dvschultz / ml-art-colabs

A list of Machine Learning Art Colabs

ML Visual Art Colabs

A list of cool Colabs on Machine Learning Imagemaking or other artistic purposes

3D Ken Burns Effect

Ken Burns Effect by Manuel Romero • Demo Video by Lia Coleman

3D Photo Inpainting

3D Photography using Context-aware Layered Depth Inpainting • Demo Video

BigBiGAN

BigBiGAN by Tensorflow

BigGan

BigGan by Tensorflow

Colorization

Image Colorizer by DeOldify

Video Colorizer by DeOldify

Coltran by Google Brain

DCGAN

TF-GAN on TPUs by Tensorflow

DeepDream

DeepDream by Alex Mordvintsev • Demo Video

Minimal DeepDream Implementation by Tensorflow

First Order Motion Model

First Order Motion by Aliaksandr Siarohin

FUNIT

FUNIT by shaoanlu

Image/Data Processing

Process WikiArt Dataset by Peter Baylies

Image Generators

Looking Glass 1.1 by bearsharktopus • Tutorial

Image-GPT by Jonathan Fly

Lucid

Lucid visualizes the networks of many convolutional neural nets

Lucid

Lucent - Lucent is a PyTorch variation of Lucid.

Next Frame Prediction

Next Frame Prediction with Pix2PixHD • Training Demo Video • Video Generation Demo

Object Detection

YOLO-v5 by Ultralytics

Object Mask Generation

U Square Net by Derrick Schultz

Shape Matching GAN

Shape Matching GAN by Derrick Schultz

SinGAN

SinGAN by Derrick Schultz • Demo Video

SinGAN Distortions by duskvirkus, inspired by the Yuma Kishi's Studies of Collage of Paintings for Humanity

StyleGAN

Flesh Digressions Loops of the constant and style layers • Demo Video

GanSpace Feature detection using PCA • Demo Video

Barycentric Cross-Network Interpolation with different layer interpolation rates by @arfafax

Network Bending • Demo Video

Network Blending by Justin Pinkney • Demo Video

StyleGAN Paintings (StyleGAN1)

StyleGAN Encoder Tutorial by Peter Baylies

StyleGAN2 by Derrick Schultz

StyleGAN2 by Mikael Christensen

SWA Playground by @arfafax

WikiArt Example Generation Peter Baylies

StyleGAN2 Activations and PCA Projection by duskvirkus, Look at lower network levels of SG2 generator.

Style Transfer

Lucid 2D Style Transfer by Google

Neural Style TF by Derrick Schultz • Demo Video

Superresolution

ESRGAN by Derrick Schultz

Image Superresolution by Erdene-Ochir Tuguldur

SRFBN by Derrick Schultz

SR Zoo ported to Colab by Derrick Schultz

Slow Motion

FILM modified by Derrick Schultz from a notebook by тоже моушн

RIFE by Derrick Schultz (modified from Towards Data Science article)

Super Slomo by Erdene-Ochir Tuguldur

Text-to-Image Generation

Aphantasia by Vadim Epstein • Tutorial

Attn-GAN The OG text-to-image generator • notebook by Derrick Schultz

Big Sleep (BigGAN controlled by CLIP) by Ryan Murdock • Demo Video

Disco Diffusion 4.1 by SOMNAI • Demo/Tutorial

IllusTrip by Vadim Epstein • Tutorial

Quick CLIP Guided Diffusion Fast CLIP/Guided Diffusion image generation • by Katherine Crowson, Daniel Russell, et al.

