Cumincad : CUMINCAD Papers : Search Results

CumInCAD is a Cumulative Index about publications in Computer Aided Architectural Design
supported by the sibling associations ACADIA, CAADRIA, eCAADe, SIGraDi, ASCAAD and CAAD futures

PDF papers
References

Hits 1 to 20 of 774

Reformat results as: short short into frame detailed detailed into frame

_id	ecaade2023_197
id	ecaade2023_197
authors	Kim, Frederick Chando, Johanes, Mikhael and Huang, Jeffrey
year	2023
title	Text2Form Diffusion: Framework for learning curated architectural vocabulary
source	Dokonal, W, Hirschberg, U and Wurzer, G (eds.), Digital Design Reconsidered - Proceedings of the 41st Conference on Education and Research in Computer Aided Architectural Design in Europe (eCAADe 2023) - Volume 1, Graz, 20-22 September 2023, pp. 79–88
doi	https://doi.org/10.52842/conf.ecaade.2023.1.079
summary	Stepping towards the machine learning age in architecture, many architects and researchers have developed creative ways of utilizing machine learning with domain-specific architectural datasets in recent years. With the rising popularity of large language-based text-to-image models, architects have relied on different strategies for developing the prompt to create satisfactory images representing architecture, which lessens the agency of the architects in the process. We explore alternative ways of working with such models and put forward the role of designers through the fine-tuning process. This research proses a fine-tuning framework of a pre-trained language model, namely Stable Diffusion, with a dataset of formal architectural vocabularies towards developing a new way of understanding architectural form through human-machine collaboration. This paper explores the creation of an annotation system for machines to learn and understand architectural forms. The results showcased a promising method combining different formal characteristics for architectural form generation and ultimately contributing to the discourse of form and language in architecture in the age of large deep learning models.
keywords	machine learning, diffusion model, architectural form, text-to-architecture
series	eCAADe
email
last changed	2023/12/10 10:49

_id	ecaade2023_266
id	ecaade2023_266
authors	Çiçek, Selen, Turhan, Gözde Damla and Özkar, Mine
year	2023
title	Reconsidering Design Pedagogy through Diffusion Models
source	Dokonal, W, Hirschberg, U and Wurzer, G (eds.), Digital Design Reconsidered - Proceedings of the 41st Conference on Education and Research in Computer Aided Architectural Design in Europe (eCAADe 2023) - Volume 1, Graz, 20-22 September 2023, pp. 31–40
doi	https://doi.org/10.52842/conf.ecaade.2023.1.031
summary	The text-to-image based diffusion models are deep learning models that generate images from text-based narratives in user-generated prompts. These models use natural language processing (NLP) techniques to recognize narratives and generate corresponding images. This study associates the assignment-based learning-by-doing of design studio with the prompt-based diffusion models that require fine-tuning in each image generation. The reference is a specific formal education setup developed within the context of compulsory courses in design programs’ curricula. We explore the implications of diffusion models for a model of the basic design studio as a case study. The term basic design implies a core and foundational element of design. To explore and evaluate the potential of AI tools to improve novice designers’ design problem solving capabilities, a retrospective analysis was conducted for a series of basic design studio assignments. The first step of the study was to reframe the assignment briefs as design problems and student design works as design solutions. The outcomes of the identification were further used as input data to generate synthetic design solutions by text-to-image diffusion models. In the third step, the design solution sets generated by students and the diffusion models were comparatively assessed by design experts with regards to how well they answered to the design problems defined in the briefs. The initial findings showed that diffusion models were able to generate a myriad of design solutions in a short time. It is conjectured that this might help students to easily understand the ill-defined design problem requirements and generate visual concepts based on written descriptions. However, the comparison indicated the value of design reasoning conveyed in the studio, as it gets highlighted with the lack of improvement in the learning curve of the diffusion model recorded through the synthetic design process.
keywords	Deep Learning, Diffusion Models, Design Education, Basic Design, Design Problems
series	eCAADe
email
last changed	2023/12/10 10:49

