Midjourney Learning Series 2 - Midjourney Advanced Road

​In the last article, I wrote a midjourney nanny-level tutorial for everyone. If you don’t know how to use Midjourney yet, click [ here ] to learn. Let us start the advanced journey of Midjourney today.

Midjourney Learning Series 2 - Midjourney Advanced Road

After the first nanny tutorial, you can already use prompt to make various AI works. But you must be curious about how to prompt the text to get better results and achieve the picture you want. You also think about how to let AI modify and create on an original picture, with unlimited creativity.

So in this article, I will refine the key technical concepts of AI drawing for you, and point out the advanced direction, so that you can avoid detours.

Here I discuss three topics, prompt, settings, and padding (that is, generating pictures from pictures). One by one, grasp the core concepts first.

about

You must first remember that the prompt is composed of three parts: screen content description + screen decoration + parameters. For example, in the generation of cats and cats in the picture above, a cat sitting on the street at night is the content description, while Bubble Mart style, Q version, 4K, etc. are the screen modification, and the final parameters such as the aspect ratio of the picture -- ar 2:3 It is the parameter.

There is no need to say more about the picture description, just explain clearly what you want to draw; picture retouching is to explain clearly how you want to draw, such as what lighting, what style, what clarity, what material, etc.; It is also very simple. For example, in the previous article, we produced a cat picture as follows

It can be seen that there is a tail, so if I want a cat without a tail all the time, then I can add a --no tail parameter, then the cat will put the tail away.

About settings

The second topic, settings. You can set the painting style (such as 2D or 3D), the version of the model, the quality of the picture, the way to modify the picture and so on. Enter /settings in the command window to see the following setting options, just click the desired option, for example, the picture below is the V5 model, high-quality picture, fast mode.

About Mat Chart

The third topic is the mat image, which is very important, because a large number of people just want to create on a certain original image, such as logo design, avatar generation, etc. The operation of the pad map is also very simple. Let me take a bat LOGO design as an example to illustrate how to use the pad map.

Bat LOGO design case tutorial

Click the + sign on the left side of the command window, click upload file, after uploading, remember to press Enter, after uploading the picture, click the picture, the picture will pop up, as shown below. Right click and select copy image address.

Then go back to the command line, enter /imagine and paste the image address just copied: https://media.discordapp.net/attachments/1099533962206007440/1099593808548724827/private-id-n-he3ehywgo-youtube-a.jpg , then enter The picture you want, for example, we want a bat, metallic luster, LOGO design style, 4K (English: a cool bat, logo design, metallic shining, 4K), you can generate the following picture

Here I will teach you a magic parameter, --iw. This parameter can control the similarity between the final generated image and the original image. The larger the value, the more similar it is. The value range is 0.5~2. Let's try to use --iw 2 if the bat produces the image. Effect:

It can be seen that the similarity with the original picture is high, especially the second picture, the background is also white.

So far we have condensedly described the three key topics of Midjourney's advanced learning. Due to space limitations, I have shown the core concepts with the most intuitive examples, so that everyone can know what they are doing and have a way at their feet. I also recommend two more specific study references.

Learning parameters: https://zhuanlan.zhihu.com/p/609658809

Learning mat map: https://zhuanlan.zhihu.com/p/620588789

Now, let’s talk about the fan benefits I prepared for you, which can be regarded as a conscience recommendation. Midjourney's official membership fee is really expensive (mainly because we are too poor~), it costs about 200 yuan a month, so we have to find a way to save money. So after some searching, I found shared accounts. After my own experience, shared accounts are really the most affordable and best way to use. Specifically, 6 people share one account, but the user experience is unexpectedly good, because they can log in at the same time, and they can build their own rooms without affecting each other, and the experience is very smooth. After in-depth experience and research, I think this is the best way to use Midjourney.

I feel that the sharing membership is really useful, so after a wave of communication with the owner of the seller, we hit it off, and after a short operation, I became a partner of a small branch ! (My family is amazing, you read that right, it is directly a partner of a newly opened small branch), hurry up to the comment area to get a large coupon, or follow the official account "Xiaoyuan Life Knowledge Museum" and reply "recommended" , make the first AI work in your life!

If you think the article has given you a better understanding of AI painting, move your beautiful hands to give the article a thumbs up~

Guess you like

Origin blog.csdn.net/Meiyuan2021/article/details/130336479