Iterative Visual Interaction with Latent Diffusion Models

Luca Sacchetto; Stefan Röhrl; Klaus Diepold

doi:10.1007/978-3-031-60606-9_24

Iterative Visual Interaction with Latent Diffusion Models

Luca Sacchetto
, Stefan Röhrl
, Klaus Diepold

Chair of Data Processing

Technical University of Munich

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

Image synthesis generative Artificial Intelligence has the potential to revolutionize many industries, from rapid prototyping of consumer goods’ design to creating works of art. However, the most widespread of these models come in the form of text-to-image models. Controlling their output is often a difficult and imprecise process; indeed, slightly different prompts can lead to significantly different outcomes. Therefore, we develop a novel way to interact with generative AI that mimics the way in which humans naturally create works of art and design. We achieve this more granular and intuitive interaction method by exploiting the latent space of Latent Diffusion Models to generate image variations.

Original language	English
Title of host publication	Artificial Intelligence in HCI - 5th International Conference, AI-HCI 2024, Held as Part of the 26th HCI International Conference, HCII 2024, Proceedings
Editors	Helmut Degen, Stavroula Ntoa
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	411-421
Number of pages	11
ISBN (Print)	9783031606052
DOIs	https://doi.org/10.1007/978-3-031-60606-9_24
State	Published - 2024
Event	5th International Conference on Artificial Intelligence in HCI, AI-HCI 2024, held as part of the 26th HCI International Conference, HCII 2024 - Washington, United States Duration: 29 Jun 2024 → 4 Jul 2024

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14734 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	5th International Conference on Artificial Intelligence in HCI, AI-HCI 2024, held as part of the 26th HCI International Conference, HCII 2024
Country/Territory	United States
City	Washington
Period	29/06/24 → 4/07/24

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 9 Industry, Innovation, and Infrastructure

Keywords

Generative AI
Human Centered AI
Latent Diffusion Models
Latent Space

Access to Document

10.1007/978-3-031-60606-9_24

Cite this

Sacchetto, L., Röhrl, S., & Diepold, K. (2024). Iterative Visual Interaction with Latent Diffusion Models. In H. Degen, & S. Ntoa (Eds.), Artificial Intelligence in HCI - 5th International Conference, AI-HCI 2024, Held as Part of the 26th HCI International Conference, HCII 2024, Proceedings (pp. 411-421). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14734 LNAI). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-60606-9_24

Sacchetto, Luca ; Röhrl, Stefan ; Diepold, Klaus. / Iterative Visual Interaction with Latent Diffusion Models. Artificial Intelligence in HCI - 5th International Conference, AI-HCI 2024, Held as Part of the 26th HCI International Conference, HCII 2024, Proceedings. editor / Helmut Degen ; Stavroula Ntoa. Springer Science and Business Media Deutschland GmbH, 2024. pp. 411-421 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{7938b082458346e9917382e94720210a,

title = "Iterative Visual Interaction with Latent Diffusion Models",

abstract = "Image synthesis generative Artificial Intelligence has the potential to revolutionize many industries, from rapid prototyping of consumer goods{\textquoteright} design to creating works of art. However, the most widespread of these models come in the form of text-to-image models. Controlling their output is often a difficult and imprecise process; indeed, slightly different prompts can lead to significantly different outcomes. Therefore, we develop a novel way to interact with generative AI that mimics the way in which humans naturally create works of art and design. We achieve this more granular and intuitive interaction method by exploiting the latent space of Latent Diffusion Models to generate image variations.",

keywords = "Generative AI, Human Centered AI, Latent Diffusion Models, Latent Space",

author = "Luca Sacchetto and Stefan R{\"o}hrl and Klaus Diepold",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.; 5th International Conference on Artificial Intelligence in HCI, AI-HCI 2024, held as part of the 26th HCI International Conference, HCII 2024 ; Conference date: 29-06-2024 Through 04-07-2024",

year = "2024",

doi = "10.1007/978-3-031-60606-9\_24",

language = "English",

isbn = "9783031606052",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "411--421",

editor = "Helmut Degen and Stavroula Ntoa",

booktitle = "Artificial Intelligence in HCI - 5th International Conference, AI-HCI 2024, Held as Part of the 26th HCI International Conference, HCII 2024, Proceedings",