_id	cdrf2023_24
id	cdrf2023_24
authors	Haoran Ma, Hao Zheng
year	2023
title	Text Semantics to Image Generation: A Method of Building Facades Design Base on Stable Diffusion Model
source	Proceedings of the 2023 DigitalFUTURES The 5st International Conference on Computational Design and Robotic Fabrication (CDRF 2023)
doi	https://doi.org/https://doi.org/10.1007/978-981-99-8405-3_3
summary	Stable Diffusion model has been extensively employed in the study of architectural image generation, but there is still an opportunity to enhance in terms of the controllability of the generated image content. A multi-network combined text-to-building facade image generating method is proposed in this work. We first fine-tuned the Stable Diffusion model on the CMP Facades dataset using the LoRA (Low-Rank Adaptation) approach, then we apply the ControlNet model to further control the output. Finally, we contrasted the facade generating outcomes under various architectural style text contents and control strategies. The results demonstrate that the LoRA training approach significantly decreases the possibility of fine-tuning the Stable Diffusion large model, and the addition of the ControlNet model increases the controllability of the creation of text to building facade images. This provides a foundation for subsequent studies on the generation of architectural images.
series	cdrf
email
last changed	2024/05/29 14:04

_id	cdrf2023_35
id	cdrf2023_35
authors	Zexi Lyu, Zao Li, Zijing Wu
year	2023
title	Research on Image-to-Image Generation and Optimization Methods Based on Diffusion Model Compared with Traditional Methods: Taking Façade as the Optimization Object
source	Proceedings of the 2023 DigitalFUTURES The 5st International Conference on Computational Design and Robotic Fabrication (CDRF 2023)
doi	https://doi.org/https://doi.org/10.1007/978-981-99-8405-3_4
summary	The intersection of technology and culture has become a topic of great interest worldwide, with China's development embracing this integration as an essential direction. One critical area where these two fields converge is in the inheritance, translation, and creative design of cultural heritage. In line with this trend, our study explores the potential of stable diffusion to produce highly detailed and visually stunning building façades. We start by providing an overall survey and algorithm fundamentals of the generative deep learning models used so far, namely, GAN and Diffusion models. Then, we present our methodology for using Diffusion Model to generate architecture façades. We then demonstrate how the fine-tuning is done for Stable Diffusion is done to yield optimal performance and then evaluate four different training methods of SD. We also compare existing GAN based façade generation method with our Diffusion based method. Our results show that our Diffusion-based approach outperforms existing methods in terms of detail and quality, highlighting the potential of stable diffusion in generating visually appealing building façades. This research contributes to the growing body of work on the integration of technology and culture in architecture and offers insights into the potential of stable diffusion for creative design applications.
series	cdrf
email
last changed	2024/05/29 14:04

_id	ascaad2023_071
id	ascaad2023_071
authors	Gabr, Rana
year	2023
title	Exploring the Integration of Mid-Journey AI for Architectural Post-Occupancy Evaluation - Prioritizing Experiential Assessment: Case Study of the Egyptian Museum and National Museum of Egyptian Civilization
source	C+++: Computation, Culture, and Context – Proceedings of the 11th International Conference of the Arab Society for Computation in Architecture, Art and Design (ASCAAD), University of Petra, Amman, Jordan [Hybrid Conference] 7-9 November 2023, pp. 518-550.
summary	The post-occupancy evaluation (POE) process, traditionally reliant on subjective user feedback and observations, has evolved in response to global concerns like climate change and sustainability. This shift towards greater objectivity and quantification reflects an increased focus on precise measurement of environmental and performance metrics. Consequently, architectural assessment is now more quantitatively oriented, moving away from a predominantly experiential emphasis. This research investigates the integration of the emerging AI tool Mid-journey into the POE process, specifically targeting the evaluation of experiential aspects in architectural design. It proposes that AI tools can be instrumental for architects and evaluators by translating user feedback into visual representations and conceptual insights. The study aims to initiate a discourse on the role of text-to-image models in assessing user experiences, potentially becoming integral to the design and concept generation process. The research combines quantitative methods like surveys and AI-driven experiments with qualitative approaches such as observations and interviews to offer enhancement proposals for Egyptian museums, comparing traditional POE solutions and frameworks with new proposed framework that incorporates AI-generated alternatives. This study emphasizes the dual role of museums as artifact custodians and platforms for public education about ancient cultures. It highlights the imperative to transform Egyptian national museums into immersive learning environments, rather than mere storage spaces. The aspiration is to create museums that securely display and preserve artifacts while fostering educational engagement in preserving our shared history.
series	ASCAAD
email
last changed	2024/02/13 14:40