S2ML Art Generator by Justin Bennington

Zoetrope 5.5 CLIP-VQGAN tool by bearsharktopus

Texture Synthesis

Neural Cellular Automata by Alex Mordvitsev

Texturize: Grass Demo • Demo Video

Texturize: Gravel Demo

TwinGAN

TwinGAN by Manuel Romero

Unpaired Image to Image Translation

CUT by Derrick Schultz

CycleGAN by Tensorflow

MUNIT by Derrick Schultz

StarGAN v2 PyTorch

ML Text Colabs

GPT-2

GPT-2 by Martin Woolf

ML Audio Colabs

Magenta

Generating Piano Music with Transformer by Magenta

Jukebox

Sampling and Co-Composing with Prompts by Anthony Matos

Music Source Separation

Demucs • Demo video by Lia Coleman

Open Unmix

Other Helpful Repositories

dl-colab-notebooks by Erdene-Ochir Tuguldur

shared_colab_notebooks by Manuel Romero

ml-art-colabs's Issues

Pix2PixHD Hanging at 'create web directory'

Hi there,
I'm running Pix2PixHD_Next_Frame_Prediction.ipynb in Google colab, following the link to 'open in colab' on this page: https://github.com/dvschultz/ml-art-colabs/blob/master/Pix2PixHD_Next_Frame_Prediction.ipynb
Everything works well until I start 'Initial Training' where it hangs at 'create web directory' & no epochs appear beneath.
Apparently I'm connected to a Tesla K80 GPU. Tried restarting the runtime a few times and it always hangs at the same place. I left it a good 30 mins or more, and the dataset is only ~350 images at 1280x736 pixels (which seemed to work well when I tried this a few months ago!) It gets as far as making the 'web' folder which contains 'loss_log.txt' and 'opt.txt' - but nothing else.

Next Frame Prediction functions.gcd deprecation

Hello, issues don't seem to be enabled on the pix2pixHD fork so I hope it is ok to write here. functions.gcd(a,b) has been deprecated and I just encountered an error when running.
The lines https://github.com/dvschultz/pix2pixHD/blob/video/train_video.py#L10-L11 can be updated to:

import math
def lcm(a,b): return abs(a * b)/math.gcd(a,b) if a and b else 0

Thank you for this awesome resource!

No matching distribution found for torch==1.4.0

i faced this when I execute !pip install torch==1.4.0 torchvision==0.5.0
ERROR: Could not find a version that satisfies the requirement torch==1.4.0 (from versions: 1.11.0, 1.12.0, 1.12.1, 1.13.0, 1.13.1, 2.0.0, 2.0.1)
ERROR: No matching distribution found for torch==1.4.0

Stylegan_nada custom --> '/content/gems.pkl'

please share the weights, and also if possible then the dataset link or gdrive, too.

Demucs.ipynb "%cd demucs/" instead of "%cd /content/demucs"

https://github.com/dvschultz/ml-art-colabs/blob/master/Demucs.ipynb?short_path=102e2a5

There is a mistake in line 145.
It should be:
"%cd /content/demucs" instead of
"%cd demucs/" otherwise the notebook won't work properly

Even in the moment of that video it is shown that previously it was
"%cd /content/demucs"

https://youtu.be/tHxsqFcx7gw?t=296

Furthemore -
!zip -r AI_01-tasnet_extra-shifts.zip ./separated/tasnet_extra/your_track_name_without_extension/*
instead of
!zip -r AI_01-tasnet_extra-shifts.zip ./separated/demucs/your_track_name_without_extension/*

otherwise it won't work

And you could add the line in the very end:
from google.colab import files files.download('AI_01-tasnet_extra-shifts.zip')
to actually allow to download the result file

