Top Features of DALL·E 3*
Next AI art generator
DALL·E 3 is an impersonation AI art machine. It is a tool for generating images, videos, and digital 4D collages in 8K resolution. It's not just another AI art generator. It's a transformation engine that can improve to varying degrees any kind of pictures, videos, and texts in digital form through the process of machine learning. 144 billion-parameter AI art model *
Let us learn from the future, to profit from the present ; )
DALL-E 2 is controlled by a 3.5-billion-parameter model trained on thousands of pairs of photos and descriptions from the internet. This method lets the model connect visual concepts and texts describing them. CLIP and DALL-E 2 datasets (approximately 650 million images in total) GPT-3 has 175 B Parameters
I have been using DALL-E 2 for two months and have experienced some things confirming that it is worth waiting for the next releases - DALL-E 3.
1. Licensing Restrictions
2. Lack of "full" control over the generated output
3. A recognizable dataset
Conceptual work on the DALL-E 2 is most appreciated. 1024 x 1024 px format is significant, but it's not enough for professional issues. Creating a concept that is, in my opinion, the best practical application.
Prohibited DALL-E 2
Restrictions related to the "forbidden dictionary" deprive artists of their creative freedom. Using multiple words results in the account being blocked. As a free artist, I feel uncomfortable self-censoring with every generation. 50 generations a day also limit your freedom.
Generating images based on the description produces images comparable to those of alternative art generators but without copyright or censorship limitations.
Instagrammable
The DALL-E 2 datasets - total 650 million images. That's a lot, but still too few not to have a "typical' visual style. A trained eye will recognize the bias of DALL-E 2. The digital sophistication of the last Instagram decade is too visible. Understandably, this is the most accessible data "material". Documentation in the form of art datasets is poor and very directed art of the northern hemisphere.
Will DALL-E 2 replace artists?
Every tool has limitations, and creativity in its use depends on the creator. DALL-E 2 is a great way to take the following steps in the art of AI. I am very excited to see what the future will bring. The recent articles on MLearning.ai show that we are only at the beginning of our AI art journey.
20 billion-parameter AI art model Text-to-Image model PARTI
*DALL·E 3 does not exist, the title quote is generated using GPT 3