_id	caadria2023_446
id	caadria2023_446
authors	Guida, George
year	2023
title	Multimodal Architecture: Applications of Language in a Machine Learning Aided Design Process
source	Immanuel Koh, Dagmar Reinhardt, Mohammed Makki, Mona Khakhar, Nic Bao (eds.), HUMAN-CENTRIC - Proceedings of the 28th CAADRIA Conference, Ahmedabad, 18-24 March 2023, pp. 561–570
doi	https://doi.org/10.52842/conf.caadria.2023.2.561
summary	Recent advances in Natural Language Processing (NLP) and Diffusion Models (DMs) are leading to a significant change in the way architecture is conceived. With capabilities that surpass those of current generative models, it is now possible to produce an unlimited number of high-quality images (Dhariwal and Nichol 2021). This opens up new opportunities for using synthetic images and marks a new phase in the creation of multimodal 3D forms, central to architectural concept design stages. Presented here are three methodologies of generation of meaningful 2D and 3D designs, merging text-to-image diffusion models Stable Diffusion, and DALL-E 2 with computational methods. These allow designers to intuitively navigate through a multimodal feedback loop of information originating from language and aided by artificial intelligence tools. This paper contributes to our understanding of machine-augmented design processes and the importance of intuitive user interfaces (UI) in enabling new dialogues between humans and machines. Through the creation of a prototype of an accessible UI, this exchange of information can empower designers, build trust in these tools, and increase control over the design process.
keywords	Machine Learning, Diffusion Models, Concept Design, Semantics, User Interface, Design Agency
series	CAADRIA
email
last changed	2023/06/15 23:14

_id	ijac202321202
id	ijac202321202
authors	Koehler, Daniel
year	2023
title	More than anything: Advocating for synthetic architectures within large-scale language-image models
source	International Journal of Architectural Computing 2023, Vol. 21 - no. 2, 242–255
summary	Large-scale language-image (LLI) models have the potential to open new forms of critical practice through architectural research. Their success enables designers to research within discourses that are profoundly connected to the built environment but did not previously have the resources to engage in spatial research. Although LLI models do not generate coherent building ensembles, they offer an esthetic experience of an AI infused design practice. This paper contextualizes diffusion models architecturally. Through a comparison of approaches to diffusion models in architecture, this paper outlines data-centric methods that allow architects to design critically using computation. The design of text-driven latent spaces extends the histories of typological design to synthetic environments including non-building data into an architectural space. More than synthesizing quantic ratios in various arrangements, the architect contributes by assessing new categorical differences into generated work. The architects’ creativity can elevate LLI models with a synthetic architecture, nonexistent in the data sets the models learned from.
keywords	diffusion models, large-scale language-image models, data-centric, access to data, discrete computation, critical computational practice, synthetic architecture
series	journal
last changed	2024/04/17 14:30

_id	ecaade2023_68
id	ecaade2023_68
authors	Mugita, Yuki, Fukuda, Tomohiro and Yabuki, Nobuyoshi
year	2023
title	Future Landscape Visualization by Generating Images Using a Diffusion Model and Instance Segmentation
source	Dokonal, W, Hirschberg, U and Wurzer, G (eds.), Digital Design Reconsidered - Proceedings of the 41st Conference on Education and Research in Computer Aided Architectural Design in Europe (eCAADe 2023) - Volume 2, Graz, 20-22 September 2023, pp. 549–558
doi	https://doi.org/10.52842/conf.ecaade.2023.2.549
summary	When designing a new landscape, such as when demolishing buildings and building new ones, visual methods are effective in sharing a common image. It is possible to visualize future landscapes by making sketches and models, but this requires a great deal of skill and effort on the part of the creator. One method for visualizing future landscapes without the need for specialized skills or labor is image generation using deep learning, and a method has been proposed of using deep learning to generate landscape images after demolishing current buildings. However, there are two problems: the inability to remove arbitrary buildings and the inability to generate a landscape after reconstruction. Therefore, this study proposes a future landscape visualization method that integrates instance segmentation and a diffusion model. The proposed method can generate both post-removal images of existing buildings and post-reconstruction images based on text input, without the need for specialized technology or labor. Verification results confirmed that the post-removal image was more than 90% accurate when the building was removed and replaced with the sky. And the post-reconstruction image matched the text content with a best accuracy of more than 90%. This research will contribute to the realization of urban planning in which all project stakeholders, both professionals and the public, can directly participate by visualizing their own design proposals for future landscapes.
keywords	landscape visualization, deep learning, diffusion model, instance segmentation, text input, text-to-image model, inpainting
series	eCAADe
email
last changed	2023/12/10 10:49