No longer seeing Swing videos outputted

depth_edge_model_ckpt: checkpoints/edge-model.pth
depth_feat_model_ckpt: checkpoints/depth-model.pth
rgb_feat_model_ckpt: checkpoints/color-model.pth
MiDaS_model_ckpt: MiDaS/model.pt
use_boostmonodepth: True
fps: 40
num_frames: 600
x_shift_range: [0.00, 0.00, -0.015, -0.015]
y_shift_range: [0.00, 0.00, -0.015, -0.00]
z_shift_range: [-0.05, -0.05, -0.05, -0.05]
traj_types: ['double-straight-line', 'double-straight-line', 'circle', 'circle']
video_postfix: ['dolly-zoom-in', 'zoom-in', 'circle', 'swing']
specific: ''
longer_side_len: 960
src_folder: image
depth_folder: depth
mesh_folder: mesh
video_folder: video
load_ply: False
save_ply: True
inference_video: True
gpu_ids: 0
offscreen_rendering: False
img_format: '.jpg'
depth_format: '.npy'
require_midas: True
depth_threshold: 0.04
ext_edge_threshold: 0.002
sparse_iter: 5
filter_size: [7, 7, 5, 5, 5]
sigma_s: 4.0
sigma_r: 0.5
redundant_number: 12
background_thickness: 70
context_thickness: 140
background_thickness_2: 70
context_thickness_2: 70
discount_factor: 1.00
log_depth: True
largest_size: 512
depth_edge_dilate: 10
depth_edge_dilate_2: 5
extrapolate_border: True
extrapolation_thickness: 60
repeat_inpaint_edge: True
crop_border: [0.03, 0.03, 0.05, 0.03]
anti_flickering: True

Potentially a limit related/caused to the input image resolution?

want to ask the result

so the result is fake_int_img in result folder right?
and i trained my own dataset why the result still someone face.. ?

Pix2PixHD_Next_Frame_Prediction: error changing start frame

Hi, in colab version of "Pix2PixHD_Next_Frame_Prediction", when trying to generate video using the parameter START FROM:

!python generate_video.py --name cuttlefish1 --dataroot ./datasets/cuttlefish1/ --fps 24 --how_many 600 --which_epoch latest --start_from 100

I get this error:
NameError: name 't' is not defined

Any clue? Thanks in advance

ValueError: Cannot feed value of shape (3, 3, 512, 256) for Tensor 'G_synthesis/64x64/Conv0_up/weight/new_value:0', which has shape '(3, 3, 512, 512)'

Hi there, I'm getting the following error when trying to run this notebook:

Traceback (most recent call last):
File "train.py", line 645, in
main()
File "train.py", line 637, in main
run_training(**vars(args))
File "train.py", line 522, in run_training
training_loop.training_loop(**training_options)
File "/content/drive/My Drive/colab-sg2-ada/stylegan2-ada/training/training_loop.py", line 129, in training_loop
G.copy_vars_from(rG)
File "/content/drive/My Drive/colab-sg2-ada/stylegan2-ada/dnnlib/tflib/network.py", line 512, in copy_vars_from
self._components[name].copy_vars_from(src_comp)
File "/content/drive/My Drive/colab-sg2-ada/stylegan2-ada/dnnlib/tflib/network.py", line 509, in copy_vars_from
self.copy_own_vars_from(src_net)
File "/content/drive/My Drive/colab-sg2-ada/stylegan2-ada/dnnlib/tflib/network.py", line 482, in copy_own_vars_from
tfutil.set_vars({self._get_vars()[name]: value for name, value in value_dict.items() if name in self._get_vars()})
File "/content/drive/My Drive/colab-sg2-ada/stylegan2-ada/dnnlib/tflib/tfutil.py", line 227, in set_vars
run(ops, feed_dict)
File "/content/drive/My Drive/colab-sg2-ada/stylegan2-ada/dnnlib/tflib/tfutil.py", line 33, in run
return tf.get_default_session().run(*args, **kwargs)
File "/tensorflow-1.15.2/python3.7/tensorflow_core/python/client/session.py", line 956, in run
run_metadata_ptr)
File "/tensorflow-1.15.2/python3.7/tensorflow_core/python/client/session.py", line 1156, in _run
(np_val.shape, subfeed_t.name, str(subfeed_t.get_shape())))
ValueError: Cannot feed value of shape (3, 3, 512, 256) for Tensor 'G_synthesis/64x64/Conv0_up/weight/new_value:0', which has shape '(3, 3, 512, 512)'

Can you please help me debug this? Since it is coming from the files I didn't write and I am a beginner, I am not sure how to tackle this. Thank you!