}

Sacchetto, L, Röhrl, S & Diepold, K 2024, Iterative Visual Interaction with Latent Diffusion Models. in H Degen & S Ntoa (eds), Artificial Intelligence in HCI - 5th International Conference, AI-HCI 2024, Held as Part of the 26th HCI International Conference, HCII 2024, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14734 LNAI, Springer Science and Business Media Deutschland GmbH, pp. 411-421, 5th International Conference on Artificial Intelligence in HCI, AI-HCI 2024, held as part of the 26th HCI International Conference, HCII 2024, Washington, United States, 29/06/24. https://doi.org/10.1007/978-3-031-60606-9_24

Iterative Visual Interaction with Latent Diffusion Models. / Sacchetto, Luca; Röhrl, Stefan; Diepold, Klaus.
Artificial Intelligence in HCI - 5th International Conference, AI-HCI 2024, Held as Part of the 26th HCI International Conference, HCII 2024, Proceedings. ed. / Helmut Degen; Stavroula Ntoa. Springer Science and Business Media Deutschland GmbH, 2024. p. 411-421 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14734 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Iterative Visual Interaction with Latent Diffusion Models

AU - Sacchetto, Luca

AU - Röhrl, Stefan

AU - Diepold, Klaus

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

PY - 2024

Y1 - 2024

N2 - Image synthesis generative Artificial Intelligence has the potential to revolutionize many industries, from rapid prototyping of consumer goods’ design to creating works of art. However, the most widespread of these models come in the form of text-to-image models. Controlling their output is often a difficult and imprecise process; indeed, slightly different prompts can lead to significantly different outcomes. Therefore, we develop a novel way to interact with generative AI that mimics the way in which humans naturally create works of art and design. We achieve this more granular and intuitive interaction method by exploiting the latent space of Latent Diffusion Models to generate image variations.

AB - Image synthesis generative Artificial Intelligence has the potential to revolutionize many industries, from rapid prototyping of consumer goods’ design to creating works of art. However, the most widespread of these models come in the form of text-to-image models. Controlling their output is often a difficult and imprecise process; indeed, slightly different prompts can lead to significantly different outcomes. Therefore, we develop a novel way to interact with generative AI that mimics the way in which humans naturally create works of art and design. We achieve this more granular and intuitive interaction method by exploiting the latent space of Latent Diffusion Models to generate image variations.

KW - Generative AI

KW - Human Centered AI

KW - Latent Diffusion Models

KW - Latent Space

UR - https://www.scopus.com/pages/publications/85196191131

U2 - 10.1007/978-3-031-60606-9_24

DO - 10.1007/978-3-031-60606-9_24

M3 - Conference contribution

AN - SCOPUS:85196191131

SN - 9783031606052

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 411

EP - 421

BT - Artificial Intelligence in HCI - 5th International Conference, AI-HCI 2024, Held as Part of the 26th HCI International Conference, HCII 2024, Proceedings

A2 - Degen, Helmut

A2 - Ntoa, Stavroula

PB - Springer Science and Business Media Deutschland GmbH

T2 - 5th International Conference on Artificial Intelligence in HCI, AI-HCI 2024, held as part of the 26th HCI International Conference, HCII 2024

Y2 - 29 June 2024 through 4 July 2024

ER -

Sacchetto L, Röhrl S, Diepold K. Iterative Visual Interaction with Latent Diffusion Models. In Degen H, Ntoa S, editors, Artificial Intelligence in HCI - 5th International Conference, AI-HCI 2024, Held as Part of the 26th HCI International Conference, HCII 2024, Proceedings. Springer Science and Business Media Deutschland GmbH. 2024. p. 411-421. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-60606-9_24