926 KiB
926 KiB
In this notebook, we will show how to load pre-trained models and draw things with sketch-rnn
In [1]:
# import the required libraries
import numpy as np
import time
import random
import pickle
import codecs
import collections
import os
import math
import json
import tensorflow as tf
from six.moves import xrange
In [2]:
# libraries required for visualisation:
from IPython.display import SVG, display
import PIL
from PIL import Image
import matplotlib.pyplot as plt
# set numpy output to something sensible
np.set_printoptions(precision=8, edgeitems=6, linewidth=200, suppress=True)
In [3]:
# !pip install -qU svgwrite
In [4]:
import svgwrite # conda install -c omnia svgwrite=1.1.6
In [5]:
tf.logging.info("TensorFlow Version: %s", tf.__version__)
In [6]:
# !pip install -q magenta
In [7]:
# import our command line tools
from magenta.models.sketch_rnn.sketch_rnn_train import *
from magenta.models.sketch_rnn.model import *
from magenta.models.sketch_rnn.utils import *
from magenta.models.sketch_rnn.rnn import *
In [8]:
# little function that displays vector images and saves them to .svg
def draw_strokes(data, factor=0.2, svg_filename = '/tmp/sketch_rnn/svg/sample.svg'):
tf.gfile.MakeDirs(os.path.dirname(svg_filename))
min_x, max_x, min_y, max_y = get_bounds(data, factor)
dims = (50 + max_x - min_x, 50 + max_y - min_y)
dwg = svgwrite.Drawing(svg_filename, size=dims)
dwg.add(dwg.rect(insert=(0, 0), size=dims,fill='white'))
lift_pen = 1
abs_x = 25 - min_x
abs_y = 25 - min_y
p = "M%s,%s " % (abs_x, abs_y)
command = "m"
for i in xrange(len(data)):
if (lift_pen == 1):
command = "m"
elif (command != "l"):
command = "l"
else:
command = ""
x = float(data[i,0])/factor
y = float(data[i,1])/factor
lift_pen = data[i, 2]
p += command+str(x)+","+str(y)+" "
the_color = "black"
stroke_width = 1
dwg.add(dwg.path(p).stroke(the_color,stroke_width).fill("none"))
dwg.save()
display(SVG(dwg.tostring()))
# generate a 2D grid of many vector drawings
def make_grid_svg(s_list, grid_space=10.0, grid_space_x=16.0):
def get_start_and_end(x):
x = np.array(x)
x = x[:, 0:2]
x_start = x[0]
x_end = x.sum(axis=0)
x = x.cumsum(axis=0)
x_max = x.max(axis=0)
x_min = x.min(axis=0)
center_loc = (x_max+x_min)*0.5
return x_start-center_loc, x_end
x_pos = 0.0
y_pos = 0.0
result = [[x_pos, y_pos, 1]]
for sample in s_list:
s = sample[0]
grid_loc = sample[1]
grid_y = grid_loc[0]*grid_space+grid_space*0.5
grid_x = grid_loc[1]*grid_space_x+grid_space_x*0.5
start_loc, delta_pos = get_start_and_end(s)
loc_x = start_loc[0]
loc_y = start_loc[1]
new_x_pos = grid_x+loc_x
new_y_pos = grid_y+loc_y
result.append([new_x_pos-x_pos, new_y_pos-y_pos, 0])
result += s.tolist()
result[-1][2] = 1
x_pos = new_x_pos+delta_pos[0]
y_pos = new_y_pos+delta_pos[1]
return np.array(result)
define the path of the model you want to load, and also the path of the dataset
In [139]:
data_dir = 'datasets/naam4'
model_dir = 'models/naam4'
In [140]:
# download_pretrained_models(models_root_dir=models_root_dir)
In [153]:
def load_env_compatible(data_dir, model_dir):
"""Loads environment for inference mode, used in jupyter notebook."""
# modified https://github.com/tensorflow/magenta/blob/master/magenta/models/sketch_rnn/sketch_rnn_train.py
# to work with depreciated tf.HParams functionality
model_params = sketch_rnn_model.get_default_hparams()
with tf.gfile.Open(os.path.join(model_dir, 'model_config.json'), 'r') as f:
data = json.load(f)
fix_list = ['conditional', 'is_training', 'use_input_dropout', 'use_output_dropout', 'use_recurrent_dropout']
for fix in fix_list:
data[fix] = (data[fix] == 1)
model_params.parse_json(json.dumps(data))
return load_dataset(data_dir, model_params, inference_mode=True)
def load_model_compatible(model_dir):
"""Loads model for inference mode, used in jupyter notebook."""
