Lorem ipsum dolor sit amet, elit eget consectetuer adipiscing aenean dolor

Follow SNS

ARATAMA 璞|Generative AI Ecosystem

Amazing 1763 sneakers made with AI "STABLE DIFFUSION"

First, introduce 5 recommended AI sneaker designs

Asymmetric color digital sneakers

What was distinctive was the sneakers with asymmetric color variations and left and right differences.
If you look closely, the color of the string is different, and the decoration, position, size, and number of fasteners are different.
It is a dynamic design that is difficult for people to design.

Exactly Evangelion style digital sneakers

Both the color scheme and the shape are cool.
I didn’t cast an Evangelion-like spell in particular, but it’s a miraculously generated sneaker.
I would like to buy it normally if it is on sale.
Since it’s CC0, what do you think? Hideaki Anno and Studio Color.

Glowing digital sneakers

It’s bright and the design is good.
A nighttime running shoe.
It was popular in the past.

Unique digital sneakers

I was terrified in the realm of art sneakers.
Somehow it looks like a living thing, and it’s adorable, so it’s likely to sell well in Harajuku.

Kirafuwa Digital Sneakers

Near future + sparkly + fluffy, these are my favorite sneakers this time.
The tiptoe angle is also automatically generated, so there must be a shot that I used as a reference…

Other mints in OpenSea can be viewed

👟AI Sneakers 1763👟

Make it

It was my first try with AI images, but I was stunned by the high quality.
If they were sold in stores as fashion sneakers, there would be many things I would want to buy at all, and the designs are also unique.
By collecting them in OpenSea, we have created a digital sneaker catalog that is fun to look at.

What surprised me was that casting the same spell didn’t produce the same thing.
The model and camera angle are also learned from the real thing, so there is a professional feeling.
The feeling of generating it for the first time was like displaying HTML in the browser for the first time, and I was hooked.

This production environment

This time, I made it in the following environment.

  • Image Generation AI: Stable Diffusion
  • GPU: Google Colab Pro
  • Same spell: text2img (generate image from text)
  • Number of spawns: about 2500
  • Size: 512px × 512px
  • Remove the ones that don’t work / fix some in Photoshop
  • Time it took: 1-2 weeks

The point of production is to get closer to the image of your imagination while grasping the pattern of the spell before generating a large amount, and to delete the bad things, and if the cool things become strange,It is necessary to switch to do the correction in Photoshop.
I think that the quality will be further improved by fine-tuning the lighting in Lightroom Classic.
In this respect, I think that people who are very good at graphic designers and retouchers can raise the degree of perfection.
As for how to cast spells, Takayuki Fukatsu’s article will be helpful, so please check it out.

“Drawing AI course to understand as magic”

Personally, I found it difficult to control textures, angles, obvious inconsistencies, letters and marks.
This time it was text2img (generating an image from text), but next time I would like to try img2img (generating an image from an image), so I plan to increase variations by combining them.

Future prospects

ARATAMA DAO, which has just started, has launched AI digital content at its core. I would like to devote myself to research and development with everyone who is interested in seeing outrageous things.
So if you are interested, please join the Discord. I want to do it noisily, but I’m lonely because there are no people at all lol

It hasn’t been long since “Midjourney” and “Stable Diffusion” were released, so I think there will be various developments in the future, including updates and other products.
Emad Mostaque, CEO of Stability AI, who is developing ‘Stable Diffusion’, said, ‘The one announced in August 2022 is the first in the series, and in September it will be voice, and then video, 3D, language, code. We will also support it.” It seems that the policy is to reduce the capacity of the dataset.

In the future, the democratization of AI will occur, and just as anyone can use the Internet and web3 (especially cryptocurrencies), AI can be used by everyone, combining what other people think with what they think. Extraction, making variations, and repeating the creation multiple times will greatly expand what you can do.
It is a revolution in which something new is born one after another from data assets.
Conversely, people who can draw originals may be featured in a different way, but I think it will lead to a creative 2.0 world line that is different from the creations that have been made up until now.
Of course, it is necessary to firmly face the law such as copyright. Recently, misleading AI-generated images have been posted on Twitter, raising moral questions. In addition, since it has a large impact on the economy, it is likely that there will be differences in the response of each country.

I understand that some people are pessimistic because I’m just a designer, but I think the future will be bright if I use my skills to enjoy expression and challenge the possibilities.
Rather than silently doing it alone, I would like to research the generation of AI digital content while having fun with everyone, spending a strange time from the second creation to the 100th creation.
Wouldn’t it be wonderful if we could build a world of token economy, including NFT, and receive rewards for the projects we’re involved in, just like web3?
It would be interesting to turn the DAO itself into an AI.



🎉🎁👟AI Sneakers 1763 free mint festival👟🎁🎉

October 2, 2022 (Sunday) from 21:00(JP) until sold out

⚡⚡⚡2nd Announcement AI Cyber ​​Gang Girls⚡⚡⚡

Scheduled to distribute white list in October 2022
Scheduled to be announced on Twitter:@aratama_en

💬Contact us💬

If you are a company, NFT founder, or creator who wants to try various possibilities with AI digital content, please let me contribute.
Please feel free to contact us, even for trivial consultations.