indexer: notabug.io
all comments

[] -0 points

your best bet is to use an avatar and speech synthesis

[] -0 points

oh, and not use youtube in the first place

[] -0 points

you could also wear a mask. Deep fake or otherwise you cannot create data that isnt present. all deepfakes are at the end of the day is a mask over the original face

[] -0 points

You lose some of the stylistic benefit here as deep fakes respond to facial expression and masks typically do not.

[] -0 points

im saying its the same thing. you are not gonna get the data back that is removed via mask, physical or digital

[] -0 points

In some of the deepfakes I’ve seen the “mask” can sometimes temporarily slip away or glitch out though.

[]x0x7 -0 points

OP here, but the deepfake mask is based in part on the face being deep faked.

It's like an instagram filter for your face that does more complicated things. If you overlay an image with a filter over one without a filter, yes, you are masking the clear image, but the filter still bleeds data through it.

What's debatable is if anyone is smart enough to use it. If you recover an image from an instagram filter at minimum you are going to have image compression and color distortion no matter what process you use, plus a fun coding assignment per instagram filter. Because a deepfake filter changes face shape the question is if it really ever can go back, and what does face shape compression look like?

You can tell who's being deepfaked here: https://www.foreignaffairs.com/sites/default/files/anchor.gif

π