary Solver Main Viewport - Cubes Indicate Dominated Solutions of Three Key Parameters: Visibility, Volume and Area

Visibility Volume Area

Evaluation.Toolkit

London, United Kingdom

Client:

Project Team:

Gruner&Friends

March 2021

Wojciech Karnowka

Calvin Sin

The Evaluation Toolkit comprises tools commissioned by Gruner&Friends, a small practice based in Switzerland, to provide reliable analytical capabilities for architectural proposals during the early design phases.

Collaborating on prior concepts like CHAM and Borgot Miranda, the office management realised that the analysis methods developed were pertinent not only to those projects but also to many early-stage designs. The project involved creating a toolkit within Rhino, scripted using Grasshopper and Python. It included a user interface catering to non-technical users, enabling analysis of designed massings.

The toolkit encompassed tools focusing on Daylight (both proposed and existing), Visibility, Area, and Topography characteristics. In collaboration with german-speaking architect, we established benchmarks and parameters for testing against Swiss building codes and compliance.

Since its inception, the toolkit has become widely utilised within the office to enhance efficiency. It has also been showcased during office presentations and at the University of Applied Sciences Northwestern Switzerland (FHNW).

Analysis of Visibility of Proposed Window Openings

Refresh Button

Every time Rhino geometry is modified, the script requires a refresh. The button will also reset the analysis.

Name used for analysis export. The default name will copy the current Rhino file name.

Directory for analysis export. The default directory is the current Rhino file path.

Before analysis begins, the user can save the layers state. Once the project path is set, the script will hide all layers beside ‘Visualisation’; this is to provide controlled and consistent visualisation. Once the path is reset, the script will restore the visibility to that saved layer state.

Analysis Geometry Section

Analysis of Visibility of Existing Window Openings

Settings - Analysis Specific

Set analysis specific settings.

Analysis - Activate Section

Toggle activates the analysis; once switched on, the results tab will appear, displaying analysis results and export options.

Display Settings Section

Controls visibility/graphic setting of Rhino also allows displaying north direction.

Ennco Render

Monroe, WA, USA

Client:

Project Team:

Ennco

August 2022

Wojciech Karnowka (lead, dev)

Jacek Karnowka (dev)

The project encompassed the development of a Rhino plug-in commissioned by Ennco, a company specialising in spectacle display panels. The primary goal was to devise a user-friendly tool, allowing Ennco’s staff, irrespective of their familiarity with Rhino or rendering technique, to assign materials and generate renders in various configurations. Additionally, the plug-in required functionality, to open files including multiple models, and their separation into separate files.

We conceived a comprehensive solution that featured a material library, an intuitive interface, and a repository of prior solutions. This database organised material lists based on past applications, considerably enhancing operational efficiency. The project’s impact on Ennco’s material preparation process was transformative, markedly expediting their workflow. Following implementation, the plug-in has facilitated the rendering of many images— potentially numbering in the thousands.

Right: Automated Renders Selection Generated by the Tool

Bottom: Selected Code Sample

Bottom Right: Automated Rendering User Interface with ETO Forms

Thesis

London, UK

Institution:

The Bartlett School of Architecture, UCL June 2023

Wojciech Karnowka dr Abel Maciel

My MArch Thesis has been accepted for publication at the Design Computation Input/Output Conference (DCIO 2023) and is currently pending reviewer feedback.

The paper examines the interplay between mental models, design models, and natural design algorithms as a means to investigate the relationship between design practitioners’ cognitive processes and artificial design algorithms.

Since the 1960s, various digital space planning algorithms have been explored, differing in their scope, objectives, and efficiency. However, none of these solutions have come close to matching human-led designs. The thesis hypothesises that the lack of consideration for some of the fundamental components present in natural design algorithms might be the missing link in artificial space planning research.