When I try to generate a picture from a Twitter profile using AI 'DALL E 2' that can generate high quality images from sentences, it looks like this



On April 6, 2022, local time, Open AI , a non-profit organization that studies AI, announced the latest version of AI ' DALL E 2 ' that generates images from text. Nick Cammarata, a researcher of OpenAI, is demonstrating using such DALL ・ E 2 'When you try to generate a picture from the text written on your Twitter profile, it looks like this'.



DALL ・ E 2 is an AI that can automatically generate illustrations and photos from a given sentence by combining natural language processing and image generation. In the following articles, you can see the images generated by DALL E 2 from words such as 'astronaut', 'riding a horse', and 'in a realistic style'.

Introducing 'DALL E 2', a high-resolution, low-latency version of 'DALL E' that creates images from input text information --GIGAZINE



Cammarata uses DALL E 2 to post 'pictures generated from a friend's Twitter profile' on Twitter. For example, below is a picture generated from the sentence 'commitments empathetic, psychedelic, philosophical' in Romeo Stevens (@ RomeoStevens76) 's profile. It is a picture that evokes a psychedelic and paranormal image.



The picture generated from Ava (@noampomsky) 's 'bookbear' profile is a lovely little polar bear holding or reading a book.



'Happy sisyphus' in the profile of Will Jack (@willwjack) is a human being who deceived the god in Greek mythology. Sisyphus was ordered to raise a huge rock to the top of the mountain as a punishment for deceiving God twice, but he will repeat this penance forever because the rock will roll down to the bottom shortly after reaching the top of the mountain. .. Looking at the paintings produced by DALL E 2, it was based on the punishment of Sisyphus in mythology.



Andrew Conner (@connerdelights) 's text 'solving the metabolic health crisis' produces a grotesque, abstract atmosphere of human organs ...



Andrés Gómez Emilsson (@algekalipso) 's 'views of a Transhuman neo-Buddhist from the future on sociology, artificial intelligence, mathematics, philosophy, neonoir film, and the post-singularity era.' , Neo-noir film, the view of the transhuman neo-buddist from the future about the era of post-singularity) ”from the profile of the Buddha image with a cyber and mysterious atmosphere. Was generated.



On Twitter of Mr. Cammarata, many pictures generated by DALL ・ E 2 are posted in addition to the ones introduced this time. In addition, Mr. Cammarata revealed that for the picture posted this time, not only entering sentences in DALL E 2 but also specifying specific styles such as 'cyberpunk' and 'oil painting', adding fun and variety. I am.



Hacker News, a social news site, is also a big topic, and users are exchanging various opinions.

Dall-E 2 illustrations of Twitter bios | Hacker News
https://news.ycombinator.com/item?id=30961385

One user commented that the DALL E 2 painting was only trained by a large amount of art, 'this would be garbage for humans.' On the other hand, another user pointed out that 'most of the creative output of human beings is combined with the evolution of existing works', and there is certainly a ' good picture ' in what DALL E 2 generated. Claims to be doing.

Also, although the pictures produced by DALL E 2 are certainly not good, there are also human 'bad artists' in the world, and DALL E 2 can be comparable to those people. Therefore, some comments argued that it was not only technically impressive but also socially important. Some users were surprised that DALL E 2 produced a fairly accurate picture, even from the vague sentence 'Happy Sisyphus'.

Related Posts:

in Software,   Art, Posted by log1h_ik