But before that let me first let you know what has worked with the new version and what hasn’t.
What Works:
1. Photorealism: The core success of V5 lies in its Photorealism. The images are so real and photorealistic that you will feel that you can even feel the pores of the skins.
2. Hands, Fingers, and Limbs:: Yes, in this version Midjourney did fix the Wobbly fingers, twisted hands, and limbs issues that plagued the earlier versions of Midjourney. A big Shout out to the Midjourney team. Awesome job here.
4. Aspect Ratio: In V4 we struggled for a higher aspect ratio for most of the time, later they supported 1:2 and 2:1 and later they lift the bar for 16:9 as well as 9:16, but in V5 we can use almost any aspect ratio and even go till 4:1, 5:1 ratios. That means even cinematic renditions of images, like anamorphic frame formats ( ratios 2.39:1 ) which are used in cinematic creations, are now possible in V5. –ar 5:1
5. Tile: In V4 they had removed the possibility of getting Tiled patterns for some reason that was available in the older version of Midjourney V3. Now with this version V5 alpha version, they have brought in the Tile pattern again. –tile
5. Tile: In V4 they had removed the possibility of getting Tiled patterns for some reason that was available in the older version of Midjourney V3. Now with this version V5 alpha version, they have brought in the Tile pattern again. –tile
Image Weights Now you can have full control of image weights that were earlier removed from V4 –iw 0 to 2
v5 is “less opinionated” (They did mention this in the official Release Document) What does that mean? This means that in V4 we had to only write smaller text prompts, and Midjourney used to do the heavy lifting.But in V5, we have to write more detailed prompts that give more details to Midjourney about what the lighting, style, or mood of the output image should look like. If we do not put anything, the default style will be Photorealism.
However, there is more room for control of the output generation through extensive text prompts. That means, there is more bond between the input text and the output image.
What Needs Improvement:
Midjourney has already released the statement that V5 is still in the Alpha state and is open to further improvements. However, in the current state, many creators feel that the way they used to prompt craft in V4 is not working the same way in V4.
Many have trained Chat GPT 3.5 to create Prompt for Midjourney V4 but they are not working in the current version. We have noticed that V5 as a platform is extremely plain Jane and if NOT prompted will create Photo realistic images by default. If we want it to generate artistic images or if we want more creative freedom, then we have to train it with extremely detailed prompts. This is one of the new challenges which creators have to face who have been suitably trained with version 4 of Midjourney.
The next big thing where V5 still lacks is Typography. Midjourney V5 still rolls out Garbled Texts when the character is more than one in a word. While, Adobe Firefly has already claimed to address the issue and they are the first in the line, we users of Midjourney have to wait and watch how fast Midjourney can role out the Typography support in their platform.
Will it be V5a or V5b or a complete new version of Midjourney, only time can tell. Till then we creators have to keep our fingers crossed. Well, did I already tell you that the crossed fingers have already been addressed by Midjourney in V5 version.
