Analiza klasičnih portretov v primerjavi s portreti generiranimi z uporabo umetne inteligence s pomočjo tehnologije sledenja očesnih premikov

Krišelj, Lea (author), Starešinič, Marica (mentor), Štampfl, Veronika (comentor)

V diplomskem delu smo primerjali fotografije s slikami, ki so generirane s pomočjo umetne inteligence. V sklopu diplomske naloge smo se odločili izdelati serijo šestih portretov in s pomočjo njihovih opisov generirati slike ter jih primerjati s pomočjo tehnologije sledenja očesnih premikov. V teoretičnem delu smo na kratko predstavili razvoj fotografije in se dotaknili teme o obraznih značilnostih. Poleg tega smo prebrali literaturo o umetni inteligenci in tehnologiji sledenja očesnih premikov. Znanje iz teoretičnega dela smo uporabili pri izdelavi praktičnega dela, kjer smo najprej fotografirali tri moške in tri ženske modele, ki so bili približno iste starosti. Njihove opise smo nato vpisali v program Adobe Firefly, ki nam je umetno generiral slike. Fotografije smo uredili, da so bile čim bolj podobne tistim, ki jih je naredila umetna inteligenca. Nato smo jih pomešali s fotografijami in naredili test za testiranje s tehnologijo sledenja očesnih premikov. Izvedli smo testiranje z napravo Tobii Pro Fusion in s pomočjo fiksacijskih in toplotnih zemljevidov pregledali rezultate.

• Language: Slovenian
• Keywords: fotografija, umetna inteligenca, generirane slike, tehnologija sledenja očesnih premikov, točka pogleda
• Work type: Bachelor thesis/paper
• Organization: NTF - Faculty of Natural Sciences and Engineering
• Year: 2024
• PID: 20.500.12556/RUL-161011
• Publication date in RUL: 06.09.2024

KRIŠELJ, Lea, 2024, Analiza klasičnih portretov v primerjavi s portreti generiranimi z uporabo umetne inteligence s pomočjo tehnologije sledenja očesnih premikov [online]. Bachelor’s thesis. [Accessed 26 April 2025]. Retrieved from: https://repozitorij.uni-lj.si/IzpisGradiva.php?lang=eng&id=161011

Secondary language

• Language: English
• Title: Analysis of traditional portraits versus AI-generated portraits using eye movement tracking technology

Abstract

The thesis compares photographs with images generated by artificial intelligence. As part of the thesis, we decided to create a series of six portraits, generate images using their descriptions and compare them using eye movement tracking technology. In the theoretical part, we briefly introduced the evolution of photography and touched upon the topic of facial features. In addition, we also looked at artificial intelligence and eye tracking technology. We applied our knowledge from the theoretical part to the practical part, where we first photographed three male and three female models of approximately the same age. Their descriptions were then entered into Adobe Firefly, which generated the images artificially. We edited the photos to look as similar as possible to those taken by the AI. We then mixed them with the photos and did a test to test the eye-tracking technology. We carried out the testing with the Tobii Pro Fusion device and reviewed the results using fixation and heat maps.

• Keywords: photography, artificial intelligence, generated images, eye movement tracking technology, gaze point