Utjecaj tehničkih parametara u postprodukciji na kvalitetu primjene deepfake tehnologije i korisničkog iskustva

master's thesis

• Author: Fabijan Nushi
• Mentor: Lidija Mandić (mentor)
• Committee member: Maja Strgar Kurečić (član povjerenstva)
• Committee member: Lidija Mandić (član povjerenstva)
• Committee member: Jesenka Pibernik (član povjerenstva)
• Granter: University of Zagreb; Faculty of Graphic Arts; Zagreb
• Defense date and country: 2021-9-24, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Graphic Technology

Abstract

Sve brži razvoj tehnologije i integracija umjetne inteligencije u digitalnu sferu, dovelo

je do toga da je u današnje vrijeme deepfake sve više prisutan među rastućim brojem

poklonika. Zbog svoje zastupljenosti i poboljšanja tehnologije, deepfake se počeo

koristiti u manipulacijske svrhe na način da se šire kreativni i zabavni sadržaji, ali i

lažne vijesti i teorije zavjere, što je označilo prekretnicu u stvaranju digitalnih

manipulacijskih sadržaja. Deepfake je tehnologija koja se bazira na strojnom učenju

(machine learning), a u posljednje vrijeme je privukla veliku pozornost, naročito od

strane medija zbog mogućnosti stvaranja nevjerojatno realističnih manipuliranih

videozapisa i zvukova. Ti videozapisi su postignuti umjetnom inteligencijom (artificial

intelligence) i pojavljuju se na društvenim mrežama i internetu te mogu imati puno

ozbiljnije posljedice i utjecaj na društvo, politiku, ekonomiju i identitet. Sadašnja i

buduća tehnologija svima će omogućavati da kreiraju lažne videozapise koji će

izgledati stvarno i to bez značajnog ulaganja u obuku i prikupljanje podataka. U ovom

istraživanju su ispitani tehnički parametri; promjena rasvjete, kut te udaljenost

snimanja, pozicija i ekspresija lica, te kompleksnost objekata u doticaju s licem.

Promjena tehničkih parametra različito utječe na kvalitetu, te su provedena subjektivan

ispitivanja korisnika s ciljem utvrđivanja utjecaja istih.

Abstract (english)

The increasingly rapid development of technology and the integration of artificial

intelligence into the digital sphere, has led to the fact that nowadays deepfake is

increasingly present among a growing number of fans. Due to its representation and

improvement in technology, deepfake began to be used for manipulative purposes in a

way to spread creative and entertainment content, but also fake news and conspiracy

theories, which marked a turning point in the creation of digital manipulative content.

Deepfake is a technology based on machine learning and has recently attracted a lot of

attention, especially from the media because of its ability to create incredibly realistic

manipulated videos and sounds. These videos are achieved by artificial intelligence and

appear on social networks and the Internet and can have much more serious consequences

and impact on society, politics, economics, and identity. Current and future technology

will allow everyone to create fake videos that will look real without significant investment

in training and data collection. In this research, the technical parameters were examined;

changes in lighting, angle, and distance of shooting, position and expression of the face,

and the complexity of objects in contact with the face. The change of technical parameters

affects the quality differently, and subjective tests of users were conducted to determine

their impact.

• Keywords: deepfake; tehnički parametri; postprodukcija
• Keywords (english): deepfake; technical parameters; postproduction
• Language: croatian
• URN:NBN: urn:nbn:hr:216:309504
• Study programme: Title: Graphic technology; specializations in: Graphic technology, Design of graphic products; Course: Graphic technology; Study programme type: university; Study level: graduate; Academic / professional title: sveučilišni/a magistar/magistra inženjer/inženjerka grafičke tehnologije (sveučilišni/a magistar/magistra inženjer/inženjerka grafičke tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2024-04-18 23:49:56