_id	acadia23_v3_49
id	acadia23_v3_49
authors	A. Noel, Vernelle; Dortdivanlioglu, Hayri
year	2023
title	Text-to-image generators: Semiotics, Semantics, Syntax, and Society
source	ACADIA 2023: Habits of the Anthropocene: Scarcity and Abundance in a Post-Material Economy [Volume 3: Proceedings of the 43rd Annual Conference for the Association for Computer Aided Design in Architecture (ACADIA) ISBN 979-8-9891764-1-0]. Denver. 26-28 October 2023. edited by A. Crawford, N. Diniz, R. Beckett, J. Vanucchi, M. Swackhamer 24-32.
summary	Text-to-image generators, such as Midjourney, DALL-E, and Stable Diffusion, are becoming increasingly popular. These generators, whose models are trained on large datasets of text-image pairs, often scraped from the web, take text prompts as input, and use them to generate images—text-to-image prompting. In this visual essay, we raise questions about the entanglement of semiotics, semantics, syntax, and society in these text-to-image generator tools. We are intrigued by how these technologies are “intrawoven” with social and cultural contexts. How are their constructions and presentations reconfigurations? How do, or might they, inform pedagogy, theory, methods, and our publics? To explore these questions, we entered six prompts related to the built environment in six different languages, eight months apart in Midjourney (“Midjourney” n.d.). The generated images (Figure 1), require that we ask deep questions of each image, in comparison with each other, across each group of four, and across time (eight months apart). We argue that text-to-image generators call for a rigorous exploration of semiotics, semantics, syntax, and the society, with implications for pedagogy, theory-building, methodologies, and public enlightenment. Furthermore, we assert that these tools can facilitate pertinent questions about the relationships between technology and society. This is just the beginning. For now, we have questions.
series	ACADIA
type	field note
email
last changed	2024/04/17 13:59

_id	ecaade2023_145
id	ecaade2023_145
authors	Dortheimer, Jonathan, Schubert, Gerhard, Dalach, Agata, Brenner, Lielle Joy and Martelaro, Nikolas
year	2023
title	Think AI-side the Box! Exploring the Usability of text-to-image generators for architecture students
source	Dokonal, W, Hirschberg, U and Wurzer, G (eds.), Digital Design Reconsidered - Proceedings of the 41st Conference on Education and Research in Computer Aided Architectural Design in Europe (eCAADe 2023) - Volume 2, Graz, 20-22 September 2023, pp. 567–576
doi	https://doi.org/10.52842/conf.ecaade.2023.2.567
summary	This study examines how architecture students use generative AI image generating models for architectural design. A workshop was conducted with 25 participants to create designs using three state-of-the-art generative diffusion models and BIM or 3D modeling software. Results showed that the participants found the image-generating models useful for the preliminary design stages but had difficulty when the design advanced because the models did not perform as they expected. Finally, the study shows areas for improvement that merit further research. The paper provides empirical evidence on how generative diffusion models are used in an architectural context and contributes to the field of digital design.
keywords	Machine Learning, Diffusion Models, Design Process, Computational Creativity
series	eCAADe
email
last changed	2023/12/10 10:49

