What should I do if I am handicapped, but still want to have my own two-dimensional waifu? no problem. Really. Just have a hand: I have to say that this AI is a bit of something, and it really looks like my future wife! Yes, as long as you can give a sketch, AI can turn it into a two-dimensional beautiful girl, and said: more grass will do.
access:
ECS year-end special: cloud server as low as 38 yuan/year
This AI developed by the official technical staff of Japanese illustration website Pixiv, “Amateur”, once topped the daily push trend list, and now has close to 160,000 likes:
After that, the author also added 512 parameter sliders, which can modify the painting style, face projection, shape and other directions. The detailed settings even reached 8192:
Netizens in the comment area directly called:
stop it! Stop making artists obsolete!
The handicapped party can also paint
The comment area below is now occupied by the three major camps.
Some people happily choose new wives from AI paintings:
I thought that “the face I like appeared in an instant, so pleasantly surprised” would flood the comment area, but it didn’t. So I posted a few favorites, so cute…
Ah, by the way, I pushed this kid.
And diverging from this little AI, seriously thinking about whether this technology is snatching the job of an illustrator.
It’s like a personal illustrator in the front row of the comment area is in a complicated mood:
If someone asks me whether this will cause me to lose my job as a painter, my answer is no. However, for people who work hard like me, this technology does make me feel a little unspeakable.
There are also people who think of AI who used to study all the works of Osamu Tezuka, the god of manga, and then “created” Osamu Tezuka’s new work “ぱいどん” in the style of Osamu Tezuka.
Regarding this trend of technological interference in creation, these netizens said that they do not know whether this is fortunate or unfortunate for real artistic creation…
But there are also many moderates who say: You think too much.
It’s like the world‘s number one princess, the virtual singer Hatsune Miku. Her appearance hasn’t reduced the number of real singers, right?
Instead, many young people fell in love with music because of her.
Therefore, no matter how advanced the technology is, there will be no need for a painter.
Some commentators also firmly believe that “technology is just a tool.”
The work of being an original artist will not be affected, on the contrary, the production efficiency of animation will be improved as a result.
Of course, in addition to these slightly fierce controversies, since the author has only released demo videos, a large number of netizens have also appeared in the comment area to scan the screen for open source and for Demo.
Development based on StyleGAN
Although the author did not reply directly, we have introduced many techniques for generating pictures and faces from sketches in the past.
And this AI should also be based on the further development of StyleGAN.
This algorithm, which turned out in 2019, can be said to be a big step forward in human face generation technology.
Although the previous GAN is good in the resolution and quality of the generated image, researchers still cannot understand the image generation process more deeply, such as how to control the random features of the image.
StyleGAN redesigned the GAN generator network structure:
Starting from the learned constant input, the generator adjusts the image “style” of each convolutional layer based on the latent code, thereby directly controlling the image characteristics, and then controlling the entire image generation process.
At the same time, since this process can be combined with the noise directly injected into the network to directly change the random attributes in the generated image, you can also see that the whole work based on StyleGAN can be said to be endless.
The author comes from the official technical team of Pixiv
In fact, we mentioned at the beginning that this AI is only an experimental amateur development by the author, and the developer himself comes from the official VRoid team of Pixiv.
The main product of this team is the 3D character creation software launched by Pixiv in 2018, VRoid Studio, which focuses on “creating 3D models in 2D”.
Like this, the painter can directly perform 2D painting on the 3D model preset in the software, and adjust the parameters according to the needs, and the adjusted results will be reflected on the 3D model in real time:
Maybe it’s career-related. The developer also said that his amateur AI “sketches turned into comics” will also add the function of 3D model generation in the future, realizing one step from sketching to three-dimensional.
In this regard, I can only say that I am waiting for the handicapped party: GKD.
.