The Midjourney V5 Model is the latest and most advanced model released on March 15, 2023.
This model has very high coherence, is excellent at interpreting natural language prompts, has higher output resolution, and supports advanced features such as repeating patterns.
The two images show the comparison of the four generated suggestion images for the same prompt. (without considering the same seed no.*)
*Seed No.: The midjourney bot uses a seed number to generate a field of visual noise, as a starting point for generating the initial image grids. Seed numbers are randomly generated for each image, but can be fixed. Using the same seed number and prompt will result in identical final images using the same model.
Image 1/left: with MJ version 4 | Image 2/right: with MJ version 5
The MJ Model 5 has proven to be a significant advance in AI-assisted image creation. My first impressions show that it produces remarkably photorealistic results and interprets the prompt more accurately, although it also appears less perfect.
My first impression is that the MJ Model 5 analyzes prompts extremely sensitively. With longer, more detailed prompts, I get more successful results here. Those who prefer more openness in the results will probably be more satisfied with the results of the MJ Model 4.
To fully exploit the potential of the new model, I plan to conduct further tests and analyze its performance in various scenarios. It would be interesting to discover in which application areas the MJ Model 5 comes into its own particularly well and its strengths can be used effectively.
New to the Midjourney Model 5
Midjourney Model Version 5 generates higher resolution proposals (by default) 1024 x 1024 px already in the image grid, without requiring an additional step to upscale each image. When using Midjourney Model Version 5, the U1, U2, U3, U4 buttons under each image grid separate the selected image from the original image grid, so they do not scale them up as they did in Model Version 4. The file format remains as before: .PNG.
Let's take a closer look. Here are the maximum image sizes in 3:2 format.
MJ V4, default upscaled: 1344 x 896 px (3:2 format)
MJ V5, proposal is already 1536 x 1024 px (3:2 format)
If we now zoom into the images, we can see the differences very well, especially in the books. There my statements made before are confirmed.
Picture left with MJ V 4 | Picture right with MJ V5
To use the latest model, the --v 5 parameter must be added to the end of the prompt, or use "5️⃣ MJ Version 5" in the /settings.
Have you already worked with Midjourney and especially the Model 5? I'm eager to hear your thoughts and experience. Insights are valuable and help us all better understand how AI-powered image creation affects our creative processes. What are your thoughts on this?
I look forward to your input!