New neural network for photorealistic images beats Midjourney. This is what she can do

No time to read?
Get a summary

The generative model Recraft V3 has appeared online. According to its creators, the latest version sets a new standard in image generation and beats all its competitors, including Midjourney.

The main advantages of Recraft V3 are the quality of text generation and understanding of human anatomy. The creators claim that the model can generate images with long texts, and that a person’s (or other humanoid beings’) hands will always have the correct number of fingers.

Recraft V3 also allows you to specify the precise placement of text, change the visual style, and edit images using numerous tools. The neural network can work with vector images and will therefore especially appeal to graphic designers, the authors think.

New neural network for photorealistic images beats Midjourney. This is what she can do

Recraft V3 can be tried out on the official website. The free version includes 50 credits to generate.

Other images from social networks:

The artist created a portrait of Call of Duty: Black Ops 6 using cracks in glass

Source: VG Times

No time to read?
Get a summary
Previous Article

Three victims were pulled from the rubble in Cherkessk

Next Article

Regina Todorenko danced in her underwear, without make-up or hair styling