# modified https://github.com/tensorflow/magenta/blob/master/magenta/models/sketch_rnn/sketch_rnn_train.py
# to work with depreciated tf.HParams functionality
model_params = sketch_rnn_model.get_default_hparams()
with tf.gfile.Open(os.path.join(model_dir, 'model_config.json'), 'r') as f:
data = json.load(f)
fix_list = ['conditional', 'is_training', 'use_input_dropout', 'use_output_dropout', 'use_recurrent_dropout']
for fix in fix_list:
data[fix] = (data[fix] == 1)
model_params.parse_json(json.dumps(data))
model_params.batch_size = 1 # only sample one at a time
eval_model_params = sketch_rnn_model.copy_hparams(model_params)
eval_model_params.use_input_dropout = 0
eval_model_params.use_recurrent_dropout = 0
eval_model_params.use_output_dropout = 0
eval_model_params.is_training = 0
sample_model_params = sketch_rnn_model.copy_hparams(eval_model_params)
sample_model_params.max_seq_len = 1 # sample one point at a time
return [model_params, eval_model_params, sample_model_params]
In [154]:
[train_set, valid_set, test_set, hps_model, eval_hps_model, sample_hps_model] = load_env_compatible(data_dir, model_dir)
In [155]:
# construct the sketch-rnn model here:
reset_graph()
model = Model(hps_model)
eval_model = Model(eval_hps_model, reuse=True)
sample_model = Model(sample_hps_model, reuse=True)
In [156]:
sess = tf.InteractiveSession()
sess.run(tf.global_variables_initializer())
In [199]:
# loads the weights from checkpoint into our model
# load_checkpoint(sess, model_dir)
saver = tf.train.Saver(tf.global_variables())
ckpt = tf.train.get_checkpoint_state(model_dir)
print(int(ckpt.model_checkpoint_path.split("-")[-1])/100)
# tf.logging.info('Loading model %s.', ckpt.model_checkpoint_path)
saver.restore(sess, "models/naam4/vector-4100")
We define two convenience functions to encode a stroke into a latent vector, and decode from latent vector to stroke.
In [159]:
def encode(input_strokes):
strokes = to_big_strokes(input_strokes, 614).tolist()
strokes.insert(0, [0, 0, 1, 0, 0])
seq_len = [len(input_strokes)]
print(seq_len)
draw_strokes(to_normal_strokes(np.array(strokes)))
print(np.array([strokes]).shape)
return sess.run(eval_model.batch_z, feed_dict={eval_model.input_data: [strokes], eval_model.sequence_lengths: seq_len})[0]
In [160]:
def decode(z_input=None, draw_mode=True, temperature=0.1, factor=0.2):
z = None
if z_input is not None:
z = [z_input]
sample_strokes, m = sample(sess, sample_model, seq_len=eval_model.hps.max_seq_len, temperature=temperature, z=z)
strokes = to_normal_strokes(sample_strokes)
if draw_mode:
draw_strokes(strokes, factor)
return strokes
In [189]:
# get a sample drawing from the test set, and render it to .svg
stroke = test_set.random_sample()
stroke=test_set.strokes[202]
draw_strokes(stroke)
Let's try to encode the sample stroke into latent vector $z$
In [190]:
z = encode(stroke)
In [197]:
_ = decode(z,temperature=1) # convert z back to drawing at temperature of 0.8
Create generated grid at various temperatures from 0.1 to 1.0
In [102]:
for i in range(10):
temp = .1 + i*.1
print(temp)
for j in range(5):
stroke = decode(draw_mode=False, temperature=temp)
draw_strokes(stroke)
# stroke_grid = make_grid_svg(stroke_list)
# draw_strokes(stroke_grid)
Latent Space Interpolation Example between $z_0$ and $z_1$
In [83]:
# get a sample drawing from the test set, and render it to .svg
z0 = z
_ = decode(z0)
In [66]:
stroke = test_set.random_sample()
z1 = encode(stroke)
_ = decode(z1)
Now we interpolate between sheep $z_0$ and sheep $z_1$
In [84]:
z_list = [] # interpolate spherically between z0 and z1
N = 10
for t in np.linspace(0, 1, N):
z_list.append(slerp(z0, z1, t))
In [85]:
# for every latent vector in z_list, sample a vector image
reconstructions = []
for i in range(N):
reconstructions.append([decode(z_list[i], draw_mode=True), [0, i]])
In [ ]:
stroke_grid = make_grid_svg(reconstructions)
draw_strokes(stroke_grid)
Let's load the Flamingo Model, and try Unconditional (Decoder-Only) Generation
In [ ]:
model_dir = '/tmp/sketch_rnn/models/flamingo/lstm_uncond'
In [ ]:
[hps_model, eval_hps_model, sample_hps_model] = load_model_compatible(model_dir)
In [ ]:
