Every frame of this video is imaginary: The places don't exist in the real world. And for every frame, the same (random) inputs are given to four different models -- one making bedrooms, one making dining rooms, one making kitchens, and one making living rooms.

6:53 AM Β· Feb 24, 2019

54
1,566
242
4,500
145,908
This Tweet is unavailable.
I love everything about this video so much. Thanks to @gwern for the suggestion of using transfer learning -- this is why the models produce things that have many of the same features, down to the pattern of the bed turning into a couch into a counter.
4
1
0
79
The results of this training are now integrated into thisairbnbdoesnotexist.com; you can now get a collection of photos that include rooms other than bedrooms.
2
36
8
127
If you're interested in seeing how the model changed over time, you can check out my new "Transfer Demo" page, which shows how the computer learned to "see" different room styles over time. thisairbnbdoesnotexist.com/t…
0
7
0
50
The crazy inception par of this is that if the training images were from furniture catalogs, they were probably computer generated also. Deep learning AI learning to make fake images of fake rooms based on fake images of fake rooms.
1
2
0
46
The source data for the LSUN image classification challenge -- according to the paper -- was Google image search, including a wide range of adjectives (e.g. "messy living room", "tidy living room", "sunny living room", etc.)
1
1
0
10
Replying to @crschmidt
my favorite part is the swooping beds
1
0
0
17
This Tweet was deleted by the Tweet author.
Replying to @crschmidt
Who made this?
5
2
0
64
Generative adversarial networks. One deep net generates something that looks real (like someone who forges art). Another deep net is the art critic and tries to figure out if an image was from a real dataset or generated by the other network.
0
1
0
1