Shulou(Shulou.com)11/24 Report--

Foreigners who are tortured by Chinese begin to learn Chinese characters with Stable Diffusion! What's more, SD is not only proficient in Chinese characters, but also can create characters on the spot. Pikachu, iPhone, and deep learning are all pinched!

Learning a foreign language is not an easy task.

Of course, the Chinese are not the only ones who eat this bitter.

On Reddit, there is a topic of "r / ChineseLanguage", which specially posts complaints about the problems encountered by foreigners in learning Chinese.

For most foreigners learning Chinese, Chinese characters are always an insurmountable hurdle.

Without correct learning methods, it is difficult for learners to remember hundreds of Chinese characters with different structures.

But it doesn't matter. Stable Diffusion will do it.

Can't you read the words? Look at the picture! On Reddit's Stable Diffusion channel, netizens shared the text generation model Depth2Image.

Code link: https://huggingface.co/ stabilityai / stable-diffusion-2-depth / blob / main / 512-depth-ema.ckpt this model uses Stable Diffusion technology to transform the Chinese character image into the meaning it represents.

For example, the word "meat" made of ham

And the word "rope" made of rope.

The blogger shares the way the image is generated, enter the material photo that needs to be presented in the model, and Depth2Image will generate Chinese characters corresponding to that material.

As soon as this post was posted, it aroused extensive discussion. While enthusiastically sharing their own words, netizens also excitedly said: the application of this in the field of education is promising in the future!

In fact, two years ago, some people were already engaged in the study of text visualization. Twitter user @ azlenelza uses neural networks to graphically express their meanings by generating visual mnemonics.

Azlen said that this approach will make more connections between the meaning and shape of each Chinese character, making each Chinese character easier to remember.

It can be found that the images generated by AI at that time belong to the category of art characters. For example, in the image in the lower right corner, it took the editor a long time to recognize the word "system".

After two years of development, the technology of text and image has become more mature. With this technology, it may really help Chinese learners to better master every Chinese character and shine in the field of Chinese teaching.

Not enough words? You have to "build"! What's more, Stable Diffusion can also make words.

After learning 10,000 Japanese characters, the Stable Diffusion model began to create "fake Chinese characters" for new things, such as Mario, Pikachu, Musk, Deep Learning, YouTube, Gunda, Singularity and so on.

In these words, we also found the Chinese point.

Compared with the original Mario, the Mario created by Stable Diffusion is not irrelevant, it can only be said to be exactly the same (boy, pure pictograph, right).

In addition, the word "Musk" created by the AI model is also domineering. It is not enough to be king, directly super double, become "king of kings".

By contrast, Curly Fu and Nicholas Cage cried and fainted in the toilet.

If you expand the size of the output image, the model will also complete the work, resulting in a "text" that is more complex than the QR code.

Secret people: 6.

Someone in the comment area said, "isn't this Xixia language?" "

Don't say, it's kind of like it. This AI, knowledge is miscellaneous.

In the eyes of Chinese and Japanese netizens, this kind of technology may be just to clean up. After all, if you are used to reading your own characters, you will feel strange even if you look at the Chinese characters in each other's language.

But for people in another language system, seeing images with different structures with such rich connotations may be more likely to feel the charm of language and words.

After the Tower of Babel, human beings are unable to communicate, and modern science and technology are building a new "Tower of Babel" while overcoming language barriers. Will mankind succeed this time?

Reference:

Https://twitter.com/hardmaru/status/1614119727416369152?s=46&t=BdjSSthlkb8_b_AWoh4BGw

Https://www.reddit.com/r/StableDiffusion/comments/10b3r24/using_depth2image_to_create_images_to_aid/

This article comes from the official account of Wechat: Xin Zhiyuan (ID:AI_era)

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.