MidjourneyV5 version of all changes, explain clearly.

MidjourneyV5 version of all changes, explain clearly.

-V5-V5 is the latest version of Mid jou mey! You may hear it called a new model, algorithm, style or data set. You may hear it described as a newly trained, bigger brain. The more it knows, the better it understands. These are the real situations of-v5, and we haven't found any more truth. How to use -v5 There are two absolute ways to ensure that you use -V5: You can go to /settings and select the [MJ Version 5] button. or

You can add the parameter -v5 III -v5 at the end of the prompt. What is the main difference between V5 and other models?

1 improves the rate of national division.

All images made with-v5 have the highest resolution and quality. It's as if they have been automatically upgraded, so your upgrade button only provides you with the images before upgrading, which are separate from the grid. This means that the u (upgrade) button does not require additional GPU time.

2 more powerful tips

In-v5, your cue should be very powerful. You must hint at what you want to see. This means that the type and style of your tips should shine! Your suggestions need to be rich, specific and relevant. If you don't specify the artist's name, media source or artistic style, the system default is photography.

3. Improve the photo quality.

In-v5, you will notice that tips designed to look like photos will look more like photos. In fact, the system default value is more biased towards photography. In order to see more images like art rather than photos, prompt the artist's name, artistic style, media source and type, and add-stv Li Ze 1000 to the prompt.

4. Increased training data

Another important difference between-v5 and other models is that the amount of training data used has increased. This makes the model better understand the relationship between different inputs and outputs, thus producing more accurate results.

5 Promote Natural Language Processing (NLP)

In -v5, it is more advantageous to write your tips in the form of sentences than in the form of lists. Try to write as you learned in school. For example, "astronauts floating in outer space" may produce more predictable results than "astronauts floating in outer space".

6. Improve the quality of mixed images

In -v5, the features of the mixed image are more accurate and powerful.

7 improve consistency

In -v5, you will find that the number of chaotic blocks is reduced. The performance of the face and hands has been greatly improved.

8 Improved fine tuning options

You can fine-tune the way Mid journey reads prompts according to the value of-stylize. In order to control the prompts to the greatest extent,

You can add-stylize N, where n is 0 (the minimum value).

1000 (maximum)

The dish is in the process of Tao.

Fine-tune the style halfway through the rendering process. The lower the value, the less influence the default style of Mid journey has on the prompt. You may find that-stylized 1000 is better in artistic image, while-stylized 0 is better in photographic image.

9 wider screen scale options

In-v5, you can create a wider or narrower canvas size. Try any picture scale you like.

10 image weight regression

When image reference is used in-v 5, its influence (image weight) can be increased by-iwN (where n is 0.5 to 2.0).

What are the new functions of V5 basic model?

-A wider range of styles and more responsive prompts-Higher image quality (2 times higher resolution), higher dynamic range, richer image details, more likely to be correct in details, and less unnecessary text.

-Improved performance of using image hinting-Support for tiling parameters of seamless tiling (experimental function)-Support for aspect ratio greater than 2: 1 (experimental function) -ar- Support for weighted image hinting-iw