{"version": "1.0", "type": "rich", "title": "we\u2019re going to start using neural networks for image compression and they\u2019ll just hallucinate the details they imagine should be...", "author_name": "kontextmaschine", "author_url": "https://kontextmaschine.com", "provider_name": "kontextmaschine", "provider_url": "https://kontextmaschine.com", "url": "https://kontextmaschine.com/post/694515293637279744/", "html": "

argumate:

baconmancr:

theunlitpath:

argumate:

ralfmaximus:

argumate:

we\u2019re going to start using neural networks for image compression and they\u2019ll just hallucinate the details they imagine should be in the image and I hate it already

It is the year 2030. File size is 0 bytes, but the file named sad_cat_wearing_a_birthday_hat.jpeg renders perfectly.

I can see it in my mind already, I gurgle

You’ll upgrade your codec and it’ll fill in details that weren’t even there in the original! It’ll automatically increase the resolution! Read plausible text on the newspaper sitting in the background of that old photo of grandpa!

Future-future archivists will lament that people aren\u2019t seeing the images as they were rendered under an era-appropriate codex, like people nowadays complaining that pixel art doesn\u2019t look right unless it\u2019s rendered on a CRT

already having that problem with emoji!

", "thumbnail_url": "https://kontextmaschine.com/media/de3a1dbe7be96a53f4fff13da686ced6421ad9a0_35032fb40629.png", "thumbnail_width": 500, "thumbnail_height": 500}