_id	ecaade2023_48
id	ecaade2023_48
authors	Doumpioti, Christina and Huang, Jeffrey
year	2023
title	Text to Image to Data
source	Dokonal, W, Hirschberg, U and Wurzer, G (eds.), Digital Design Reconsidered - Proceedings of the 41st Conference on Education and Research in Computer Aided Architectural Design in Europe (eCAADe 2023) - Volume 2, Graz, 20-22 September 2023, pp. 541–548
doi	https://doi.org/10.52842/conf.ecaade.2023.2.541
summary	Recent advancements in text-to-image technology have transformed the landscape of computational creativity by enabling the generation of conceptual images. By implementing innovative standards for image generation, we can now shift our focus from the constraints of notational design communication to more purposeful reflection, opening up new design possibilities for future architectures informed by contemporary ideas, concepts, and concerns. In light of the pressing climatic crisis, this paper specifically explores the relationship between text-to-image generation and the integration of environmental sensibility, aiming to explore how digital information (bits) can translate into physical reality (atoms). Our case study focuses on specific residential building typology and its façade morphology to analyse the environmental responsiveness of the design. We propose a workflow that merges creative and analytic processes, through different stages, including diffusion-generated conceptual images, 2D to 3D through depth-mapping and point-cloud meshing, semantic segmentation analysis and sunlight simulation. The paper describes the methods and their combination into a coherent workflow, outlines encountered setbacks, and suggests stages for further improvement.
keywords	Computational Creativity, Text-to-Image, Simulation, Environmental Responsiveness, Machine Intelligence
series	eCAADe
email
last changed	2023/12/10 10:49

_id	acadia23_v2_420
id	acadia23_v2_420
authors	Guida, George; Escobar, Daniel; Navarro, Carlos
year	2023
title	3D Neural Synthesis: Gaining Control with Neural Radiance Fields
source	ACADIA 2023: Habits of the Anthropocene: Scarcity and Abundance in a Post-Material Economy [Volume 2: Proceedings of the 43rd Annual Conference for the Association for Computer Aided Design in Architecture (ACADIA) ISBN 979-8-9860805-9-8]. Denver. 26-28 October 2023. edited by A. Crawford, N. Diniz, R. Beckett, J. Vanucchi, M. Swackhamer 420-428.
summary	This research introduces a novel, 3D machine-learning, aided design approach for early design stages. Integrating language within a multimodal framework grants designers greater control and agency in generating 3D forms. The proposed method leverages Stable Diffusion and Runway's Gen1 through the generation of 3D Neural Radiance Fields (NeRFs), surpassing the limitations of 2D image-based outcomes in aiding the design process. This paper presents a flexible machine-learning workflow taught to students in a conference workshop, and outlines the multimodal methods used - between text, image, video, and NeRFs. The resultant NeRF design outcomes are contextualized within a Unity agent-based, virtual environment for architectural simulation, and are expe- rienced with real-time VFX augmentations. This hybridized design process ultimately highlights the importance of feedback loops and control within machine-learning, aided-design processes.
series	ACADIA
type	paper
email
last changed	2024/04/17 13:59

_id	sigradi2023_114
id	sigradi2023_114
authors	Huang, Sheng-Yang, Wang, Yuankai and Jiang, Qingrui
year	2023
title	(In)Visible Cities: Exploring generative artificial intelligence'screativity through the analysis of a conscious journey in latent space
source	García Amen, F, Goni Fitipaldo, A L and Armagno Gentile, Á (eds.), Accelerated Landscapes - Proceedings of the XXVII International Conference of the Ibero-American Society of Digital Graphics (SIGraDi 2023), Punta del Este, Maldonado, Uruguay, 29 November - 1 December 2023, pp. 717–728
summary	The rise of generative AI has redefined architectural design by introducing latent space, challenging traditional methods. This paper aims to explore, structure, and analyse latent journeys, drawing from analytical design discourses. We construct journeys towards 'Isaura' from 'Invisible Cities' by Italo Calvino, bridging literature and visual narratives, utilising the text-image generating software, Midjourney. The objective is to identify spatial configurations that align with the designer's interpretation of the text, ensuring the accuracy of visual elements. Structured as a Markov (stochastic) process, the experiment encompasses four primary stages to offer a rational explanation for the journey and the role of each segment. Findings emphasise the potential of latent space in augmenting architectural design and underscore the necessity for analytical tools to avert the reduction of design to trivial formalism. The study's outcome suggests that understanding and leveraging the traits of latent space can nurture a more meaningful engagement with AI-driven design, presenting a novel approach to architectural creativity.
keywords	Latent Space, Generative Artificial Intelligence, Text-to-image Generation, Architectural Creativity, Spatial Analysis
series	SIGraDi
email
last changed	2024/03/08 14:07

