AIGC or the two-thorned ape can play: The Niewa community is booming, and you can become a fandom boss if you have a brain.

The west wind in the house comes from the Aofei Temple
Qubit | Official account QbitAI

It turns out that this is how the two-spined apes nowadays play——

With the help of AIGC, even with zero drawing skills, you can easily create your own fantasy. If you don’t have a favorite character, you have nothing to worry about!

Gosho Aoyama has not allowed the New Zealand Party in "Detective Conan" to get his wish for 28 years, so he quickly arranged for the two of them to get married:

"Demon Slayer: Kimetsu no Yaiba", "Spy's House", "Spell Return"... Make enough food and clothing by yourself:

Good guy, this "game" called " Pinch Ta " has made the two-spined apes have fun: someone used it to pinch hundreds of their own babies last year (shocked face) .

After asking around, I found that it is not only a small program, but also an AI-driven UGC co-creation community. On average, there are tens of thousands of UGC contributions per event.

It is highly recognized by serious two-dimensional people, and the founder behind it is also a doge .

Travel into a fantasy world

Speaking of pinching it, its main function is very focused, which is to pinch pictures.

Because of the use of AIGC capabilities and ready-made prompt words for people to choose, even a novice in painting + Prompt can easily create the two-dimensional child in his mind in seconds (and it is not OOC ) .

Moreover, it can better maintain the consistency of the character and cover all aspects of the detailed characteristics of the character, so the pinched pictures can be used to make self-made comic strips:

You can even create a whole family story:

As long as you have a big enough brain, you can figure out anything:

As for the specific operation, there are generally three pinching methods: text pinching , picture pinching , and CP picture pinching.

There are hundreds of characters, hundreds of styles, and various settings in the "Spell Dictionary" to choose from. The different "spells" selected will be highlighted like a programmer's coder.

Judging from our experience, the control is as strong as the cute Midouzi’s biceps:

For picture pinching, you can choose to upload a picture from the mobile phone album or select a previously generated picture for "secondary creation", and the pinching strength can be adjusted. Just like text pinching, you can design your own prompts, or you can choose them directly.

CP diagrams need to describe two roles separately, and there are also CP templates to choose from.

It is not a problem to let the CP in your heart join hands to succeed:

Of course, good ideas are not satisfied with fan creations. You can also customize your own original character - Niebao . Then you can pinch your own baby in the pinch picture, which is commonly known as "raising a baby":

If you have no inspiration, it is recommended to focus on experiencing the magical gameplay of " Qi Brigade ". Players can choose existing characters or their own created characters as "agents" to experience the plot and story and develop their personality.

(Coconut is our own special agent named and created by ourselves)

To put it simply, the gameplay of Qilu is a systematic description of the plot. The player pinches the corresponding picture according to the plot, and the character will also give you an evaluation:

Match the pictures with the plot text, and a comic strip belonging to you and your little one will be born.

Knock on the blackboard! From the characters to the lines, to the plot, to the final pictures, the entire journey is all the work of AIGC.

According to the news from Bagua, Pinta is based on the Stable Diffusion open source framework. At the same time, the basic model is retrained, and continuous fine-tuning and alignment iterations are continued. The team also revealed that an open source two-dimensional model will be launched in the future.

In addition, the model uses the MoE of multiple expert fine-tuning networks to fix the role, instead of just using feature extraction from one reference image. This kind of character modeling covers all aspects of the character's detailed characteristics, rather than a single angle or just the characteristics of the face.

Therefore, on the premise that the details are more consistent, the character's expressive power is stronger , and it can look completely different from a reference picture.

This kind of fixation does not require any model training time and has strong batch capabilities, so it is not a problem for users to make 500 characters a year.

Be “MiHoYo + Little Red Book in the AI ​​Era”

Once you experience it, the characteristics of pinching it will be obvious.

First of all, it is an AI creation community that was born based on the hot topics of the times .

Utilizing the multi-modal generation capabilities of the underlying AI large model, it provides users with the basic elements of a new generation of UGC creation: characters and stories.

Secondly, it can bring users a personalized experience .

Because AIGC participates in the entire process, plus the reserved character library and event library, any pinched child can be used to advance the plot, just like setting up a camera in a different world, just waiting to press the shutter.

Okay, next is a reserved project - what kind of team is behind this AI community?

Qubit learned that the company behind PinchTa was established in December 2022 and is called Jianjian Concept . The team currently has 15 members.

The founder and CEO is Hu Xiuhan , who himself is a 2D enthusiast - so he can capture the enthusiasm of 2D users for spontaneous memes and the desire for a good creative platform. The name "Pinch Ta" is not only vivid and makes people want to pinch it , but it is also the pronunciation of "Geng" ( Neta) in Japanese .

Hu Xiuhan met the partner who saw the concept while working at Meta. At that time, Hu Xiuhan served as the technical team leader and was responsible for a full set of video-related technical solutions for the three platforms of Facebook, WhatsApp, and Instagram, including content publishing tools, content management, copyright data analysis, advertising system docking, etc. In the earliest era of Pix2Pix and StyleGAN, AI functions were successfully implemented into the creator ecosystem covering more than one billion people around the world.

After returning to China, Hu worked for a period of time at Tezan, a content technology unicorn.

At the end of 2022, ChatGPT fired its first shot. Hu Xiuhan felt that this wave of AI was an opportunity to have a core impact on the UGC ecosystem.

What we are targeting may be the biggest opportunity, which is similar to the "MiHoYo + Little Red Book" opportunity.
Everyone has their own new Avatar, and they can exchange "tags" with others in the community to create it together. This was something that could not be done before the virtual human technology was not C-end. Even if the character could be designed, the character could not have its own The story grows.

In addition, I saw that the concept team still had technical capital.

Team members have been deeply involved in multiple open source community projects, participating in the development of the ComfyUI plug-in ecosystem, the most used SD installation image, and also contributed to AnimateDiff, AIDXL, one of the best open source models of the second dimension, Chat Haruhi Suzumiya, etc.

These projects are not limited to Vincentian drawings, but also include animation and language generation.

The accumulation of past technologies can ensure to the greatest extent that the AI ​​multi-modal technology behind Pinch Ta will not fall behind.

What we hope to do is to support the accumulation of technical capabilities in multiple directions such as text, voice, comics, video, etc. at the lowest cost, and at the same time obtain a large amount of real demand feedback.

At the same time, the open source spirit that team members have always possessed has become the DNA of the current NieTa community.

Hu Xiuhan said: "ACG has a very good content collaboration gene. The subtitle group, animation group, and Guiwu Second Creation are all former content collaboration organizations, which are crucial to embracing an open co-creation environment."

After all, “powering for love” means being able to keep the costs of talent and community organizations as low as possible.

"If everyone in the future world has several AI partners throughout their lives, then we hope that they will be wonderful, warm, and memorable." Hu Xiuhan said that what the team hopes is to create tools to shape AI partners and let everyone's story There is more flexibility in life.

“The opportunity to experience the entire universe in a short period of time.