A Style-Based Generator Architecture for Generative Adversarial Networks
Anime Crop Datasets: Faces, Figures, & Hands § Danbooru2019 Portraits
Danbooru2018 Is a Large-Scale Anime Image Database With 3.3m+ Images Annotated With 92.7m+ Tags; It Can Be Useful for Machine Learning Purposes such as Image Recognition and Generation.
Large Scale GAN Training for High Fidelity Natural Image Synthesis
https://www.reddit.com/r/NovelAi/comments/xu8xpg/novelai_image_generation_launch_announcement/
Soumith/dcgan.torch: A Torch Implementation of Https://arxiv.org/abs/1511.06434
Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
Danbooru2020 Is a Large-Scale Anime Image Database With 4.2m+ Images Annotated With 130m+ Tags; It Can Be Useful for Machine Learning Purposes such as Image Recognition and Generation.
StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks
StackGAN++: Realistic Image Synthesis with Stacked Generative Adversarial Networks
Synthesizing Programs for Images using Reinforced Adversarial Learning
CAN: Creative Adversarial Networks, Generating "Art" by Learning About Styles and Deviating from Style Norms
IllustrationGAN: A Simple, Clean TensorFlow Implementation of Generative Adversarial Networks With a Focus on Modeling Illustrations.
Towards the Automatic Anime Characters Creation with Generative Adversarial Networks
Illustration2Vec: a semantic vector representation of illustrations
https://www.reddit.com/r/MachineLearning/comments/akbc11/p_tag_estimation_for_animestyle_girl_image/
NoGAN: Decrappification, DeOldification, and Super Resolution
DINO: Emerging Properties in Self-Supervised Vision Transformers
Semantic Image Synthesis with Spatially-Adaptive Normalization
Progressive Growing of GANs for Improved Quality, Stability, and Variation
Progressive Growing of GANs for Improved Quality, Stability, and Variation
ProGAN: Progressive Growing of GANs for Improved Quality, Stability, and Variation [Video]
Improved Precision and Recall Metric for Assessing Generative Models
One Limitation of StyleGAN Is That It Generates a ‘Pyramid’ of Images. The First Layer Makes a 4×4 Image, Which Is Upscaled and Passed through the next Layer (8×8), and so On, Until out Pops the Final 1,024×1,024. by the Time You Reach 32×32, the Overall Structure of the Object Is Established (is This a Face? Is It a Dog?) yet Only the First 4 Layers of the Model Were Allowed to Contribute to That Decision! For a 1,024×1,024 Model, That Means 6 out of 10 Layers of Weights Are Irrelevant.
A Style-Based Generator Architecture for Generative Adversarial Networks [Video]
[StyleGAN] A Style-Based Generator Architecture for GANs, Part 1 (algorithm Review)
[StyleGAN] A Style-Based Generator Architecture for GANs, Part2 (results and Discussion)
GenForce: an Efficient PyTorch Library for Deep Generative Modeling (StyleGANv1v2, PGGAN, Etc)
https://www.lyrn.ai/2018/12/26/a-style-based-generator-architecture-for-generative-adversarial-networks/
AdaIN: Arbitrary Style Transfer in Real-time with Adaptive Instance Normalization
The relativistic discriminator: a key element missing from standard GAN
Variational Discriminator Bottleneck: Improving Imitation Learning, Inverse RL, and GANs by Constraining Information Flow
Spectral Norm Regularization for Improving the Generalizability of Deep Learning
Spectral Normalization for Generative Adversarial Networks
Figure 16: (a) A Typical Architectural Layout for BigGAN-Deep's _G_
CS231n Convolutional Neural Networks for Visual Recognition
A Technical Report on Convolution Arithmetic in the Context of Deep Learning
T04glovern/stylegan-Pokemon: Generating Pokemon Cards Using a Mixture of StyleGAN and RNN to Create Beautiful & Vibrant Cards Ready for Battle!
