(Series of pictures created by Twitter author @ChromedSets using Waifu Diffusion)
The original author's drawing tutorial link has been attached at the end of the article. Friends who understand English can take a look at it themselves.
At that time, I also tried to follow the instructions of @chromedSets and tried the waifu diffusion, but found that the links in the tutorial he gave needed to use the secret key from a certain forum, otherwise it would not be used.
continued to look for other resources and found that another link could draw pictures. However, because of nsfw, the system automatically blocks it and can only display pure black pictures. About 80-90% of the chances are black pictures, and the remaining 10-20% of the pictures are not very satisfied, so I just forgot it and didn't continue to use it.
(Note: NSFW refers to Not safe for work, something that is not suitable for browsing at work. We can understand that it is not suitable for children.)
Until two days ago, He Caitou released the "Concise AI Painting Tutorial for Readers' Children", which mentioned the waifu diffusion, and then found a useful entrance. Finally, I started playing happily.
2.
Step 1, you need to open this link: www.stable-diffusion.click
Step 2, enter what you want to generate in the upper input box, such as:
Japanese girl, white hair, with big sword (Japanese girl, white hair, holding a big sword)
Step 3, click the "Generate" button below the input box to start generating the picture. Then a prompt box will pop up saying that the queue is being generated, and there are still a few people who are still your turn, please wait. Click yes.
Step 4, slide the mouse down, you can see other people's predictions and pictures. You can see your own props and pictures after a while.
Step 5, click on the image you want, it will automatically enlarge. Click the share/download image button. If the download does not respond, just take a screenshot and save it.
. Next, you can test the various functions of this website to generate various interesting pictures.
For example,
(refer to @chromedSets's props)
(refer to @chromedSets' props)
(stable-diffusion.click Picture generated by other users)
(stable-diffusion.click Picture generated by other users)
(stable-diffusion.click Picture generated by other users)
On the website, can see the good pictures drawn by others, and the corresponding prop description and system settings can be used to directly refer to and debug other people's props to generate the pictures you want.
Now, you won’t know how to write prompts.
Hehe, it's fun.
OK. Before I knew it, the AI painting series has been written for three articles.
History two articles can also be viewed:
The most popular AI drawing stable diffusion tutorial
AI painting introductory tutorial: Disco Diffusion
3,
Some friends will ask, why are they all AI drawings from foreign websites, aren't there any Chinese ones? Can't use Chinese? Of course,
is OK. Interested friends can also try Baidu Wenxin Yige (yige.baidu.com). The experience is good, it is quite fun, and of course there is a lot of room for development.
(non-stakeholder, but not many platforms that support Chinese are available now)
Because they are all Chinese, I won’t do tutorials, just use the mouse to do it.
However, the semantic understanding of Chinese is relatively difficult to do than in English. I also look forward to the development of domestic AI drawings as soon as possible~
Reference materials:
1. Hecaitou: A concise AI painting tutorial for readers' children
2. chromedSets: I USE STABLE DIFFUSION USING DANBOORU/WAIFU MODEL , rentry.co/8vaaa (need to magic to surf the Internet)
3. 36Kr: Will AI Art be the next NFT?丨Digital Forecast·Global Industry Mapping