Creator of "This Anime Does Not Exist" model (thisanimedoesnotexist.ai). Generator generator, *GAN surgeon. AI / ML / Art 🤖 🎨 Making anime real.

Joined August 2019
Pinned Tweet
The AI makes them blink and gives them teeth... 😬 (take 2)
54
2,721
217
10,151
334,077
aydao retweeted
messing around with ru-dalle (source code: github.com/sberbank-ai/ru-da…) prompt "a happy blonde haired fox anime girl" (in auto-translated Russian) most images are quite messy, these are cherry-picked:
3
1
19
Show this thread
aydao retweeted
We've added a search input to the explore page to make it easier to find models. ⚡️🔎⚡️ replicate.com/explore?query=…
2
4
aydao retweeted
Our sibling app ProsePainter is almost ready for public use! Folks in our discord are already using it to take their Artbreeder creations to the next level. A few early examples of Artbreeder images before and after ProsePainter! This is just scratching the surface.
5
11
3
91
Show this thread
aydao retweeted
スケッチからイラストと3Dモデルをリアルタイムで自動生成する実験です。 右上がワイヤフレーム、左下が法線、右下がテスクチャを貼り付けた3Dモデルになります(イラストと区別が付きにくいため軽く回転させています)。 パラメータスライダーでの画風や形状、落ち影、顔向きの変更も反映しています。
50
10,021
429
36,860
750,557
Show this thread
aydao retweeted
StyleGAN-V: A Continuous Video Generator with the Price, Image Quality and Perks of StyleGAN2 abs: arxiv.org/abs/2112.14683 project page: universome.github.io/stylega… can generate arbitrarily long videos at arbitrary high frame rate
4
56
3
327
13,462
Show this thread
ruDALL-E finetuned on Warhols
9
14
112
Show this thread
aydao retweeted
anime face detection using mmdetection and mmpose, trained via TADNE model, includes colab+pre-trained models github.com/hysts/anime-face-…
5
21
1
100
Show this thread
Final epoch (31) of the danbooru 500m diffusion training, model out once inference script is done and training tips written up. Soon (TM)
Epoch 24 of the danbooru 2020 500m diffusion training is learning to draw the arms.
2
11
1
36
Ahem. ATTENTION ML TWITTER: *grabs megaphone* The Russians trained their own DALL-E (ruDALL-E 1.3B), and it looks amazing. Googling "ruDALL-E" brings up nothing whatsoever. Apparently @l4rz is the only one who noticed it was released a few hours ago!
"тян из аниме, похожая на Хомуру" ruDALL-E 1.3B
Show this thread
5
19
4
120
Show this thread
Vehicles are coming soon to Artbreeder! 🚀🚗🚲 Create weird mixes of: cars, motorcycles, mech-warriors, spaceships, airplanes and tanks.
10
87
7
557
15,970
Projected GANs Converge Faster Projected GANs match the previously lowest FIDs up to 40x faster, cutting the wall-clock time from 5 days to less than 3 hours given the same computational resources! arxiv.org/abs/2111.01007
5
62
1
335
Show this thread
Spooky StyleGAN surgery art from way back when
1
1
31
Show this thread
aydao retweeted
アニメキャラの上半身を生成するStyleGANで遊んでみる ブログ:cedro3.com/ai/anime/
1
19
1
67
1,250
Using @AydaoAI’s Flesh Digressions script to mangle the output of my botanical #StyleGAN model. Things are always more interesting when they go a bit wrong. #latentbotany #aiart #artist #ArtificialIntelligence
1
6
23
455
I’ve gotten a lot of questions about setting up training for stylegan3, and I will always recommend you read the repo before doing anything. There’s a link to this doc with this guide in there that covers a lot of the training arguments: github.com/dvschultz/stylega…
1
1
1
28
Show this thread
aydao retweeted
Vectorization of Raster Manga by Deep Reinforcement Learning abs: arxiv.org/abs/2110.04830
3
121
10
445