Here's a Link to My Colab If You'D like to Give It a Go Yourself. This Codebase Builds off of Previous Work from Many People including @advadnoun @RiversHaveWings @NerdyRodent as well as ClipDraw from @kvfrans @crosslabstokyo @err_more and @okw
CLIP: Connecting Text and Images: We’re introducing a neural network called CLIP which efficiently learns visual concepts from natural language supervision. CLIP can be applied to any visual classification benchmark by simply providing the names of the visual categories to be recognized, similar to the ‘zero-shot’ capabilities of GPT-2 and GPT-3
https://towardsdatascience.com/creating-new-scripts-with-stylegan-c16473a50fd0
Conditional Implementation for NVIDIA's StyleGAN Architecture
Someone Used a Neural Network to Draw Doom Guy in High-Res: A Series of Algorithms Turned the Famous Pixelated Face into an HD Portrait
https://www.reddit.com/r/computervision/comments/bfcnbj/p_stylegan_on_oxford_visual_geometry_group/
This President Does Not Exist: Generating Artistic Portraits of Donald Trump Using StyleGAN Transfer Learning: Theory and Implementation in Tensorflow
https://www.reddit.com/r/MachineLearning/comments/bkrn3i/p_stylegan_trained_on_album_covers/
Tired of Books Written by Authors? Try Books Written by AI
https://web.archive.org/web/20230604002332/https://thiseyedoesnotexist.com/story/
Curated Output from a StyleGAN 2 Model Trained on Images That Trigger Pareidolia in the Viewer—Scraped from the `#iseefaces` and `#pareidolia` Hashtags on Instagram.
I Trained a StyleGAN on Images of Butterflies from the Natural History Museum in London.
End-to-End Chinese Landscape Painting Creation Using Generative Adversarial Networks
ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness
Approximating CNNs with Bag-of-local-Features models works surprisingly well on ImageNet
The Origins and Prevalence of Texture Bias in Convolutional Neural Networks
Naoto0804/pytorch-AdaIN: Unofficial Pytorch Implementation of ‘Arbitrary Style Transfer in Real-Time With Adaptive Instance Normalization’ [Huang+, ICCV2017]
https://openaccess.thecvf.com/content_ICCV_2017/papers/Zhang_StackGAN_Text_to_ICCV_2017_paper.pdf#page=7
GAN Dissection: Visualizing and Understanding Generative Adversarial Networks [Blog]
Spatially Controllable Image Synthesis with Internal Representation Collaging
Generative Models: What do they know? Do they know things? Let’s find out!
Unsupervised Discovery of Interpretable Directions in the GAN Latent Space
Object Segmentation Without Labels with Large-Scale Generative Models
DatasetGAN: Efficient Labeled Data Factory with Minimal Human Effort
BigDatasetGAN: Synthesizing ImageNet with Pixel-wise Annotations
Beyond Surface Statistics: Scene Representations in a Latent Diffusion Model
Image2StyleGAN: How to Embed Images Into the StyleGAN Latent Space?
Enhanced SRGAN. Champion PIRM Challenge on Perceptual Super-Resolution
ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks
Joeyballentine/ESRGAN: A Modified Version of the Original ESRGAN Test.py Script With Added Features
CC BY-NC 4.0 Deed Attribution-NonCommercial 4.0 International
Comment Regarding Request for Comments on Intellectual Property Protection for Artificial Intelligence Innovation
https://www.copyright.gov/comp3/chap300/ch300-copyrightable-authorship.pdf#Compendium%20300.indd%3A.122046%3A96431
https://scholarship.law.duke.edu/cgi/viewcontent.cgi?article=1023&context=dltr#pdf
https://www.rutgerslawreview.com/wp-content/uploads/2017/07/Robert-Denicola-Ex-Machina-69-Rutgers-UL-Rev-251-2016.pdf
https://files.osf.io/v1/resources/np2jd/providers/osfstorage/59614dec594d9002288271b6?action=download&version=1&direct#pdf
https://journal.atp.art/the-next-rembrandt-who-holds-the-copyright-in-computer-generated-art/
We’ve Been Warned about AI and Music for over 50 Years, but No One’s Prepared
LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop
Nagadomi/lbpcascade_animeface: A Face Detector for Anime/manga Using OpenCV
Nagadomi/waifu2x: Image Super-Resolution for Anime-Style Art
Provide Demonstration Script for Producing Images Cropped to the Face
Nagadomi/animeface-2009: Face and Landmark Detector for Anime/Manga. This Is 2009s Version of Imager::AnimeFace, but It Works on Recent System.
Animating GAnime With StyleGAN: Part 1—Introducing a Tool for Interacting With Generative Models
BigGAN: Large Scale GAN Training for High Fidelity Natural Image Synthesis § 4.2 Characterizing Instability: The Discriminator
StyleGAN2-ADA: Training Generative Adversarial Networks with Limited Data
Differentiable Augmentation for Data-Efficient GAN Training
Here We Analyze the Performance of BigGAN [2] With Different Amounts of Data on CIFAR-10. As Plotted in Figure 1, Even given 100% Data, the Gap between the Discriminator’s Training and Validation Accuracy Keeps Increasing, Suggesting That the Discriminator Is Simply Memorizing the Training Images...Figure 6 Analyzes That Stronger DiffAugment Policies Generally Maintain a Higher Discriminator’s Validation Accuracy at the Cost of a Lower Training Accuracy, Alleviate the Overfitting Problem, and Eventually Achieve Better Convergence.
Figure 1a Shows Our Baseline Results for Different Subsets of FFHQ. Training Starts the Same Way in Each Case, but Eventually the Progress Stops and FID Starts to Rise. The Less Training Data There Is, the Earlier This Happens. Figure 1b, Figure 1c Shows the Discriminator Output Distributions for Real and Generated Images during Training. The Distributions Overlap Initially but Keep Drifting Apart As the Discriminator Becomes More and More Confident, and the Point Where FID Starts to Deteriorate Is Consistent With the Loss of Sufficient Overlap between Distributions. This Is a Strong Indication of Overfitting, Evidenced Further by the Drop in Accuracy Measured for a Separate Validation Set.
BigGAN: Large Scale GAN Training For High Fidelity Natural Image Synthesis § 5.2 Additional Evaluation On JFT-300M
Retrieval-Augmented Diffusion Models: Semi-Parametric Neural Image Synthesis
Styleganime2/misc/ranker.py at Master · Xunings/styleganime2
Removing Blob Artifact from StyleGAN Generations without Retraining. Inspired by StyleGAN-2
2019-03-08-Stylegan-Animefaces-Network-02051-021980.pkl.xz
Megapixel Size Image Creation using Generative Adversarial Networks
https://colab.research.google.com/gist/kikko/d48c1871206fc325fa6f7372cf58db87/stylegan-experiments.ipynb
doc2vec: Distributed Representations of Sentences and Documents
Conditional Image Generation and Manipulation for User-Specified Content § Pg3
Improved Consistency Regularization for GANs § 2.1 Balanced Consistency Regularization (bCR)
https://cdn.openai.com/papers/Learning_Transferable_Visual_Models_From_Natural_Language_Supervision.pdf#page=4
Contrastive Representation Learning: A Framework and Review
https://colab.research.google.com/drive/1WLU1dIWJ4YeNlMk3Jz9q-1dhLfL23-r-
Tag-Based Anime Generation: This Model Uses Doc2vec Embeddings of Danbooru Tags, Combined With a Conditional StyleGAN2 Model, to Generate Anime Characters Based on Tag Inputs.
2021-01-19-gwern-stylegan2ext-danbooru2019-3x10montage-1.png
2021-01-19-gwern-stylegan2ext-danbooru2019-3x10montage-2.png
2021-01-19-gwern-stylegan2ext-danbooru2019-3x10montage-3.png
Here Are 120K 𝑤 Samples from @AydaoAI’s Large Anime Model (aka TADNE) Clustered into a Set of 256 Centroids. 𝘸𝘢𝘵𝘤𝘩 𝘪𝘵 𝘴𝘩𝘪𝘯𝘦
https://colab.research.google.com/drive/1gbqukfE5f4yYOuHWFW-85zuXW8JtWS09
This Anime Does Not Exist—Interpolation Videos: This Notebook Generates Interpolation Videos from the Model Used for Https://thisanimedoesnotexist.ai by @aydao
https://colab.research.google.com/drive/1QzttnjpQiVHJ8bnhEP0JaSwBX62V1ieG
This Is Great! Now That the Model Can Be Used in PyTorch, I'Ve Starting Playing With @AydaoAI's Anime StyleGAN Directly Guided by CLIP. Starting Slow by Searching for Asuka by Name in the Latent Space.
