OpenAI, a non-profit organization working in the field of artificial intelligence, has introduced a new version of its neural network called DALL-E. It should be noted that, This neural network is able to generate photos thanks to a simple text description.
Lent.az reports citing foreign media that, DALL-E 2 has more opportunities and Among those opportunities is the ability to work with ready-made photos. The main difference of the new algorithm from the previous version is that that, it includes the ability to edit ready-made photos. Previously, this neural network could only take photos based on text descriptions. Now, thanks to it, you can turn a text image into a deeper photo. The user can select any part of the photo and the algorithm will send a request for the necessary changes in the selected part.
Photo generated by the algorithm for querying a shiba-inu dog in a long-sleeved red sweater with a beret:
For example, this algorithm replaces the table in the photo with another and or generally delete and can add any other object to the table. The algorithm can edit objects by taking into account details such as the direction of shadows in the room. Another feature that DALL-E 2 has is similar to a non-existent photo search tool. In this case the user enters the system and or uploads another photo and The neural network, in turn, resembles a loaded photo. The created photos have a size of 1024 × 1024 pixels.
Photo generated by the algorithm for requesting a monster-like soup bowl woven from wool:
The DALL-E 2 neural network is based on a technology called CLIP. This is a computer vision system. OpenAI announced it last year. The neural network is instructed on the basis of a database that excludes unwanted material. That’s it according to also from the possibility of creating prohibited content free has been published. Using a function designed to protect against abuse, a neural network cannot generate recognizable faces based on any names. DALL-E 2 is currently undergoing a closed beta test. However, the developers of OpenAI noted that, In the near future it will be made available for public use and it will be possible to use it in various application formats.
NETWORK -VIDEO that can create photos based on simple texts
Published an article entitled “lent.az”