# construct the sketch-rnn model here:
reset_graph()
model = Model(hps_model)
eval_model = Model(eval_hps_model, reuse=True)
sample_model = Model(sample_hps_model, reuse=True)
In [ ]:
sess = tf.InteractiveSession()
sess.run(tf.global_variables_initializer())
In [ ]:
# loads the weights from checkpoint into our model
load_checkpoint(sess, model_dir)
In [ ]:
# randomly unconditionally generate 10 examples
N = 10
reconstructions = []
for i in range(N):
reconstructions.append([decode(temperature=0.5, draw_mode=False), [0, i]])
In [ ]:
stroke_grid = make_grid_svg(reconstructions)
draw_strokes(stroke_grid)
Let's load the owl model, and generate two sketches using two random IID gaussian latent vectors
In [ ]:
model_dir = '/tmp/sketch_rnn/models/owl/lstm'
In [ ]:
[hps_model, eval_hps_model, sample_hps_model] = load_model_compatible(model_dir)
# construct the sketch-rnn model here:
reset_graph()
model = Model(hps_model)
eval_model = Model(eval_hps_model, reuse=True)
sample_model = Model(sample_hps_model, reuse=True)
sess = tf.InteractiveSession()
sess.run(tf.global_variables_initializer())
# loads the weights from checkpoint into our model
load_checkpoint(sess, model_dir)
In [ ]:
z_0 = np.random.randn(eval_model.hps.z_size)
_ = decode(z_0)
In [ ]:
z_1 = np.random.randn(eval_model.hps.z_size)
_ = decode(z_1)
Let's interpolate between the two owls $z_0$ and $z_1$
In [ ]:
z_list = [] # interpolate spherically between z_0 and z_1
N = 10
for t in np.linspace(0, 1, N):
z_list.append(slerp(z_0, z_1, t))
# for every latent vector in z_list, sample a vector image
reconstructions = []
for i in range(N):
reconstructions.append([decode(z_list[i], draw_mode=False, temperature=0.1), [0, i]])
In [ ]:
stroke_grid = make_grid_svg(reconstructions)
draw_strokes(stroke_grid)
Let's load the model trained on both cats and buses! catbus!
In [ ]:
model_dir = '/tmp/sketch_rnn/models/catbus/lstm'
In [ ]:
[hps_model, eval_hps_model, sample_hps_model] = load_model_compatible(model_dir)
# construct the sketch-rnn model here:
reset_graph()
model = Model(hps_model)
eval_model = Model(eval_hps_model, reuse=True)
sample_model = Model(sample_hps_model, reuse=True)
sess = tf.InteractiveSession()
sess.run(tf.global_variables_initializer())
# loads the weights from checkpoint into our model
load_checkpoint(sess, model_dir)
In [ ]:
z_1 = np.random.randn(eval_model.hps.z_size)
z_1 = z
_ = decode(z_1)
In [ ]:
z_0 = np.random.randn(eval_model.hps.z_size)
_ = decode(z_0)
Let's interpolate between a cat and a bus!!!
In [ ]:
z_list = [] # interpolate spherically between z_1 and z_0
N = 50
for t in np.linspace(0, 1, N):
z_list.append(slerp(z_0, z_1, t))
# for every latent vector in z_list, sample a vector image
reconstructions = []
for i in range(N):
reconstructions.append([decode(z_list[i], draw_mode=False, temperature=0.15), [0, i]])
In [ ]:
stroke_grid = make_grid_svg(reconstructions)
draw_strokes(stroke_grid)
Why stop here? Let's load the model trained on both elephants and pigs!!!
In [ ]:
model_dir = '/tmp/sketch_rnn/models/elephantpig/lstm'
In [ ]:
[hps_model, eval_hps_model, sample_hps_model] = load_model_compatible(model_dir)
# construct the sketch-rnn model here:
reset_graph()
model = Model(hps_model)
eval_model = Model(eval_hps_model, reuse=True)
sample_model = Model(sample_hps_model, reuse=True)
sess = tf.InteractiveSession()
sess.run(tf.global_variables_initializer())
# loads the weights from checkpoint into our model
load_checkpoint(sess, model_dir)
In [ ]:
z_0 = np.random.randn(eval_model.hps.z_size)
_ = decode(z_0)
In [ ]:
z_1 = np.random.randn(eval_model.hps.z_size)
_ = decode(z_1)
Tribute to an episode of South Park: The interpolation between an Elephant and a Pig
In [ ]:
z_list = [] # interpolate spherically between z_1 and z_0
N = 10
for t in np.linspace(0, 1, N):
z_list.append(slerp(z_0, z_1, t))
# for every latent vector in z_list, sample a vector image
reconstructions = []
for i in range(N):
reconstructions.append([decode(z_list[i], draw_mode=False, temperature=0.15), [0, i]])
In [ ]:
stroke_grid = make_grid_svg(reconstructions, grid_space_x=25.0)
In [ ]:
draw_strokes(stroke_grid, factor=0.3)
In [ ]: