Session:Demystifying Deepfakes

From 36C3 Wiki
Revision as of 20:33, 26 December 2019 by A0b1 (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Description I wanted to know what the fuzz is all about in the newspapers and tried to understand some details in deepfake video generation. This talk aims to be slightly more technical than a youtube introduction on deepfakes and slightly less useful for creating your own deepfake video.
Website(s)
Type Talk
Kids session No
Keyword(s) political, software, science
Tags deepfakes, faceswap, educational
Person organizing A0b1
Language en - English
en - English
Other sessions... ... further results

(Click here to refresh this page.)

Starts at 2019/12/xx xx:00
Ends at Error: Invalid time.
Duration 60 minutes
Location Room:Seminar room 13
<ul><li>"2019/12/xx xx:00" contains an extrinsic dash or other characters that are invalid for a date interpretation.</li> <!--br--><li>"Error: Invalid time." contains an extrinsic dash or other characters that are invalid for a date interpretation.</li></ul>

Newspaper articles on deepfakes are usually titled something like "The very serious dangers of deepfakes", "How the internet will be destroyed in 5 seconds" or, at best, "Deepfakes - should we be concerned? Spoiler: yes". This is the talk of a concerned citizen with not enough computing power wanting to find out what we are actually talking about when we are talking deepfakes. I will go through the steps involved when creating a deepfake including some technical details and some common sense. Deep learning knowledge is not required, but some deep learning may appear.