_id	ijac202321203
id	ijac202321203
authors	Kudless, Andrew
year	2023
title	Hierarchies of bias in artificial intelligence architecture: Collective, computational, and cognitive
source	International Journal of Architectural Computing 2023, Vol. 21 - no. 2, 256–279
summary	This paper examines the prevalence of bias in artificial intelligence text-to-image models utilized in the architecture and design disciplines. The rapid pace of advancements in machine learning technologies, particularly in text-to-image generators, has significantly increased over the past year, making these tools more accessible to the design community. Accordingly, this paper aims to critically document and analyze the collective, computational, and cognitive biases that designers may encounter when working with these tools at this time. The paper delves into three hierarchical levels of operation and investigates the possible biases present at each level. Starting with the training data for large language models (LLM), the paper explores how these models may create biases privileging English-language users and perspectives. The paper subsequently investigates the digital materiality of models and how their weights generate specific aesthetic results. Finally, the report concludes by examining user biases through their prompt and image selections and the potential for platforms to perpetuate these biases through the application of user data during training.
keywords	Bias in artificial intelligence, language bias, aesthetic bias, latent diffusion models, digital materiality
series	journal
last changed	2024/04/17 14:30

_id	sigradi2023_416
id	sigradi2023_416
authors	Machado Fagundes, Cristian Vinicius, Miotto Bruscato, Léia, Paiva Ponzio, Angelica and Chornobai, Sara Regiane
year	2023
title	Parametric environment for internalization and classification of models generated by the Shap-E tool
source	García Amen, F, Goni Fitipaldo, A L and Armagno Gentile, Á (eds.), Accelerated Landscapes - Proceedings of the XXVII International Conference of the Ibero-American Society of Digital Graphics (SIGraDi 2023), Punta del Este, Maldonado, Uruguay, 29 November - 1 December 2023, pp. 1689–1698
summary	Computing has been increasingly employed in design environments, primarily to perform calculations and logical decisions faster than humans could, enabling tasks that would be impossible or too time-consuming to execute manually. Various studies highlight the use of digital tools and technologies in diverse methods, such as parametric modeling and evolutionary algorithms, for exploring and optimizing alternatives in architecture, design, and engineering (Martino, 2015; Fagundes, 2019). Currently, there is a growing emergence of intelligent models that increasingly integrate computers into the design process. Demonstrating great potential for initial ideation, artificial intelligence (AI) models like Shap-E (Nichol et al., 2023) by OpenAI stand out. Although this model falls short of state-of-the-art sample quality, it is among the most efficient orders of magnitude for generating three-dimensional models through AI interfaces, offering practical balance for certain use cases. Thus, aiming to explore this gap, the presented study proposes an innovative design agency framework by employing Shap-E connected with parametric modeling in the design process. The generation tool has shown promising results; through generations of synthetic views conditioned by text captions, its final output is a mesh. However, due to the lack of topological information in models generated by Shap-E, we propose to fill this gap by transferring data to a parametric three-dimensional surface modeling environment. Consequently, this interaction's use aims to enable the transformation of the mesh into quantifiable surfaces, subject to collection and optimization of dimensional data of objects. Moreover, this work seeks to enable the creation of artificial databases through formal categorization of parameterized outputs using the K-means algorithm. For this purpose, the study methodologically orients itself in a four-step exploratory experimental process: (1) creation of models generated by Shap-E in a pressing manner; (2) use of parametric modeling to internalize models into the Grasshopper environment; (3) generation of optimized alternatives using the evolutionary algorithm (Biomorpher); (4) and classification of models using the K-means algorithm. Thus, the presented study proposes, through an environment of internalization and classification of models generated by the Shap-E tool, to contribute to the construction of a new design agency methodology in the decision-making process of design. So far, this research has resulted in the generation and classification of a diverse set of three-dimensional shapes. These shapes are grouped for potential applications in machine learning, in addition to providing insights for the refinement and detailed exploration of forms.
keywords	Shap-E, Parametric Design, Evolutionary Algorithm, Synthetic Database, Artificial Intelligence
series	SIGraDi
email
last changed	2024/03/08 14:09

