# stylegan2 machinelearning contemporaryart ai hibiscus morshu aiartist digitalart nebula galaxy curious curiosity aiart discodiffusion weirdwonderfulaiart artificialintelligence aiartists aiartcommunity gan stylegan3 generativeart vqganclip artist nft nftart nftcollectors salvaukraini war robot artificial generate generative dream deepdream vqgan painting computervision a1 abstractart conceptart scifi cyberpunk dystopia nowreading nftartwork nfts nftcommunity aiartwork workinprogress
画質 高画質

さらにAIさんを鍛え中3日目ぐらい。もはや普通に売れるレベルの絵になってきた。やはりGPUとかTPUとか筋肉で殴る感じやな。

6 78
2022-05-23

The art of creating a dataset (for StyleGAN interpolation animation)
It's all about finding the balance between diversity and similarity. Too diverse, and the training process 'collapses'. Too similar, and the results are uninteresting. Currently making these with VQGAN/CLIP.

5 33
2022-05-21

A happy (?) mistake whilst trying to recreate the snapshot grid during a StyleGAN training run...

0 6
2022-05-17

Taking hundreds of these images, I created a dataset to train a STYLEGAN2 model on in hopes to create "new" versions of my existing work.

Instead, I found that the stills and videos created were more beautiful than anything I'd done before.
4/

0 5
2022-05-05

I used an AI that made Morshu realistic. I added the hat later.

1 3
2022-05-03

GN Twitter! 🪄✨💎
stylegan-xl

7 41
2022-04-12

DualStyleGANを使って無理やりアイコンから実写っぽい画像を生成して遊んでる
https://t.co/amXvmaCFZg

2 7
2022-04-03

Fixed various issues, animegan quality improved a lot from earlier, only at 64x64 for now. Has the "blob" issue from StyleGAN now so I'll have to try adding weight demodulation next.

0 0
2022-04-01

Fair to say my StyleGAN experiments are yielding some nice results.

1 11
2022-03-28

i changed the Generator to stylegan3 and OMG it becomes REALLY GOOD WOW

1 13
2022-03-23

what i did today instead of generating an image to match a text description why not just generate image to match another image and it worked! very cool i also tried different gan generators like projected gan, stylegan2, and stylegan2 ada (top is generated bottom is the real one)

0 9
2022-03-23

通常、GANでアニメを生成すると言えば顔だけですが、これはStyleGAN２でアニメの上半身を生成するモデルです。

作成したのは、AIアーティストAydao（）さんで、膨大なアニメ画像を使った転移学習・拡張によってアニメの上半身の生成を可能にしています。

ブログ：https://t.co/FWkEPeK4CN

87 247
2022-03-15

Ai art thread. No links. Post some ai art and then come to our spaces tomorrow. (I will be doing a stylegan training on discord afterwards)

https://t.co/KZkUBHNhgv

59 266
2022-03-14

Self-Distilled StyleGAN [Mokady+, 2022]
Web上の画像でGANを訓練する際の課題は外れ値と多峰性である。自己蒸留によるデータの選別とWのクラスタ中心へのtrucationによって従来よりも多様なデータセットで高品質な画像生成を可能にした。データセットも公開。
https://t.co/Kika7V6uEc

5 30
2022-02-28

Double (Mario Klingemann) score on - 'Plastic Trash in a Storm at Sea', an AI-generated painting based on a text prompt and proceeds going to . The other is 'Arti-Choke', a portrait generated with GLIDE/StyleGAN2

1 11
2022-02-25

Another new collection in progress, Cyberpunk Apocalypse, a series mixing landscapes with StyleGAN2+CLIP avatars. Here they are separately, working on image segmentation to combine them!

1 8
2022-02-02