Alias-Free Generative Adversarial Networks

Tero Karras, Miika Aittala, Samuli Laine, Erik Härkönen, Janne Hellsten, Jaakko Lehtinen, Timo Aila

Tutkimustuotos: Artikkeli kirjassa/konferenssijulkaisussaConference article in proceedingsScientificvertaisarvioitu

410 Sitaatiot (Scopus)

Abstrakti

We observe that despite their hierarchical convolutional nature, the synthesis process of typical generative adversarial networks depends on absolute pixel coordinates in an unhealthy manner. This manifests itself as, e.g., detail appearing to be glued to image coordinates instead of the surfaces of depicted objects. We trace the root cause to careless signal processing that causes aliasing in the generator network. Interpreting all signals in the network as continuous, we derive generally applicable, small architectural changes that guarantee that unwanted information cannot leak into the hierarchical synthesis process. The resulting networks match the FID of StyleGAN2 but differ dramatically in their internal representations, and they are fully equivariant to translation and rotation even at subpixel scales. Our results pave the way for generative models better suited for video and animation.

AlkuperäiskieliEnglanti
OtsikkoAdvances in Neural Information Processing Systems 34 - 35th Conference on Neural Information Processing Systems, NeurIPS 2021
ToimittajatMarc'Aurelio Ranzato, Alina Beygelzimer, Yann Dauphin, Percy S. Liang, Jenn Wortman Vaughan
KustantajaNeural Information Processing Systems Foundation
Sivut852-863
Sivumäärä12
ISBN (elektroninen)9781713845393
TilaJulkaistu - 2021
OKM-julkaisutyyppiA4 Artikkeli konferenssijulkaisussa
TapahtumaConference on Neural Information Processing Systems - Virtual, Online
Kesto: 6 jouluk. 202114 jouluk. 2021
Konferenssinumero: 35
https://neurips.cc

Julkaisusarja

NimiAdvances in Neural Information Processing Systems
Vuosikerta2
ISSN (painettu)1049-5258

Conference

ConferenceConference on Neural Information Processing Systems
LyhennettäNeurIPS
KaupunkiVirtual, Online
Ajanjakso06/12/202114/12/2021
www-osoite

Sormenjälki

Sukella tutkimusaiheisiin 'Alias-Free Generative Adversarial Networks'. Ne muodostavat yhdessä ainutlaatuisen sormenjäljen.

Siteeraa tätä