_id	acadia23_v2_560
id	acadia23_v2_560
authors	Saldana Ochoa, Karla; Huang, Lee-Su; Guo, Zifeng; Bokhari, Adil
year	2023
title	Playing Dimensions: Images / Models / Maps: Conceptualizing Architecture with Big Data and Artificial Intelligence
source	ACADIA 2023: Habits of the Anthropocene: Scarcity and Abundance in a Post-Material Economy [Volume 2: Proceedings of the 43rd Annual Conference for the Association for Computer Aided Design in Architecture (ACADIA) ISBN 979-8-9860805-9-8]. Denver. 26-28 October 2023. edited by A. Crawford, N. Diniz, R. Beckett, J. Vanucchi, M. Swackhamer 560-568.
summary	This article presents a novel architecture design workflow that explores the intersection of Big Data, Artificial Intelligence (AI), and storytelling by scraping, encoding, and mapping data, which can then be implemented through Virtual Reality (VR) and Augmented Reality (AR) technologies. In contrast to conventional approaches that consider AI solely as an optimization tool, this workflow embraces AI as an instrument for critical thinking and idea generation. Rather than creating new AI models, this workflow encourages architects to experiment with existing ones as part of their practice. The workflow revolves around the concept of ""Canonical architecture,"" where data-driven techniques serve to traverse dimensions and representations, encompassing text, images, and 3D objects. The data utilized consists of information specific to the project, gathered from social media posts, including both images and text, which provide insights into user needs and site charac- teristics. Additionally, roughly 9,000 3D models of architectural details extracted from 38 different architectural projects were used. The primary objective is to assist architects in developing a workflow that does not suggest starting from scratch or a tabula rasa, but to work with already hyper-connected objects, be it text, images, 3D models, et cetera. These conceptualizations can then be enacted in game engines and/or experimented with in AR/ VR platforms, while keeping their connections alive. Through this process, the framework aims to develop a sensibility of working with large amounts of data without losing focus, and letting the electric grounds of the internet help us in articulating projects.
series	ACADIA
type	paper
email
last changed	2024/04/17 13:59

_id	ascaad2023_134
id	ascaad2023_134
authors	Salman, Huda; Dounas, Theodoros; Clarke, Connor
year	2023
title	Fluency of Creative Ideas in the Digital Age: Exploring Emergent AI Influences on Design Methodology and Visual Thinking in Architectural Education
source	C+++: Computation, Culture, and Context – Proceedings of the 11th International Conference of the Arab Society for Computation in Architecture, Art and Design (ASCAAD), University of Petra, Amman, Jordan [Hybrid Conference] 7-9 November 2023, pp. 815-832.
summary	Research has explored the concept of originality in visual thinking and architectural education, using different methods. The new state of Artificial Intelligence (AI) in architectural design represents another shift from traditional modes of architectural design and education, into a more authentic approach to the digital age. An experiment is designed to highlight the originality of this approach in design thinking and its futuristic trends and impact on education and creativity studies. The intent of the study we present here is twofold: one to revisit key design studies of design exploration and secondly to explore students' design activity while interacting with text-to-image diffusion machine learning (ML) generative models such as Midjourney, DALL-E and Stable Diffusion, as these might have the potential to change the way that architectural students approach the concept stages of designing projects and products. In addition, we are interested in how the new shift in interfaces and modes of stimulus will influence the students' design process and perceptions. Participants in the design process are final year students who had spent at least four years in a school of architecture and can be classified as semi-experienced designers. Further within the evaluation also lies a critique of the diffusion ML tools themselves as producers of architectonic images, rather than complete concepts for architecture that encapsulate spatial, formal, structural arrangements of elements.
series	ASCAAD
email
last changed	2024/02/13 14:41