StyleGAN Anime Sliders: This Notebook Demonstrate How to Learn and Extract Controllable Directions from ThisAnimeDoesNotExist. This Takes a Pretrained StyleGAN and Uses DeepDanbooru to Extract Various Labels from a Number of Samples. It Then Uses Those Labels to Learn Various Attributes Which Are Controllable With Sliders
Controlled GAN-Based Creature Synthesis via a Challenging Game Art Dataset—Addressing the Noise-Latent Trade-Off
Stabilizing Training of Generative Adversarial Networks through Regularization
Update: the XXXL Model (250M Parameters, Doubled Latent Size)
Progressive Growing of GANs for Improved Quality, Stability, and Variation: 3. Increasing Variation Using Minibatch Standard Deviation
TensorFlow Research Cloud (TRC): Accelerate your cutting-edge machine learning research with free Cloud TPUs
Danbooru2019 Is a Large-Scale Anime Image Database With 3.69m+ Images Annotated With 108m+ Tags; It Can Be Useful for Machine Learning Purposes such as Image Recognition and Generation.
Top-K Training of GANs: Improving GAN Performance by Throwing Away Bad Samples
Jukebox: We’re introducing Jukebox, a neural net that generates music, including rudimentary singing, as raw audio in a variety of genres and artist styles. We’re releasing the model weights and code, along with a tool to explore the generated samples.
VQ-GAN: Taming Transformers for High-Resolution Image Synthesis
not-so-BigGAN: Generating High-Fidelity Images on Small Compute with Wavelet-based Super-Resolution
DALL·E 1: Creating Images from Text: We’ve trained a neural network called DALL·E that creates images from text captions for a wide range of concepts expressible in natural language
Some Heavily Cherrypicked Samples from Transfer Learning Using @AydaoAI’s Enhanced StyleGAN-2 Anime Model After 2 Days.
https://drive.google.com/file/d/1qNhyusI0hwBLI-HOavkNP5I0J0-kcN4C/view
https://drive.google.com/file/d/1A-E_E32WAtTHRlOzjhhYhyyBDXLJN9_H/view
Some AI Koans § Http://www.catb.org/esr/jargon/html/koans.html#id3141241
Image2StyleGAN: How to Embed Images Into the StyleGAN Latent Space? § Pg2
https://www.reddit.com/r/SpiceandWolf/comments/apazs0/my_holo_face_collection/
https://www.reddit.com/r/SpiceandWolf/comments/apbz6r/all_those_cropped_holo_faces_uprimarypizza_posted/
2019-02-10-Stylegan-Holofaces-Networksnapshot-00015-011370.pkl
https://www.reddit.com/r/evangelion/comments/apmkjm/brighten_your_monday_with_some_asukas_album_of_130/
https://mega.nz/#!0JVxHQCD!C7ijBpRWNpcL_gubWFR-GTBDJTW1jXI6ThzSxwaw2aE
https://www.reddit.com/r/MachineLearning/comments/apq4xu/p_stylegan_on_anime_faces/egf8pvt/
https://www.reddit.com/r/MachineLearning/comments/apq4xu/p_stylegan_on_anime_faces/egmyf60/
https://www.reddit.com/r/touhou/comments/gl180j/here_have_a_few_marisa_portraits/
Played around With @gwern’s TWDNEv2 Model to Generate Images of Hayasaka Ai! This Is After ~9 Hours of Training (n = 300+). Stopped Working on It After a Bit, so a Bunch of Potential Improvements. More Thoughts Here: https://github.com/ZKTKZ/thdne/bl
Hayasaka.ai/StyleGAN2_Tazik_25GB_RAM.ipynb at Master · Taziksh/hayasaka.ai
https://www.kaggle.com/datasets/andy8744/rezero-rem-anime-faces-for-gan-training
https://www.kaggle.com/code/andy8744/predict-anime-face-using-pre-trained-model/data
https://github.com/ultralytics/yolov5/issues/6998#issue-1170533269
https://www.kaggle.com/code/andy8744/generating-ganyu-from-trained-model/notebook
2019-04-30-Stylegan-Danbooru2018-Portraits-02095-066083.pkl
https://mega.nz/#!OEFjWKAS!QIqbb38fR5PnIZbdr7kx5K-koEMtOQ_XQXRqppAyv-k
GAN Explorations 011: StyleGAN2 + Stochastic Weight Averaging
Averaging Weights Leads to Wider Optima and Better Generalization
Toonify: Resolution Dependent GAN Interpolation for Controllable Image Synthesis Between Domains
‘Network Blending in StyleGAN: Swapping Layers between Two Models in StyleGAN Gives Some Interesting Results. You Need a Base Model and a Second Model Which Has Been Fine-Tuned from the Base.’, Buntworthy
I Just Tried My StyleGAN Layer Swapping Method the Other Way round to What I’d Been Doing Before. So Making the Ukiyo-E Model Human (rather Than the Other Way Around) and I Love the Results!
