Midjourney, a firm specializing in the development of artificial intelligence products, did not want to miss the opportunity to overshadow —or, at least, try to— OpenAI and the recent launch of GPT-4. The independent research laboratory headed by David Holz has announced Midjourney V5a new version of its AI capable of generating images by means of a brief text description that comes, among other novelties, with the possibility of create higher quality designs. Although also with some other disadvantage compared to the previous version.
The company has confirmed in a tweet that Midjourney V5, currently available in beta for some users, has “much higher image quality” than Midjourney v4. The new version also allows you to create images with “a wider stylistic range” and supports “fluid textures”, which can make the detail of the designs much more precise and, therefore, generate more realistic photographs.
Midjourney V5, on the other hand, now can output images with wider aspect ratios, as well as with a greater dynamic range. The firm also ensures that the new version of the AI is also capable of receiving better image indications. Everything seems to indicate, yes, that Midjourney V5 is slower processing images compared to Midjourney V4. Mainly, given the significant improvement in quality.
Midjourney V5 generates significantly higher quality images than the previous version
Everything seems to indicate, yes, that Midjourney V5 is slower processing images compared to Midjourney V4. Mainly, given the considerable improvement in quality. ButAre there really important differences in the images generated by Midjourney v5 versus Midjourney v4?
As Phil Desforges, a photographer dedicated to exploring the capabilities of AI, shows, Midjourney V5 takes an incredible leap in just about every type of prompt AI can receive. The function that allows you to generate designs based on the indications of an image, for example, can create a photography with considerably more realistic textures and elements and with better worked shadows.
Midjourney V5 too improves imaging in other types of indications. For example, it is now possible to ask you to create designs in specific formats. One of the best examples to understand the improvements of the new version of the AI, is through these two images that Desforges has generated through the following description: “photo of the street in new york, medium shot, natural lighting, filmed in Jujifilm, detailed and realistic environment, cinematic –Aspect Ratio 2:1”. We can see how the image generated by Midjourney v5 creates much more realistic skin and clothing textures than the one created by version 4.