This Food Does Not Exist
AI
image generation
SHOW HN2022
How HN took it
16 positive, 38 neutral or asking, 3 mixed, 13 critical, out of 70 comments about the project. Thread score +0.04.
Idea: 11 for, 7 againstExecution: 1 for, 4 againstPricing: 1 for, 0 againstDesign: 1 for, 0 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Images look convincingly food-like and appetizing
From 2 comments
Everything looks delish!
ComputerCat on HNThe food looks great! I suppose these models could use some extra training with dishes, though. The plates and glasses look wobbly, which is an instant giveaway. Otherwise, I can…
n4bz0r on HN
Interesting use for food blogs and stock photos
From 2 comments
This is the most disturbing “does not exist” yet. A food blog could write itself
georgeburdell on HNWe're looking at the complete collapse of the stock photography market.
beej71 on HN
Objections
Generated images may closely copy training data
From 2 comments
I wonder how close the nearest match from the training data is. Was there a cheesecake that looked almost like these generated images?
forgotusername6 on HNLooks impressive but I can't escape the notion that surely some of the generated images will be very close to the some of the training images? How am I to assess how original the…
jdthedisciple on HN
Plates and glasses look wobbly, a giveaway
From 1 comment
The food looks great! I suppose these models could use some extra training with dishes, though. The plates and glasses look wobbly, which is an instant giveaway. Otherwise, I can…
n4bz0r on HN
Requests
Way to generate a fresh image on demand
From 1 comment
Is there a way to trigger a fresh image on demand? That's kind of what I expect when I see a does-not-exist site.
herpderperator on HN
Information on training dataset size and licensing
From 2 comments
How big was your training dataset?
Sebbecking on HNand what was the licensing for the training data that you used?
waynesonfire on HN
Questions
How original are outputs compared to training data
From 2 comments
Looks impressive but I can't escape the notion that surely some of the generated images will be very close to the some of the training images? How am I to assess how original the…
jdthedisciple on HNI wonder how close the nearest match from the training data is. Was there a cheesecake that looked almost like these generated images?
forgotusername6 on HN
Are there techniques to detect generated images versus photos
From 1 comment
Are there any analysis techniques that can easily distinguish between these and real photographs? Do simple things like edge detections or histograms reveal any anomalies?
wyldfire on HN
Alternatives named
DALL-E and diffusion models
From 2 comments
At least with DALL-E you can be sure the food has a name. For a moment I was worried this would produce vaguely food-like images where on closer look you realise you have no idea…
xg15 on HNWe have trained four StyleGAN2 image generation models and are releasing checkpoints and training code. We are exploring how to improve/scale up StyleGAN training, particularly…
MasterScrat on HN
Imagen
From 1 comment
I'm honestly surprised that they trained a StyleGAN. Recently, the Imagen architecture has been show to be both easier in structure, easier to train, and even faster to produce…
fxtentacle on HN
thisxdoesnotexist.com collection
From 1 comment
Aggregate "does not exist" website for anyone who's interested. https://thisxdoesnotexist.com/
tmountain on HN