Combining My Cross-Model Interpolation With @Buntworthy‘s Layer Swapping Idea. Here the Different Resolution Layers Are Being Interpolated at Different Rates between Furry, FFHQ, and @KitsuneKey’s Foxes. P0 Is 4x4 and 8x8, P1 Is 16x16 to 128x128, and P2 Is 256x256 to 512x512.
Cross-Model Interpolations Are One of Those Neat Hidden Features That Arise from Transfer Learning. Here I‘M Interpolating between 5 StyleGAN2 Models: Furry, FFHQ, Anime, Ponies, and @KitsuneKey’s Fox Model. All Were Trained off the Same Base Model, Which Makes Blending Possible.
Inverting The Generator Of A Generative Adversarial Network (II)
Reinventing the Wheel: Discovering the Optimal Rolling Shape With PyTorch
Galton Boards Are Fun and All, but What about Asymmetric Galton Board 🎉😇 By Tuning (thanks #autodiff !!) the Probabilities of Going to the Left/right, One Can Pretty Much Obtain Any Desired Final Distribution 😍 #probability #python #jax
Mining gold from implicit models to improve likelihood-free inference
A Steepest-Ascent Method for Solving Optimum Programming Problems
Synthesizing the preferred inputs for neurons in neural networks via deep generator networks
Unadversarial Examples: Designing Objects for Robust Vision
Style Generator Inversion for Image Enhancement and Animation
Style Generator Inversion for Image Enhancement and Animation
Interpreting the Latent Space of GANs for Semantic Face Editing
SummitKwan/transparent_latent_gan: Use Supervised Learning to Illuminate the Latent Space of GAN for Controlled Generation and Edit
Generating Custom Photo-Realistic Faces Using AI: Controlled Image Synthesis and Editing Using a Novel (Transparent Latent-Space GAN) TL-GAN Model
https://www.reddit.com/r/MachineLearning/comments/aq6jxf/p_stylegan_encoder_from_real_images_to_latent/
https://github.com/Puzer/stylegan-encoder-encoder/blob/master/Play_with_latent_directions.ipynb
https://colab.research.google.com/drive/1LiWxqJJMR5dg4BxwUgighaWp2U_enaFd#offline=true&sandboxMode=true
https://www.reddit.com/r/AnimeResearch/comments/aul582/modification_of_anime_face_stylegan_disentangled/
https://www.reddit.com/r/MediaSynthesis/comments/c6axmr/close_the_world_txen_eht_nepo/
This Fursona Does Not Exist—Fursona Editor (Tensorflow Version)
https://colab.research.google.com/drive/1g-ShMzkRWDMHPyjom_p-5kqkn2f-GwBi
MSG-GAN: Multi-Scale Gradients for Generative Adversarial Networks
2020-01-11-Skylion-Stylegan2-Animeportraits-Networksnapshot-024664.pkl.xz
https://hivemind-repo.s3-us-west-2.amazonaws.com/twdne3/twdne3.pt
https://hivemind-repo.s3-us-west-2.amazonaws.com/twdne3/twdne3.onnx
Implementation of Analyzing and Improving the Image Quality of StyleGAN (StyleGAN 2) in PyTorch
https://colab.research.google.com/drive/1Pv8OIFlonha4KeYyY2oEFaK4mG-alaWF
StyleGAN → BigGAN: Import the StyleGAN Large 8x512 FC _z_ → _w_ Embedding Trick
EndingCredits/Set-CGAN: Adaptation of Conventional GAN to Condition on Additional Input Set
Image Generation From Small Datasets via Batch Statistics Adaptation
https://www.reddit.com/r/MachineLearning/comments/e23ezq/p_using_stylegan_to_make_a_music_visualizer/
Pretrained Anime StyleGAN-2: Convert to Pytorch and Editing Images by Encoder by Allen Ng Pickupp
Video Shows off Hundreds of Beautiful AI-Created Anime Girls in Less Than a Minute
https://towardsdatascience.com/stylegan-v2-notes-on-training-and-latent-space-exploration-e51cf96584b3
Deep Generative Modeling: A Comparative Review of VAEs, GANs, Normalizing Flows, Energy-Based and Autoregressive Models
Generative Modeling by Estimating Gradients of the Data Distribution
Wikipedia Bibliography:
