Research Data Leeds Repository

Citation

Kelly, Tom and Guerrero, Paul (2019) FrankenGAN Powerpoint presentation. University of Leeds. [Dataset] https://doi.org/10.5518/608

Dataset description

Powerpoint presentation from Siggraph Asia (Conference and Exhibition on Computer Graphics and Interactive Techniques in Asia) 2018, including 3D models and animation.

Coarse building mass models are now routinely generated at scales ranging from individual buildings through to whole cities. For example, they can be abstracted from raw measurements, generated procedurally, or created manually. However, these models typically lack any meaningful semantic or texture details, making them unsuitable for direct display. We introduce the problem of automatically and realistically decorating such models by adding semantically consistent geometric details and textures. Building on the recent success of generative adversarial networks (GANs), we propose FrankenGAN, a cascade of GANs to create plausible details across multiple scales over large neighborhoods. The various GANs are synchronized to produce consistent style distributions over buildings and neighborhoods. We provide the user with direct control over the variability of the output. We allow her to interactively specify style via images and manipulate style-adapted sliders to control style variability. We demonstrate our system on several large-scale examples. The generated outputs are qualitatively evaluated via a set of user studies and are found to be realistic, semantically-plausible, and style-consistent.

Keywords:	generative adversarial networks, urban modeling, computer graphics
Divisions:	Faculty of Engineering and Physical Sciences > School of Computing
Related resources:	Location Type https://vcg.leeds.ac.uk/projects/frankengan/ Website https://doi.org/10.1145/3272127.3275065 Publication https://github.com/twak/chordatlas Software https://github.com/twak/bikegan Software https://web.archive.org/web/20190814135633/https://sa2018.siggraph.org/en/ Website
License:	Creative Commons Attribution 4.0 International (CC BY 4.0)
Date deposited:	18 Nov 2019 15:30
URI:	https://archive.researchdata.leeds.ac.uk/id/eprint/605
Additional details Additional details Creators or authors: Creators ORCID Other ID Email Primary affiliation Primary affiliation ID Primary affiliation ID type Secondary affiliation Secondary affiliation ID Secondary affiliation ID type Kelly, Tom https://orcid.org/0000-0002-6575-3682 University of Leeds https://ror.org/024mrxd33 ror Guerrero, Paul https://orcid.org/0000-0002-7568-2849 University College London https://ror.org/02jx3x895 ror Type of data: Dataset Research funders: ERC Starting Grant, KAUST-UCL, ERC PoC Grant, KAUST Office of Sponsored Research, EPSRC, Google Faculty Award Grant numbers: SmartGeometry StG-2013-335373, OSR-2015-CCF-2533, SemanticCity, OSR-CRG2017-3426, EP/M013685/1, UrbanPlan Publication date: 2019 Resource language: English Metadata language: English Research project title: FrankenGAN: Guided Detail Synthesis for Building Mass Models using Style-Synchonized GANs Publisher: University of Leeds Contact email: T.W.A.Kelly@leeds.ac.uk Last Modified: 24 Jul 2024 13:19

Files

Data

Powerpoint deck presented at Siggraph 2018. [605MB] Powerpoint deck presented at Siggraph 2018. [605MB] Powerpoint deck presented at Siggraph 2018. [605MB] Security: Anyone File name: frankengan_slides.pptx Content type: Data Mime-Type: application/vnd.openxmlformats-officedocument.presentationml.presentation License: Creative Commons Attribution 4.0 International (CC BY 4.0) File size: 605MB

3D models and animation for the Madrid animation. [608MB] 3D models and animation for the Madrid animation. [608MB] 3D models and animation for the Madrid animation. [608MB] Security: Anyone File name: madrid_anim_with_sky.blend Content type: Data Description: 3D models and animation for the Madrid animation License: Creative Commons Attribution 4.0 International (CC BY 4.0) File size: 608MB

3D models and animation for the London Regent Street animation. [919MB] 3D models and animation for the London Regent Street animation. [919MB] 3D models and animation for the London Regent Street animation. [919MB] Security: Anyone File name: regent_anim_slow.blend Content type: Data Description: 3D models and animation for the London Regent Street animation License: Creative Commons Attribution 4.0 International (CC BY 4.0) File size: 919MB

3D models and animation for the distribution examples animation. [763MB] 3D models and animation for the distribution examples animation. [763MB] 3D models and animation for the distribution examples animation. [763MB] Security: Anyone File name: distribution_examples_anim.blend Content type: Data Description: 3D models and animation for the distribution examples animation License: Creative Commons Attribution 4.0 International (CC BY 4.0) File size: 763MB

Network weights for the bikegan network (github.com/twak/bikegan/) [2GB] Network weights for the bikegan network (github.com/twak/bikegan/) [2GB] Network weights for the bikegan network (github.com/twak/bikegan/) [2GB] Security: Anyone File name: bikegan_weights.zip Content type: Data Description: Network weights for the bikegan network (https://github.com/twak/bikegan/) Mime-Type: application/zip License: Creative Commons Attribution 4.0 International (CC BY 4.0) File size: 2GB

3D animation showing results under varying lighting. [62MB] 3D animation showing results under varying lighting. [62MB] 3D animation showing results under varying lighting. [62MB] Security: Anyone File name: lighting_demo_anim.blend Content type: Data Description: 3D animation showing results under varying lighting License: Creative Commons Attribution 4.0 International (CC BY 4.0) File size: 62MB

3D models and animation giving an overview of FrankenGAN's pipeline. [61MB] 3D models and animation giving an overview of FrankenGAN's pipeline. [61MB] 3D models and animation giving an overview of FrankenGAN's pipeline. [61MB] Security: Anyone File name: overview_anim.blend Content type: Data Description: 3D models and animation giving an overview of FrankenGAN's pipeline License: Creative Commons Attribution 4.0 International (CC BY 4.0) File size: 61MB