Recently, Midjourney released the 5th version of its eponymous artificial intelligence generation tool as beta. In the announcement made on Twitter, it was stated that the final version has much improved image quality, more diverse results, wider range of styles, seamless textures and other improvements. We can say that these discourses are of incomprehensible importance on paper, because the images produced with Midjourney v5 have reached an “incredible” level in the literal sense of the word.Version 5 of Midjourney released
Even though Midjourney was already showing great results in the v4 version, it often had difficulties with the hands and eyes. When you asked for a landscape picture, yes, the results were excellent, but when you asked for a human picture, there were occasional distortions in the hands and face. In the new version, artificial intelligence is now much more advanced. Therefore, the hands and facial features are extremely realistic. The images created are so realistic that you often think they were taken by a great photographer.The difference is huge
With Midjourney v5, AI art enthusiasts are also divided. However, this disagreement is not because the new version is bad, but because it is very successful. While some reviewers said “scary” and “too perfect” about the new version, others could not hide their admiration. You can see the difference between the previous versions and the latest version in the image below. The same prompt was used to produce this image: “A muscular barbarian with a gun next to a CRT television set, cinematic, 8K, studio lighting”
Spooky factor aside, the Midjourney v5’s quality has improved significantly compared to the V4. Graphic designer Julie Wieland has been using Midjourney v4 (released last November) for a while and says version 5 has “incredibly realistic” skin textures. Lighting effects are also much better, including reflections, flares, and shadows. Perhaps most importantly, AI often produces hands and eyes that look natural.
Midjourney v5 is as if you didn’t wear your glasses for a long time and then put them on again. However, the new version has increased the native resolution from 512×512 pixels to 1024×1024 pixels. In addition, the rendering of images therefore takes a little more time than before. Midjourney works like image synthesizers like Stable Diffusion and DALL-E, generating images based on text descriptions called “promises” using an artificial intelligence model trained on millions of man-made artworks. If you want to try it, you can generate “prompts” by logging into the Discord servers here. In the gallery below, you can take a look at the images produced by Midjourney v5.
29938