_id	acadia23_v2_616
id	acadia23_v2_616
authors	Kuang, Zheyuan; Zhang, Jiaxin; Huang, Yiying; Li, Yunqin
year	2023
title	Advancing Urban Renewal: An Automated Approach to Generating Historical Arcade Facades with Stable Diffusion Models
source	ACADIA 2023: Habits of the Anthropocene: Scarcity and Abundance in a Post-Material Economy [Volume 2: Proceedings of the 43rd Annual Conference for the Association for Computer Aided Design in Architecture (ACADIA) ISBN 979-8-9860805-9-8]. Denver. 26-28 October 2023. edited by A. Crawford, N. Diniz, R. Beckett, J. Vanucchi, M. Swackhamer 616-625.
summary	Urban renewal and transformation processes necessitate the preservation of the histor- ical urban fabric, particularly in districts known for their architectural and historical significance. These regions, with their diverse architectural styles, have traditionally required extensive preliminary research, often leading to subjective results. However, the advent of machine learning models has opened up new avenues for generating building facade images. Despite this, creating high-quality images for historical district renovations remains challenging, due to the complexity and diversity inherent in such districts. In response to these challenges, our study introduces a new methodology for automatically generating images of historical arcade facades, utilizing Stable Diffusion models conditioned on textual descriptions. By classifying and tagging a variety of arcade styles, we have constructed several realistic arcade facade image datasets. We trained multiple low-rank adaptation (LoRA) models to control the stylistic aspects of the gener- ated images, supplemented by ControlNet models for improved precision and authenticity. Our approach has demonstrated high levels of precision, authenticity, and diversity in the generated images, showing promising potential for real-world urban renewal projects. This new methodology offers a more efficient and accurate alternative to conventional design processes in urban renewal, bypassing issues of unconvincing image details, lack of precision, and limited stylistic variety. Future research could focus on integrating this two-dimensional image generation with three-dimensional modeling techniques, providing a more comprehensive solution for renovating architectural facades in historical districts.
series	ACADIA
type	paper
email
last changed	2024/04/17 13:59

_id	ecaade2023_250
id	ecaade2023_250
authors	Sebestyen, Adam, Özdenizci, Ozan, Hirschberg, Urs and Legenstein, Robert
year	2023
title	Generating Conceptual Architectural 3D Geometries with Denoising Diffusion Models
source	Dokonal, W, Hirschberg, U and Wurzer, G (eds.), Digital Design Reconsidered - Proceedings of the 41st Conference on Education and Research in Computer Aided Architectural Design in Europe (eCAADe 2023) - Volume 2, Graz, 20-22 September 2023, pp. 451–460
doi	https://doi.org/10.52842/conf.ecaade.2023.2.451
summary	Generative deep learning diffusion models have been attracting mainstream attention in the field of 2D image generation. We propose a prototype which brings a diffusion network into the third dimension, with the purpose of generating geometries for conceptual design. We explore the possibilities of generating 3D datasets, using parametric design to overcome the problem of the current lack of available architectural 3D data suitable for training neural networks. Furthermore, we propose a data representation based on volumetric density grids which is applicable to train diffusion networks. Our early prototype demonstrates the viability of the approach and suggests future options to develop deep learning generative 3D tools for architectural design.
keywords	Artificial Intelligence, Generative Deep Learning, Neural Networks, Diffusion Models, Parametric Design, 3D Data Representations
series	eCAADe
email
last changed	2023/12/10 10:49

_id	sigradi2023_65
id	sigradi2023_65
authors	Cheung, Lok Hang, Dall'Asta, Juan Carlos and Di Marco, Giancarlo
year	2023
title	Exploring Large Language Model as a Design Partner through Verbal and Non-verbal Conversation in Architectural Design Process
source	García Amen, F, Goni Fitipaldo, A L and Armagno Gentile, Á (eds.), Accelerated Landscapes - Proceedings of the XXVII International Conference of the Ibero-American Society of Digital Graphics (SIGraDi 2023), Punta del Este, Maldonado, Uruguay, 29 November - 1 December 2023, pp. 1059–1070
summary	This paper proposes a framework for applying Large Language Models (LLM) as a design partner in architectural design processes instead of a passive question-answering machine. The proposed design framework integrates LLM and Conversation Theory (CT) into a standard parametric design tool for architectural designers. The program establishes an ongoing conversation with the designer through verbal and non-verbal feedback by tracking brain activity and modelling commands. The program can collect conversation data for fine-tuning, thus progressively improving conversation effectiveness. The paper contributes to the knowledge area of architectural design by introducing a novel approach to integrating LLM and CT into the design process, simulated as a proof-of-concept pilot study within a commonly used design software.
keywords	Large Language Model, Human-Computer Interaction, Conversation Theory, Architectural Design Process
series	SIGraDi
email
last changed	2024/03/08 14:08

For more results click below: